<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Michael Zochowski on Medium]]></title>
        <description><![CDATA[Stories by Michael Zochowski on Medium]]></description>
        <link>https://medium.com/@mzochowski?source=rss-f9bef90711a5------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*nHqojO8ertKJfPm5bAJxCQ.jpeg</url>
            <title>Stories by Michael Zochowski on Medium</title>
            <link>https://medium.com/@mzochowski?source=rss-f9bef90711a5------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 02:03:09 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mzochowski/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Joining Ripple’s Xpring]]></title>
            <link>https://medium.com/logos-network/joining-ripples-xpring-e018943a256d?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/e018943a256d</guid>
            <category><![CDATA[xpring]]></category>
            <category><![CDATA[ripple]]></category>
            <category><![CDATA[blockchain]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Fri, 27 Sep 2019 16:46:45 GMT</pubDate>
            <atom:updated>2019-09-28T03:25:37.492Z</atom:updated>
            <content:encoded><![CDATA[<p>We are excited to announce that we joined Ripple this week, where the Logos team will continue working on the cutting edge of blockchain technology as part of the company’s Xpring initiative. Xpring helps innovative blockchain projects grow through partnerships and investment and builds the best crypto infrastructure to grow the XRP ecosystem.</p><p>Since the Logos project was launched almost two years ago, we have continually pushed the envelope of blockchain speed and scalability, and we are eager to bring that same energy and passion to the Xpring team.</p><p>Ripple has always been highly visible to us as one of the original innovators in crypto and fintech. Over the past few months and as we met the broader team, we have been repeatedly impressed by the scope of expertise and clarity of vision present at all levels of the organization. Ripple offers a peerless platform in the blockchain space, and the opportunity to contribute to and shape its growing ecosystem of projects is unparalleled.</p><p>While this chapter of Logos is closing, the ideas and spirit behind Logos will continue in our work at Xpring. We have not yet finalized the plans for the Logos core technology, which has continued to set the bar for blockchain performance, but we hope to have exciting news to share in the near future.</p><p>As we look ahead to our next chapter at Ripple, we expect that our team will work on a variety of projects at all levels of the Xpring platform, with a particular focus on enabling a wide range of DeFi applications. We continue to strongly believe that the future of payments and finance lies in blockchain and decentralization, and Ripple is the perfect place for us to help build that future.</p><p>We’re excited to work with you along the way!</p><p>The Logos Team</p><p>For more information on the deal, please see <a href="https://medium.com/xpring/welcoming-logos-3586dd8cb552">Xpring’s blog post</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e018943a256d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/joining-ripples-xpring-e018943a256d">Joining Ripple’s Xpring</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Benchmarking Hash and Signature Algorithms]]></title>
            <link>https://medium.com/logos-network/benchmarking-hash-and-signature-algorithms-6079735ce05?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/6079735ce05</guid>
            <category><![CDATA[algorithms]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[cryptography]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Wed, 27 Mar 2019 21:26:57 GMT</pubDate>
            <atom:updated>2019-03-27T21:28:51.281Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Peng Wang, Carl Hua, and Michael Zochowski</em></p><h3>Introduction</h3><p>Hash and signature algorithms are the cornerstone of blockchain and crypto networks. Among other uses, they enable unforgeable identification and messaging, immutable recording of transactions, and efficient data storage. These properties make them a critical part of consensus, transaction verification, and, in proof-of-work networks, mining.</p><p>With the exception of proof-of-work (which, aside from extreme cases, is decoupled from the hash algorithm performance), the performance of the chosen hash and signature algorithms in a network has a major impact on performance. Every block and each transaction contained therein requires multiple signature verifications, hash verifications, and — for validators — a signature production. We estimate that the average crypto network node spends 25–65% of its CPU usage for hashing and signing, excluding mining. Optimizing these functions is one of the best ways to boost overall network performance, including transaction capacity and confirmation latency.</p><p>When we were architecting the Logos system, we were faced with the question of what signature and hashing algorithms best match our needs for high performance and minimal overhead. While there are many benchmarking results available on the web, they can vary significantly depending on the exact system environment. In particular, Logos uses proof-of-stake and a heavily optimized PBFT consensus scheme. We require multisignatures for efficient consensus, single signatures for transaction approval, and hashing for message digests, Merkle Trees, hashchains, and other data structures.</p><p>We therefore tested the performance of several hash algorithms and signature algorithms in our development environment last summer. For this benchmarking, we used AWS t2.2xlarge instances with 8 vCPUs from an E5–2686 v4 CPU at 2.30GHz running Ubuntu 16.04 with kernel 4.4.0.</p><p>In this article, we summarize those benchmarking findings. You can find the raw code we used published on our Github <a href="https://github.com/LogosNetwork/crypto-function-benchmark">here</a>. This article is intended to be a technical overview and assumes familiarity with the various concepts and algorithms we tested.</p><h3>Hash Algorithms</h3><p>We tested some of the hash algorithms available in Openssl’s libcrypto, version 1.1.1. The figure below is the per byte time (nanoseconds) of the hash algorithms with different message sizes. The x-axis is in log scale. The table underneath is the raw data, sorted by the column of message size 1024 bytes.</p><p>We can see that (1) for smaller messages up to 3KB, the cost per byte decreases when the message size increases, and for larger messages, the cost per byte is constant; and (2) the sha256 and sha512 have good performance, and blake2b512 is even faster.</p><p>Please note that sha1 and md5 were included only for reference, and they were not considered for use in Logos due to their <a href="https://security.googleblog.com/2017/02/announcing-first-sha1-collision.html?">security issues</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*qYBP92aytBNAXDIM" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*HU4-Dnjuk2Z10158" /></figure><p>Messages in Logos are typically in the range of 512B to 2KB. The next figure shows the cost of hashing 512B, 1KB, and 2KB messages. Please note the y-axis is per message time. This clearly confirms the intuition that when the message size doubles, cost also roughly doubles.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*8vJKBDaDkErwlAR0" /></figure><p>We chose blake2 for Logos’s hashing algorithm due to its speed. We specifically use blake2b256 because it targets 64 cpu architecture and provides 128-bit security.</p><p>In addition to its use in data structures and message digests, hashing algorithms are used in Logos for a minor proof-of-work spam prevention mechanism. For this role, we additionally considered Scrypt, Balloon, and Argon2id, but exclude them here for simplicity.</p><h3>Signature Algorithms</h3><p>We tested the performance of the following signature algorithms: ECDSA (Openssl), EdDSA (Nano), EC-Schnorr (Zilliqa), and BLS (Keep).</p><h4>Single signatures</h4><p>First, we show the cost of single signatures in the figures below. Similar to the previous section, we have the “Time per byte” figure for all message sizes, its raw data table, and the “Time per message” figure for 512B, 1KB, and 2KB messages, where time is measured in microseconds. We can see that (1) EdDSA is very fast, (2) for any algorithm, the cost of signing (verifying) 512B, 1KB, or 2KB messages are about the same.</p><p>In general, a signature algorithm signs the digest of messages. I.e., a message is hashed to a fixed size digest, then the signature is computed on the digest. So the cost includes two parts (1) hashing cost which grows with message size, (2) signature operation cost which is constant. This is clearly shown in the last figure in this section. For this reason, we included blake2b512 in all the figures as a reference. In general, the cost of (2) significantly dominates that of (1).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*4BpTLXSassJ7imEi" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*xo__pAv0tLB07mV5" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*V96aVVId3nVuE2Xw" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bnTmgoQSm9f2qwi0" /></figure><p>We decided to use EdDSA for Logos since it was the most efficient signature scheme and has a number of robust, open source implementations.</p><h4>Multisignatures</h4><p>Transaction validation in Logos involves a set of 32 elected delegates running consensus instances in parallel. Each consensus instance involves a batch of transactions and is run by a primary delegate with 31 backups. A supermajority (greater than ⅔) of delegates must sign off on each batch of transactions for them to be confirmed. Safety is strongly guaranteed through a combination of mathematically proven distributed algorithms and game theory.</p><p>If Logos were to implement this system using single signatures, validating that a batch (and thus the contained transactions) is valid as well as storing the network history would be extremely inefficient in both CPU and storage space since each batch involves at least 22 signatures. We therefore use a multisignature scheme to make signature size constant, regardless of the number of individual signatures it contains.</p><p>We tested three multisignature algorithms. The Schnorr multisignature is a two-round interactive algorithm. The BLS_Diff_Msg is a simple aggregation algorithm included in herumi’s code “<a href="http://github.com/herumi">github.com/herumi</a>”. It does not have a public key aggregation algorithm and is not secure when the messages could be the same. The BLS_Same_Msg is built on top of herumi’s code. It has a public key aggregation algorithm, and it requires all the messages are the same.</p><p>In this section, all the messages are 512B, since we already know message sizes only affect the hashing cost of signatures, from the previous section. The figure below shows the time cost of the operations. Some observations:</p><ol><li>“Public key aggregation” cost is small, and is a one time cost for the same set of delegates.</li><li>The Schnorr algorithm’s signing operation is a bit higher than BLS.</li><li>BLS_Diff_Msg’s signature aggregation cost is very low and BLS_Same_Msg’s signature aggregation cost is much higher.</li><li>The signature verification cost of BLS_Diff_Msg is very high.</li><li>No matter which algorithm, if some of the input to the signature aggregation function are corrupted, the cost of finding the corrupted ones (marked as Response Verify in the figure) is very high.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*cSBxiG-LfmHZXim5" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*WRiP5GzlDMBfaAan" /></figure><p>The next figure is the cost of the delegates of one multi-signature generation. We considered two cases, (A) all 31 delegates returned good individual signatures (or equivalent in Schnorr) to the primary, (B) some (&lt;1/3) individual signatures are corrupted. Other cases such as missing signatures, or too many corrupted signatures which resulted in a failed round should have less cost as compared to cases (A) and (B).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*c9wuMTRQQ41C-QSu" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*yotKROdBhFuSdwDF" /></figure><p>Thus, Schnorr outperforms BLS in CPU space, but BLS is within an order of magnitude. What these results do not show is the cost of messaging. In particular, a Schnorr multisignature requires one more round of messages between the aggregator and single signers than BLS, which complicates the core consensus protocol and imposes a significant time overhead due to latency and communication. Assuming a 200ms latency (a multisignature is limited by the highest latency signer), the Schnorr multisignature would have an additional latency of 200,000 microseconds. This means that BLS is drastically more time efficient than Schnorr in a real world setting. In addition, BLS requires only half the data as Schnorr for the equivalent security level. For example, the signature size of Schnorr multisignature scheme is 64 bytes for 128-bit security, while the signature size of BLS is only 32 bytes.</p><p>For these reasons, among others, we decided to use BLS multisignatures in Logos. To enable signature aggregation, we extended publicly available implementations based on <a href="https://crypto.stanford.edu/~dabo/pubs/papers/BLSmultisig.html">Boneh et al.</a> We’ll release that implementation publicly when we open-source the Logos core codebase.</p><h3>More Benchmarking Data: eBACS</h3><p>While it’s always important to test out the algorithms in the specific context you wish to deploy them, ECRYPT Benchmarking of Cryptographic Systems (<a href="http://bench.cr.yp.to/">eBACS</a>) is a great source of general benchmarking results of many cryptographic algorithms.</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6079735ce05" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/benchmarking-hash-and-signature-algorithms-6079735ce05">Benchmarking Hash and Signature Algorithms</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The PoS Cash Flow Valuation Model or: How I Learned to Stop Worrying and Love Velocity]]></title>
            <link>https://medium.com/logos-network/the-pos-cash-flow-valuation-model-or-how-i-learned-to-stop-worrying-and-love-velocity-cb4c240791cd?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/cb4c240791cd</guid>
            <category><![CDATA[finance]]></category>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Tue, 12 Mar 2019 17:35:55 GMT</pubDate>
            <atom:updated>2019-03-12T17:35:55.546Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BgGmLq0IL-6ZfRe4grH3tw.jpeg" /></figure><p>In our <a href="https://medium.com/logos-network/velocity-and-demand-for-money-models-e5bf71333934">last valuation article</a>, we explored more sophisticated frameworks for thinking about velocity, in particular inventory models of demand for money such as Baumol-Tobin. While an interesting theoretical extension, we ultimately concluded that such models are not particularly useful in a practical valuation context due to: (1) significant uncertainty in economic literature around model metaparameters such as the income elasticity of money demand; and (2) replacing one variable (velocity) with multiple new variables that are equally, if not more, ambiguous.</p><p>Given this rather unsatisfying conclusion, we must turn to other approaches for estimating velocity. We’ve previously discussed how both <a href="https://medium.com/logos-network/empirical-velocity-estimates-and-artificial-volumes-in-ethereum-9aa3d93bb60">empirical observations</a> and <a href="https://medium.com/logos-network/cryptoasset-valuation-2-the-velocity-problem-8bbb4111c9c7">first principles arguments</a> both suggest that velocity will not become arbitrarily high but instead will stick in the 1 to 100 range. However, given the uncertainty around and immaturity of decentralized networks generally, we must consider the possibility that our arguments are incorrect and that velocity does indeed diverge.</p><p>The obvious question that arises is: <em>What are the implications for cryptoasset value if velocity is arbitrarily high?</em></p><p>In my <a href="https://medium.com/logos-network/cryptoasset-valuation-2-the-velocity-problem-8bbb4111c9c7">first article on velocity</a>, I sketched a counterexample to the claim that cryptoassets would have arbitrarily low value with an arbitrarily high velocity based on cross-sectional transaction volumes. This showed that the commonly held “doomsday” arguments around velocity levels are misleading.</p><p>In this article, we more formally consider this scenario through an alternative valuation model specific to proof-of-stake (PoS) networks that decomposes network value into a variety of sources, several of which are independent from velocity. As we shall see, the end result establishes that networks can accrue significant value, even with huge velocities. Formalizing this valuation framework enables analysts to better understand value accrual and stakeholder dynamics in decentralized networks and increase the precision of estimates.</p><h3>The PoS Cash Flow Model</h3><p>For lack of a better alternative, we’ll call this new model the Proof-of-Stake Cash Flow Model. The key idea behind the model is that staking validators in a proof-of-stake network (as a group) earn recurring revenues from network fees. The value of this cash flow, and thus the value of the tokens staked to earn that cash flow, can easily be analyzed by standard financial analysis.</p><p>This model has been touched on by a few papers and industry commentators,¹ but it is far less prominent than the EoE model. Furthermore, it is valid for a much narrower range of networks than the EoE model and has some additional limitations that we will touch on below. Nevertheless, it is very useful in understanding the true drivers of value in a staking-based network. In this article, we will add in a few wrinkles to the basic model and formalize some of the key concepts behind it.</p><h4>Model Setup</h4><p>For simplicity, we will make a number of assumptions around the structure of the network. All of these simplifying assumptions can be relaxed without impacting the high level conclusions, but doing so significantly increases the mathematical complexity of the model.²</p><p>Our idealized network will be a payments network with two tokens: a base utility token and a stablecoin pegged to $1 per token. The base utility token is used to pay for transaction fees, while the stablecoin serves as the unit of account and medium of exchange in transactions. There are a fixed number of utility tokens in the system, <em>N</em>.</p><p>Transactions are validated by a group of validators that are paid pro rata to the number of tokens they stake out of the pool of transaction fees. There is no inflation in the system, so fees consist solely of direct fees paid by transactors. Assume that in period <em>t</em> (e.g. a year) the network processes <em>Y_t </em>transaction volumes (measured in dollars), and that validators receive <em>c*Y_t </em>in fees. Here, the fee <em>c</em> is a fraction of transaction size, but we could easily reparameterize to a constant fee per transaction by making assumptions around average transaction size. We assume that cost of validation is zero (which is reasonable in a PoS network), but any costs can be incorporated into <em>c</em> by netting.</p><p>Assume further that both fees and transactions grow at a constant rate <em>g</em>. This will be useful for defining the basic model, but note that we can explicitly model out future growth into discrete periods plus a terminal growth rate, akin to a standard DCF analysis.</p><p>In addition to the validator staking, there are two additional sources of utility token sinks. First, users hold utility tokens as working capital for fee payments, represented without loss of generality as a fraction of total transaction fees <em>h*c*Y_t</em>. Here, <em>h</em> roughly corresponds to average holding period, and thus is inversely proportional to velocity. We can substitute in a more detailed model such as Baumol-Tobin to solve for money demand, but as noted previously these models tend to complicate without edifying the model.</p><p>Second, each account on the system requires a small deposit (denominated in dollars) <em>d</em>. In addition to tying up tokens proportional to network use (and thus adding value to the network independent of velocity), this deposit mechanism is primarily useful as an anti-spam measure. If each account transacts <em>y</em> dollars per year, then there are <em>Y_t / y * d </em>dollars tied up in deposits.</p><p>Note that all cash flows and deposits are paid in tokens but are valued in dollars. This is the most natural way to look at things, as we are evaluating the value of validation as a USD-based business. In particular, the fees <em>c*Y_t</em> paid are independent of token value — as token value changes, so too does the number of tokens required for each fee, but the fee itself per transaction does not change. This reflects the operating reality of a standard business, where both revenue and expenses are measured in a single fiat currency like USD.</p><p>Most importantly, we assume that the network has achieved some equilibrium value, <em>V_t</em>, at time <em>t</em>. While non-trivial, it is also not an overly restrictive assumption. For example, the equilibrium value could be $0. This equilibrium value is valid for time <em>t</em>, and thus the model defines a cross-sectional snapshot of value at some period rather than a time-varying solution.</p><p>Let the fraction of tokens held by validators in this equilibrium be <em>φ</em>, the fraction held by transactors as working capital be <em>θ</em>, and the fraction held as account deposits be <em>δ</em>. In other words, the utility token value held by validators is <em>V_t*φ. </em>Assuming economic efficiency, these three uses of the network utility tokens are comprehensive, so:</p><p><em>φ + θ + δ = 1</em></p><p>Lastly, we need to specify the discounting rate, <em>r</em>, that encapsulates the risk free rate, opportunity cost of capital, and any risk premium in the system.</p><h4>Model specification</h4><p>We now can solve for the value corresponding to the three uses of utility tokens and derive overall equilibrium value <em>V_t</em>.</p><p>First, the validators. In period <em>i</em>, validators receive fees with net present value</p><p><em>c*Y_t * ((1+g)/(1+r))^i</em></p><p>Adding up all the periods <em>i </em>in perpetuity gives a <a href="https://en.wikipedia.org/wiki/Geometric_series#Formula">geometric series</a> that corresponds to a standard <a href="https://en.wikipedia.org/wiki/Annuity">annuity</a> or <a href="http://macabacus.com/valuation/dcf/terminal-value">DCF terminal value</a>. Its value is equivalent to</p><p><em>V_t*φ = c*Y_t / (r — g)</em></p><p>Second, we have the working capital holders, whose holdings value we have defined as</p><p><em>V_t*θ = h*c*Y_t</em></p><p>Third, we have the account deposits, which we defined as</p><p><em>V_t*δ = Y_t / y * d</em></p><p>Solving this system of equations yields a simple closed form solution for <em>V_t</em>:</p><p><em>V_t = Y_t * (c/(r-g) + h*c + d/y)</em></p><p>A few notes on this result. Analyzing the variables in isolation (i.e. holding all others constant) conforms with prior expectations. As overall transaction volumes <em>Y_t</em> increase, so too does value. Value also increases with fee level <em>c</em>, growth rate <em>g, </em>holding period <em>h</em>, and deposit <em>d</em>, while it decreases with discount rate <em>r</em> and transaction volume per account <em>y</em>. Note that these relationships are not necessarily true in practice, where changes in one variable impact other variables.</p><p>The valuation formula very nicely does not depend on future value <em>Y_t</em>′. Otherwise, we would need to use more advanced statistics to account for the stochasticity of this process.</p><p>Velocity is captured indirectly by holding period, its inverse. As expected, as velocity increases, holding period decreases, and network value decreases. However, only one of the three sources of value accrual is impacted by velocity. <strong>If velocity rises arbitrarily and <em>h</em> converges to zero, <em>V_t </em>does not converge to zero. Decoupling overall value from velocity is the key contribution of the model.</strong></p><h3>A Worked Example</h3><p>Let’s assume a reasonably efficient transaction network has the following values:</p><ul><li>It processes <em>Y_t </em>= $100 billion in transaction volumes per year</li><li>Constant growth rate in volumes is 5% per year</li><li>Discounting rate (including risk premium) is 10% per year</li><li>Validators earn an average of <em>c = </em>10 basis points per transaction</li><li>Holding period for users’ working capital of utility tokens is <em>h = </em>1 week ≈ 0.02 (corresponding to a velocity of 50)</li><li>Network mandated account deposits are <em>d = </em>$1 per account</li><li>Average transaction volume per account per year is <em>y = </em>$1000</li></ul><p>Plugging these in gives network value:</p><p><em>V_t = $100 billion * ( 0.1%/(10%-5%) + 0.02*0.1% + $1/$1000) = </em>$2.1 billion</p><p>We can further decompose this value into its components:</p><ul><li>Validator value = <em>V_t*φ = $100 billion * 0.1%/(10%-5%)</em> = $2 billion</li><li>Working capital value = <em>V_t*θ = h*c*Y_t = $100 billion * 0.02 * 0.001 = </em>$2 million</li><li>Account deposit value = <em>V_t*δ = Y_t / y * d = $100 billion * $1/$1000 = </em>$100 million</li></ul><p>This corresponds to the following breakdown of network tokens between the groups:</p><ul><li><em>Φ = 95.14%</em></li><li><em>θ = 0.01%</em></li><li><em>δ = 4.75%</em></li></ul><p><strong>So only 0.01% of network value is contingent on a reasonable velocity!</strong> Even if velocity were infinite, value would be substantially unchanged. This is an extreme example — it is a 2 token system where the value-accruing token does not denominate transactions, and a velocity of 50 is 10x what we see empirically — but it fits the narrative of valuation skeptics.</p><h3>Conclusion</h3><p>Despite its limitations, the PoS Cash Flow Model offers a significant improvement on the vanilla EoE model in several ways by cleanly separating out the sources of value accrual in a network. First, this decomposition allows for more nuanced analysis and isolation of assumptions of variable values that have high uncertainties. Second, it enables network designers to better understand the impact of features like account deposits on network value. Third, it greatly improves precision of business and valuation models as we move increasingly into a proof-of-stake paradigm.</p><p>The most important issue addressed by the model, though, is challenging the conventional wisdom, held by many, that velocity will erode any meaningful network value. As shown in these results, that presumption is categorically false, provided that the network is designed in an intelligent way to provide for value sinks like staking. The design process should always be bottoms-up and aimed at maximizing network utility rather than value, but luckily there is heavy overlap between the two objectives.</p><p>In future articles, we will explore more advanced versions of this model, in particular decomposing the value into discrete time periods where we can model out network volumes and TAMs as well as adding in stochasticity to model features like inflation.</p><p>[1] For a recent example, see section 7.1 <a href="https://dtr.org/wp-content/uploads/2019/01/2019-01-16-Decentralized-Payment-Systems-Principles-and-Design.pdf">here</a>.</p><p>[2] In particular, removing the simplifying assumptions typically requires adding stochasticity to the model.</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cb4c240791cd" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/the-pos-cash-flow-valuation-model-or-how-i-learned-to-stop-worrying-and-love-velocity-cb4c240791cd">The PoS Cash Flow Valuation Model or: How I Learned to Stop Worrying and Love Velocity</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Velocity and Demand for Money Models]]></title>
            <link>https://medium.com/logos-network/velocity-and-demand-for-money-models-e5bf71333934?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/e5bf71333934</guid>
            <category><![CDATA[economics]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[finance]]></category>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Tue, 05 Mar 2019 15:56:26 GMT</pubDate>
            <atom:updated>2019-03-07T19:59:11.664Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Written by </em><a href="https://medium.com/@mzochowski"><em>Michael Zochowski</em></a><em> and </em><a href="https://medium.com/@joe_29624"><em>Joe Alie</em></a></p><p>We’ve <a href="https://medium.com/logos-network/cryptoasset-valuation-1-improving-the-equation-of-exchange-model-6525aa748d28">talked in the past</a> about using the quantity theory of money or (a.k.a. Equation of Exchange) model to value cryptoassets. We labeled this the “base” model, since it is broadly applicable to all networks that have transactions denominated in a native token, which covers almost all networks.</p><p>One issue with the Equation of Exchange model is that it depends on assumptions around velocity and its relationship with the other variables in the formula <em>M*V = P*Q</em>. Typically, some constant velocity or functional form that is independent of the other variables is assumed. Note that this is not an unusual assumption for economic analysis, but it does impose some limitations on the model. Here’s what noted economist Greg Mankiw says about constant velocity:</p><blockquote>“As with many of the assumptions in economics, the assumption of constant velocity is only a simplification of reality. Velocity does change if the money demand function changes. For example, when automatic teller machines were introduced, people could reduce their average money holdings, which meant a fall in the money demand parameter k and an increase in velocity V. Nonetheless, experience shows that the assumption of constant velocity is a useful one in many situations.”¹</blockquote><p>The admitted difference between this standard economic analysis and forecasting done in the EoE cryptoasset valuation model is that the former has pretty high confidence around predominant levels of all the variables, while the latter involves substantially more uncertainty.</p><p>It’s been noted in several places, including <a href="https://medium.com/logos-network/empirical-velocity-estimates-and-artificial-volumes-in-ethereum-9aa3d93bb60">one of our recent articles</a>, that empirically observed velocities of cryptoassets are quite low, particularly when accounting for “self churn” transactions — often lower than fiat currency velocity, despite arguments to the contrary. Nevertheless, we are still in the extreme early stages of crypto adoption and infrastructure, so current empirical estimates have limited utility in informing our posterior expectations.</p><p>We’ve also <a href="https://medium.com/logos-network/cryptoasset-valuation-2-the-velocity-problem-8bbb4111c9c7">argued before</a> from first principles that there are many reasons to believe that velocity will not become arbitrarily high but instead will stick in the 1 to 100 range. While I believe there is little empirical or logical evidence for another result, such arguments are inherently informal.</p><p>The natural question that arises is: <em>Are there any more rigorous models that can give us more insight into realistic values of velocity going forward?</em></p><p>In this article, we’ll explore some common economic models for demand for money (which directly determines velocity), empirical evidence for these models in real economies, and implications for crytpoasset valuations.</p><h3>Transaction Models of Monetary Demand</h3><p>Velocity ultimately is determined by the <a href="https://en.wikipedia.org/wiki/Demand_for_money">demand for money</a>. Demand for money is a well-studied topic in economics, with a variety of proposed approaches. The most directly relevant class of models for the purposes of modeling cryptoasset velocities are inventory models for transactions demand.</p><h4>The Baumol-Tobin Model</h4><p>The most commonly used inventory model is the <a href="http://terpconnect.umd.edu/~jneri/Econ330/files/ch22apx1.pdf">Baumol-Tobin model</a>. <a href="https://medium.com/blockchannel/on-value-velocity-and-monetary-theory-a-new-approach-to-cryptoasset-valuations-32c9b22e3b6f">Alex Evans</a>, among others, has previously proposed the Baumol-Tobin model as a means of estimating velocity.</p><p>The idea is that individuals who know they are going to spend <em>Y</em> dollars over a period will optimize the amounts held in their interest-bearing savings account and in cash.² The money in the savings account earns interest <em>i</em>, while each withdrawal costs <em>C</em> (with the biggest component of cost typically being the value of someone’s time). Given this parameterization, we can use calculus to solve for the optimal number of withdrawals, <em>N</em>.</p><p>The end result of the Baumol-Tobin model is the average money holdings (adjusted for price level):</p><p><em>H = M/P = (C* Y / (2 * i))^(½)</em></p><p>The logic around an individual decision maker can be generalized to the entire economy, assuming uniform behavior (more on this in a bit), so the above equation also applies generally to a token GDP of <em>Y = P*Q</em>.</p><p>Given the average money holdings, we can easily solve for velocity by dividing <em>Y</em> by <em>H</em>.</p><p>There are a couple of issues with using Baumol-Tobin and other models. First, it recasts a single variable (V) in the M*V = P*Q model into multiple variables (C, i) for which we don’t necessarily have higher estimate precision. Second, the meta-assumptions of the Baumol-Tobin model — and any other monetary demand model, for that matter — end up being quite large and impactful in any conclusions. Of particular consequence is the income elasticity of demand, as noted by Greg Mankiw:</p><blockquote>“The Baumol–Tobin model, for example, makes precise predictions for how income and interest rates influence money demand. The model’s square-root formula implies that the income elasticity of money demand is 1/2: a 10-percent increase in income should lead to a 5-percent increase in the demand for real balances. It also says that the interest elasticity of money demand is 1/2: a 10-percent increase in the interest rate (say, from 10 percent to 11 percent) should lead to a 5-percent decrease in the demand for real balances. Most empirical studies of money demand do not confirm these predictions. They find that the income elasticity of money demand is larger than 1/2 and that the interest elasticity is smaller than 1/2. Thus, although the Baumol–Tobin model may capture part of the story behind the money demand function, it is not completely correct.”³</blockquote><p>The effect of these meta-assumptions is that Baumol-Tobin can give some crazy velocity numbers in a cryptoasset valuation model. In <a href="https://docs.google.com/spreadsheets/d/1a1SzF2H1Y3twTvqAlGAwm8Q2jG-CPnP1Q-7qopN-4LE/edit#gid=428912142">Alex Evans’s model</a>, for example, velocity ranges from 105 to 37,962!</p><h4>Income Elasticity of Money Demand</h4><p>What exactly is going on here? It turns out that income elasticity of money demand is a relatively poorly understood economic variable, and model-indicated values often differ substantially from empirical evidence in mature economies. Ultimately, it is this discrepancy that limits the value of these models — we are trying to develop better precision around velocity, but we’re no better off if the model we introduce has even less certainty!</p><p>Understanding the implications of income elasticity of money demand requires a bit of math. Elasticities measure the percentage change in a dependent variable <em>Y</em> for a unit percentage change in another variable <em>X</em>. This allows us to measure the unitless sensitivity of an output to its inputs.</p><p>Mathematically, elasticity is the <a href="https://economics.stackexchange.com/questions/13910/why-are-elasticities-defined-as-logarithmic-derivatives">partial derivative of the logarithmic variables</a>:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/538/0*I956IX_TA1eQNrif" /></figure><p>With any exponential function <em>Y = X^b</em>, similar to the Baumol-Tobin money demand equation referenced earlier, elasticity will be equal to the exponent <em>b</em>. Solving the above equation:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/641/0*_6BrZHjRMeBNmrnw" /></figure><p>Transforming the Baumol-Tobin equation <em>H = (C*Y / 2*i)^(½)</em> into logarithmic terms gives</p><p><em>log(H) = (½)log(Y) + (½)log(C) — (½)log(2i)</em></p><p>The income elasticity of money demand in the Baumol-Tobin model is given by the coefficient of log(Y), and therefore the model implies an income elasticity of ½. Other variants of the inventory model (e.g. Miller &amp; Orr 1966) suggest an income elasticity range between ⅓ and ⅔, yet empirical observations of income elasticity of money demand indicate much higher levels (see chart below, <a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.578.9410&amp;rep=rep1&amp;type=pdf">A Meta-Analysis of Money Demand</a>).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/470/0*JAXoVmeryZCxuxHy" /><figcaption>Empirically observed ranges for income elasticity of money demand. <a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.578.9410&amp;rep=rep1&amp;type=pdf">Source</a>.</figcaption></figure><p>We can therefore conclude a reasonable range for income elasticity is ⅓ on the low end (given by analytical models) and 1.34 on the high end (given by empirical observations for the EU). This seems to be a relatively small range, but, as we will see in the next section, even small differences in assumed income elasticity lead to massively different estimates of velocity.</p><p>We can adjust the Baumol-Tobin equation to reflect this range in income elasticity by exponentiating Y by the desired elasticity <em>ε</em> rather than the constant ½.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/169/0*xeh_a-xDyrJRAO9W" /></figure><h4>Velocity under different inventory model assumptions</h4><p>The following table shows velocities implied by the range of model-derived and empirically observed income elasticities of money demand.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*5k0Zx2K3gpFD_-kJ" /><figcaption>Velocity sensitivity: income elasticity of demand vs transaction volume</figcaption></figure><p>As we can see, velocity is extremely sensitive to the income elasticity of demand. While the model-derived elasticities suggest velocities that are very high, empirically observed elasticities suggest velocities that are extremely low. This variance in velocities directly corresponds to high variance in modelled cryptoasset value since value is inversely proportional to velocity.</p><p>We can replicate this analysis for other input variables in the Baumol-Tobin model, such as the cost of conversion from reference asset to cryptoasset, <em>C</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zSBR_UW9YbzhbXhF" /><figcaption>Velocity sensitivity: income elasticity of demand vs transaction cost</figcaption></figure><p>What causes these outsized changes in velocity for small changes in income elasticity? Elasticity manifests itself as an exponent in money demand models, which means it is extremely sensitive to small perturbations in elasticity.</p><p>Incidentally, the <a href="https://fred.stlouisfed.org/series/M1V">velocity of the U.S. M1 Money Supply</a> has hovered between 5 and 10, which is roughly in line with the implied velocities corresponding to the USA Empirical income elasticity of 0.84 in the sensitivity tables above.</p><h3>Conclusions</h3><p>Due to the large uncertainty around what level of income elasticity is appropriate to use (let alone other assumptions like conversion costs and interest elasticity) and the extreme sensitivity of the model to this assumption, it is quickly apparent that money demand models are very limited in their ability to provide additional insights into velocity. In addition to the sensitivity, these models are further limited due to the fact they replace one uncertain variable (velocity) with multiple, equally or more uncertain variables.</p><p><strong>At best, they can providing cross-validation and a sanity check of prior assumptions around velocity.</strong></p><p>[1] N. Gregory Mankiw, <em>Macroeconomics</em>, 2010, Section 4–2.</p><p>[2] The model logic can easily be extended to two return-generating alternative assets.</p><p>[3] Mankiw, Section 19–2.</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e5bf71333934" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/velocity-and-demand-for-money-models-e5bf71333934">Velocity and Demand for Money Models</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Empirical Velocity Estimates and Artificial Volumes in Ethereum]]></title>
            <link>https://medium.com/logos-network/empirical-velocity-estimates-and-artificial-volumes-in-ethereum-9aa3d93bb60?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/9aa3d93bb60</guid>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[ethereum]]></category>
            <category><![CDATA[data-science]]></category>
            <category><![CDATA[finance]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Tue, 26 Feb 2019 22:21:39 GMT</pubDate>
            <atom:updated>2019-02-26T22:21:39.432Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Eliminating self-churn from Ethereum transactions.</em></p><p>In a <a href="https://medium.com/logos-network/cryptoasset-valuation-2-the-velocity-problem-8bbb4111c9c7">previous article</a>, we discussed token velocity and its impact on token valuation. In particular, I argued qualitatively that there are many costs associated with swapping tokens that make it unrealistic to expect velocities above 100. This can be quantified with a demand-for-money model such as <a href="https://en.wikipedia.org/wiki/Baumol%E2%80%93Tobin_model">Baumol-Tobin</a>, but ultimately we end up relying on several murky assumptions. Furthermore, there are key considerations around the correlation or dependence between velocity and token GDP that complicates an analytical approach.</p><p>A more revealing, if less intellectually satisfying, approach is to empirically observe velocities of different assets. The easiest point of comparison is the US dollar, for which velocity data is <a href="https://fred.stlouisfed.org/series/M1V">readily available</a>. Recently, M1 USD velocity has hovered around 5.5. Since the Federal Reserve has pretty good insight into USD transactions, these estimates are likely accurate.</p><p>Similarly, we can naively estimate cryptoasset velocity in a relatively straightforward manner. It is easy to observe total Bitcoin transaction volumes and total Bitcoin supply and take their ratio, arriving at a velocity of around 2.1. However, there are a few issues with this calculation. For example, large numbers of Bitcoins are “hodl’d”, meaning they are effectively removed from circulation. This tends to artificially depress velocity. On the other hand, many Bitcoin transactions are not real transactions where one person is sending value to another person (more on this in a bit). This tends to artificially increase naive velocity.</p><p>Can we do better? Yes! Many of the issues arising from using Bitcoin as a reference arise because it is quite a poor payment rail (for target transaction use cases) right now — fees are high, finality times are slow, the currency is volatile, and so on. Luckily, we now have many better assets to reference — stablecoins. Since there is no possibility of their price increasing, there are no rational “hodl’ers” to account for.¹ Furthermore, they are more directly akin to true money, where holders are plausibly using it in real transactions.</p><p>To eliminate any further noise, it makes sense to set aside Tether — recognizing that there is some controversy around some of Tether’s funds, without commenting on the validity of these claims — and algorithmic stablecoins like DAI — where there are complicating factors around creation, redemption, and loans. We are then left with the audited full reserve stablecoins. Now that they all have existed for a few months, we have good, but still limited, data for <a href="https://etherscan.io/token/0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48">USDC</a>, <a href="https://etherscan.io/token/0x056fd409e1d7a124bd7017459dfea2f387b6d5cd">GUSD</a>, <a href="https://etherscan.io/token/0x8e870d67f660d95d5be530380d0ec0bd388289e1">PAX</a>, and <a href="https://etherscan.io/token/0x0000000000085d4780B73119b644AE5ecd22b376">TUSD</a>.</p><p>But there are still issues with the artificial transactions. If a user sends money between 2 addresses he owns, those transactions shouldn’t be counted in our volume numbers. Before diving into the data, then, it is worth first clearly defining our goal for calculating velocity.</p><h3>The Goal: Real Economic Transactions</h3><p>What do we want in our estimate of velocity? Most superficially, we are looking for the value that makes the basic M*V = P*Q true.² At a deeper level, P*Q represents the real economic transactions that are denominated in a money asset, such as the stablecoin. Regular users of the money asset will hold on to some inventory, and a marginal user that wishes to acquire the asset to transact will increase the value of M (whether through additional token issuance or value appreciation). Given that we can observe M, what we want is an accurate estimate of P*Q — the <em>bona fide</em> economic transaction volume on the network.</p><h3>The Problem: Self Churn and Artificial Volume</h3><p>As mentioned previously, the naive analysis is complicated by the fact that we know that many of the transactions observable on a blockchain are <em>not</em> real economic transactions.</p><p>The most common example of a transaction that we want to exclude is what has been dubbed “<em>self churn</em>”. In a UTXO network like Bitcoin, transactions have a change output that represents funds returned to the sender. Similarly, users will often consolidate or sweep their outstanding UTXOs into a single UTXO to clean up their wallet. Obviously, none of these transactions should be accurately counted as part of P*Q.</p><p>Self churn was perhaps first formalized by <a href="https://cseweb.ucsd.edu/~smeiklejohn/files/imc13.pdf">Meiklejohn et al.</a> in 2013. Subsequently, it has become common for analysts to use heuristics to remove self churn from the system. For example, <a href="https://arxiv.org/pdf/1709.02489.pdf">BlockSci</a> (section 3.4) proposes to eliminate (1) outputs controlled by an address linked to an input address (eliminates change transactions) and (2) outputs that are spent within <em>k</em> blocks. Similar heuristics have been used by <a href="https://www.blockchain.com/charts/n-transactions-excluding-chains-longer-than-100">Blockchain.com</a> and <a href="https://coinmetrics.io/introducing-adjusted-estimates/">Coinmetrics</a>.</p><p>Note that the heuristics are just that, and they are subject to false positives and false negatives; nevertheless, they are very useful in deriving order-of-magnitude estimates for transaction volume and velocity. BlockSci, for example, found that the naive method overstates BTC velocity <em>by a factor of 4</em>!</p><p>There is an issue when trying to apply these same heuristics to our selected stablecoins — the heuristics only work for a UTXO token, but the stablecoins are all ERC-20 tokens running on Ethereum, which uses an <a href="https://github.com/ethereum/wiki/wiki/Design-Rationale#accounts-and-not-utxos">account system</a>. This means there is no way to trace individual coins through the system, and we can’t know if a particular send/receive within <em>k</em> blocks used the same tokens.</p><p>There has been very little work on how to identify self churn and other artificial volume in Ethereum or other account-based tokens. Coinmetrics proposed a “<a href="https://coinmetrics.io/realized-capitalization/">virtual UTXO</a>” mapping, but this is not sufficiently deterministic to apply the UTXO heuristics. Coinmetrics also published a case study on <a href="https://coinmetrics.io/introducing-adjusted-estimates/">eliminating transactions attributable to the mixer</a> that dominated transactions in 2017, but none of the stablecoins we are interested in were active during that period, and, besides, it does not actually eliminate many other sources of artificial transactions.</p><p><strong>The goal of this article is to lay out a methodology for identifying the artificial volumes in ERC-20 stablecoins that can give us better estimates of P*Q and V.</strong></p><h3>The Data: BigQuery</h3><p>All transaction data for decentralized networks can be downloaded from a full node, but that can take days (and is expensive) to download. A better alternative is to use Google BigQuery, which contains <a href="https://cloud.google.com/blog/products/data-analytics/ethereum-bigquery-public-dataset-smart-contract-analytics">SQL-ized Ethereum data</a> updated on a daily basis.</p><p>Here’s an example of how you can scrape all PAX transactions:</p><pre>SELECT</pre><pre>DATE(block_timestamp) as TxDate,</pre><pre>token_transfers.*</pre><pre>FROM</pre><pre>`bigquery-public-data.ethereum_blockchain.token_transfers` AS token_transfers</pre><pre>WHERE</pre><pre>token_transfers.token_address = ‘0x8e870d67f660d95d5be530380d0ec0bd388289e1’</pre><pre>ORDER BY</pre><pre>block_number</pre><pre># GUSD: 0x056fd409e1d7a124bd7017459dfea2f387b6d5cd</pre><pre># PAX: 0x8e870d67f660d95d5be530380d0ec0bd388289e1</pre><pre># TUSD: 0x8dd5fbce2f6a956c3022ba3663759011dd51e73e</pre><pre># USDC: 0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48</pre><p>You can find the token addresses on <a href="https://etherscan.io/token/0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48">Etherscan</a>.</p><p>From there, it’s easy to download the data into a CSV or pull it directly into your preferred data analysis environment (I like Python).</p><h3>Identifying Artificial Volume In Ethereum Stablecoins</h3><p>While none of the UTXO heuristics work directly with Ethereum’s account system, we can adapt several of them to filter out many of the artificial transactions. Furthermore, we can dig into the specific stablecoin ecosystem to identify some additional sources of inflated volume.</p><p>I have uploaded a <a href="https://github.com/mczochowski/articles/blob/master/velocity/stablecoin_velocities.py">Python script</a> to my Github that shows how these filters can be applied in practice. It is also handy for exploring stablecoins data generally. There are many cool insights that we may write about later, including inventory churn, transaction sizes, daily active users, and more.</p><h4>Filter 1: Remove self-sends</h4><p>There are a small number of transactions whose sending and receiving addresses are the same. By definition, no funds have changed hands, so we cut these transactions out.</p><h4>Filter 2: Remove Minting and Burning Activity</h4><p>All the stablecoins of interest have minting and burning features. When a user deposits fiat at the stablecoin issuer, an equivalent number of stablecoins are minted and delivered to the user. Similarly, when a user sends stablecoins to the issuer, the stablecoins are burned and an equivalent value of fiat is credited to their account.</p><p>This minting and burning activity appears as a real transaction in the blockchain, but it should not be counted for our purposes. In reality, it is just an exchange of one form of money for the other, but ownership never changes. As an extreme example, imagine a user mints and then immediately burns $1mm repeatedly 1000 times. That apparent $2bn in volume (both minting and burning are counted) is obviously not related to P*Q.</p><p>We can easily identify minting and burning transactions by those that are sent from or to the root address (0x0000000000000000000000000000000000000000), respectively.</p><h4>Filter 3: Identifying Intermediate Mints and Burns</h4><p>Stablecoin issuers typically keep a “working capital” supply of stablecoins to make creation and redemption smoother. Thus, the lifecycle of a new coin involves minting by the token issuer to an intermediate address controlled by the issuer, and then a send to the user. As a result, the direct minting and burning filter only captures half the artificial volumes. While the token issuer effectively “finances” the stablecoins initially, it still is just a user switching from one form of USD to another rather than any real economic transactions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/478/0*jeJJ3BWQPi9v-v88" /><figcaption>Minting and burning addresses for various stablecoins</figcaption></figure><p>As shown in the table, the number of unique minting and burning addresses for all four stablecoins is minuscule compared to the total number of unique addresses involved in all transactions.</p><p>We can remove this artificial volume by removing any transactions sent from the intermediate minting addresses and any sent to the intermediate burning addresses.</p><h4>Filter 4: Exclude Known Exchanges</h4><p>Deposits into and withdrawals from exchanges cause multiple sources of artificial volume that do not reflect actual economic transactions.</p><p>First, a send into or withdrawal from an exchange is just moving money from one account to another controlled by the same user. While the stablecoin is then typically used to trade against other tokens, none of these transactions reflect those we are interested in — payment for a real good or service. Similarly, we don’t count brokerage account deposits or stock trading volumes in estimates of the USD M1 velocity or US GDP. There is some potential for false positives when people use the exchange as their liquid transactional wallet, but on a volume weighted basis these should be small compared to normal exchange usage.</p><p>Second, exchanges typically generate unique addresses for a particular user to deposit into (to make it easy for the exchange to track user activity), and then they are immediately sent to a consolidated, commingled account. See, for example, <a href="https://etherscan.io/token/0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48?a=0x5b273629b9fe1ac6ac013aa2370ceb4539dbe27f">these deposits of USDC into Huobi</a>. These clearly are spurious transactions for the purposes of calculating P*Q.</p><p>Third, there has been an idiosyncratic arbitrage opportunity related to how exchanges handle stablecoins and competitive practices of stablecoin issuers that has led to additional artificial volumes, particularly for GUSD and PAX. You can find a summary of these issues <a href="https://medium.com/@neutralproject/a-case-study-on-husd-752b16a2d1f8">here</a> and <a href="https://www.coindesk.com/stablecoin-crypto-price-war-paxos-gemini">here</a>.</p><p>Given these issues, we can justify eliminating all transactions involving known exchange addresses. Since exchanges generate unique addresses for specific users, it’s difficult to identify all addresses, but some of the later heuristics will help. Even some of the commingled addresses are not correctly identified. We can conclude that our estimates of volumes, therefore are generous, and the resulting estimates of velocity are conservative (biased high).</p><p>TokenAnalyst gives <a href="https://medium.com/tokenanalyst/classifying-ethereum-users-using-blockchain-data-dd6edb867de3">some background</a> on identifying exchange addresses, but I decided to rely on Etherscan’s <a href="https://etherscan.io/accounts/1?&amp;l=Exchange">exchange mapping</a>.</p><h4>Filter 5: Remove Temporary Addresses</h4><p>The last heuristic aims to replicate the standard UTXO assumption of removing temporary UTXOs.</p><p>We do so by finding all addresses that currently have a balance of 0 and the time difference between its first receive and its last send is less than <em>k</em> blocks. Here <em>k</em> must be larger than in Bitcoin due to the faster blocktime. I choose 60 blocks, or about 15 minutes, which is a standard confirmation wait time in Ethereum.</p><p>This heuristic will have more false negatives than its UTXO equivalent due to the structure of addresses. For example, an exchange deposit address (unique to the user) that is used twice, one day apart, and whose balance is immediately swept into a commingled address will not be flagged. Nevertheless, it is difficult to distinguish between these false negatives and legitimate addresses. Further work could explore matching up IN and OUT transactions and percentage of life at zero balance.</p><h3>Result: Adjusted Stablecoin Velocity Estimates</h3><p>The charts below show the rolling 30 day velocity estimates (in annualized terms) estimated first under the naive methodology (total volume / outstanding supply) and second using the self-churn filters.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/484/0*b55alziC3TJlTREP" /><figcaption>Velocities without removing self-churn appear to be ~10x USD velocity.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/498/0*nMPE9n9e0ec8XWiN" /><figcaption>Velocities with self-churn removed are in line with or lower than USD velocity.</figcaption></figure><p>The difference is quite stark. <strong>While the naive method indicates velocities ranging from 25 to 60, removing the artificial volumes shows a range of 2 to 10</strong>, in line with historically observed values for USD.</p><h4>Why the high velocities early?</h4><p>One interesting feature on both charts is the tendency for velocity to spike soon after the stablecoin debuts. There are a number of reasons for this unrelated to any real economic velocity.</p><p>Once the token is launched, significant quantities are minted and sent to early partners. While our minting filter can eliminate the initial transaction, it is very possible that those early partners then shifted around or tested the functionality of the tokens.</p><p>Additionally, token supply has historically been lowest at launch. Any transfer has an outsized impact during this period versus later periods. Partners that are testing integrations or even the stablecoin issuer itself sending test transactions will have their transactions magnified, and our filters cannot catch most of this activity.</p><p>In our internal models, we mitigate this issue by weighting by outstanding supply. Alternatively, the measurements can be smoothed by calculating total volume divided by average supply over the window. For simplicity, though, we stick with the basic calculation in this article and instead focus on the filters.</p><h3>Conclusion</h3><p>Velocity and token GDP are critical inputs to valuation models, and analytical models have led to wildly variable and unrealistic estimates. While data is limited, we can look at empirical evidence to get a better idea of what values velocity might take. Stablecoins are the closest asset we have to real-world use and eliminate many of the confounding factors like “hodl’ing”, and so they are the best source of empirical estimates.</p><p>Nevertheless, most of the transactions are not of interest in a velocity estimate for an economic valuation model. While some work has been done to identify such artificial volumes in Bitcoin and UTXO systems, little research has been done into account based systems like Ethereum and ERC-20-based stablecoins. We present a set of heuristic filters that can eliminate many, but not all, “self churn” transactions. We further estimate that these heuristics are conservative (more false negatives than false positives), meaning that velocities are overestimated. In spite of this, the filters are able to remove up to 90% of transaction volumes from the estimate.</p><p>There is a caveat to drawing definitive conclusions from these velocity estimates of 2 to 10. Ethereum and, by extension, its stablecoins have very limited usefulness for most real economic transactions. High fees, slow confirmation times, user experience frictions, and relative immaturity are all significant barriers.</p><p>Indeed, our conservative heuristics reveal that almost all transactions are not real economic transfers, and I would not be surprised if <em>bona fide</em> volumes are close to zero. This meshes with what we hear from the stablecoin issuers — right now, they are targeting traders and the nascent (but basically non-existent) smart contract payout economy rather than real payments until better platforms are found.</p><p>Going forward, ecosystem infrastructure will continue to improve, while networks like Logos will make transfers vastly more efficient, cost effective, and rapid, opening up many real-world payments use cases. These advances will plausibly reduce user frictions and increase velocity. As the cost of converting decreases, demand for working capital and thus average holding period decreases.</p><p>Nevertheless, we can confidently say that empirical evidence currently shows that decentralized stablecoins so far do not have meaningfully higher velocities than fiat currencies. To the extent that conclusions can be drawn, data supports the low velocity camp rather than the high velocity camp. The primary contribution of the article, the framework to eliminate artificial volumes in account-based systems, will be useful in monitoring changes as the ecosystem matures.</p><p>[1] This is a generalization; near stablecoin creation, there are potentially unanticipated nuances resulting from the issuer bootstrapping the network.</p><p>[2] Even if you are skeptical of projecting velocity values, accurately estimating P*Q should be of interest independent of model choice.</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9aa3d93bb60" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/empirical-velocity-estimates-and-artificial-volumes-in-ethereum-9aa3d93bb60">Empirical Velocity Estimates and Artificial Volumes in Ethereum</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Proof-of-Work Is Not Viable in the Long-Term]]></title>
            <link>https://medium.com/logos-network/why-proof-of-work-is-not-viable-in-the-long-term-dd96d2775e99?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/dd96d2775e99</guid>
            <category><![CDATA[bitcoin]]></category>
            <category><![CDATA[network]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[blockchain-technology]]></category>
            <category><![CDATA[algorithms]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Tue, 19 Feb 2019 19:24:44 GMT</pubDate>
            <atom:updated>2019-02-19T19:45:02.631Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A literature review of the challenges facing Nakamoto Consensus.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RCwrJu7mG8AsmA85zQEU0w.jpeg" /></figure><p><em>Note: Proof-of-Work (PoW) and Proof-of-Stake (PoS) are not consensus algorithms but rather Sybil control mechanisms that are compatible with a variety of consensus algorithms. For the sake of simplicity, in this article we will stick with popular convention and use PoW as a synonym for Nakamoto Consensus and PoS as a synonym for staking coupled with consensus algorithms like PBFT.</em></p><p>The success of Bitcoin and its pioneering Nakamoto Consensus based on proof-of-work showed incontrovertibly that decentralized networks can work and accrue value for users and network stakeholders by solving key problems such as Sybil control and economic incentivization. Nevertheless, it has been plagued in recent years by performance issues and inefficiencies, even with only marginal real-world adoption. Simultaneously, newer consensus models that promise greater efficiency and effectiveness have emerged, challenging proof-of-work’s hegemony as the choice for decentralized consensus.</p><p>In the face of these alternatives gaining traction, Bitcoin maximalists have become increasingly entrenched and dogmatic in their views, to the exclusion of reasonable discourse and empirical and mathematical evidence that challenges their views. Nevertheless, there is a growing pile of results that suggest that proof-of-work is not viable in the long-term for a healthy, mature network. While it served a key role as a proof-of-concept, it is strictly inferior to newer systems across all important aspects — security, cost, efficiency, performance.</p><p>In this article, I summarize the key issues that proof-of-work faces, both currently and as the network matures. For the sake of accessibility, I will keep things high level but will reference papers or articles that give additional technical details and proofs.</p><p>I will also keep the focus on proof-of-work and its challenges rather than its competitors, but the comparison throughout (whether explicitly or implicitly) is to proof-of-stake. For a more detailed discussion of PoS, I refer the reader to the myriad papers and articles that mathematically show how PoS improves on PoW, such as the<a href="https://www.logos.network/whitepaper"> Logos white paper</a>.</p><h3><strong>1. PoW has weak security compared to PoS and may be unstable at maturity</strong></h3><h4><strong>1a.</strong> PoW has poor economic/game theoretic security that results in instability</h4><p>Any trustless network must answer two questions: (1) what is the Byzantine security level (e.g. the network is secure assuming an attacker holds less than 50% of power), and (2) what is <strong>the probability that security level is met</strong>. Most projects answer (1), but (2) is just as critical. Distributed algorithm analysis gives us the Byzantine security level, but that alone is insufficient as it begs the question, <em>what is the probability a validator becomes malicious</em>? This is fundamentally a <em>game theory</em> question and can be answered via an analysis of the economic incentive system.</p><p>Nakamoto Consensus has a relatively weak economic model, and so incentive alignment for validators is quite poor. This is due to several reasons, including:</p><p><strong>1.</strong> Inability to harshly penalize validators (like in slashing). Since mining is inherently anonymous and open, it is impossible to blacklist a malicious validator from participating in the network, and there is no way to confiscate their mining equipment as punishment for misbehaving.</p><p><strong>2.</strong> Low correlation between the network value and the economic collateral of the validator. Mining equipment typically has a low “beta” (less than 1) to the value of the underlying network. While there is secular exposure, an attack on any particular network is unlikely to meaningfully impact the payout of a potential attacker. Ironically, the efforts of networks to thwart ASICs, which contribute to centralization (as discussed below), reduce this correlation to very low levels. Together with the lack of a direct penalization mechanism, this means that an attacker has only weak economic incentives to behave honestly.</p><p><strong>3.</strong> High variance of validator payoffs. PoW has a lottery effect: either you get a big reward by chance (block rewards and transaction fees), or you get nothing. This variance in payout leads to unstable economic equilibria and perverse incentives.</p><p><strong>4.</strong> Transaction finality is probabilistic, not guaranteed. After a transaction is included in a mined block, there is always a chance that a longer chain comes along and reverses the transaction, although this probability decreases as additional blocks are added.</p><p>These features lead to several known attack vectors and likely many more that are currently unknown that undermine the security and stability of the PoW system at maturity. Unfortunately, none of the root causes of these issues are correctable in PoW, which means that, even with increasing awareness of these attacks, there is no way to harden the system against most of them.</p><h4><strong>Original selfish miner attack</strong>:</h4><p>The first major game theoretic attack that can compromise the long-term viability of PoW consensus is the<a href="https://www.cs.cornell.edu/~ie53/publications/btcProcFC.pdf"> selfish miner attack</a>. Cornell researchers found that miners are economically incentivized to hide blocks they mine from the rest of the network to maximize their own payout. The idea is that only the selfish miner has the longest chain and so is advantaged in finding subsequent blocks; if another block is found on the old chain, the selfish miner can release their block and trigger a race condition that they might win.</p><p>This has major implications for both network decentralization and security:</p><blockquote>“We present an attack with which colluding miners obtain a revenue larger than their fair share. This attack can have significant consequences for Bitcoin: Rational miners will prefer to join the selfish miners, and the colluding group will increase in size until it becomes a majority. At this point, the Bitcoin system ceases to be a decentralized currency.”</blockquote><p>By hiding their blocks from the rest of the network, the selfish miners also reduce the effective hash power of the network since other miners are mining a stale chain and not contributing to the security of the network. If every miner is selfish, then effective hash power will only be a small fraction of deployed hash power. None of this requires miners to be outright malicious, just economically rational!</p><h4><strong>Selfish miner attack is way worse than previously thought</strong>:</h4><p>The original selfish miner paper proposed a modification to PoW that would eliminate the selfish miner attack for mining pools that control less than 25% of the hash rate. Nevertheless, subsequent research from Princeton found<a href="https://www.cs.princeton.edu/~smattw/CKWN-CCS16.pdf"> additional selfish miner problems</a> that occur when transaction fees grow relatively larger than the block rewards (<a href="https://freedom-to-tinker.com/2016/10/21/bitcoin-is-unstable-without-the-block-reward/">summary</a>). The paper outlines several deviant miner strategies that are economically rational, whereby miners undercut each other by promoting forks to capture large transaction fees:</p><blockquote>“Bitcoin provides two incentives for miners: block rewards and transaction fees. The former accounts for the vast majority of miner revenues at the beginning of the system, but it is expected to transition to the latter as the block rewards dwindle. There has been an implicit belief that whether miners are paid by block rewards or transaction fees does not affect the security of the block chain. We show that this is not the case. Our key insight is that with only transaction fees, the variance of the block reward is very high due to the exponentially distributed block arrival time, and it becomes attractive to fork a “wealthy” block to “steal” the rewards therein. <em>We show that this results in an equilibrium with undesirable properties for Bitcoin’s security and performance, and even non-equilibria in some circumstances</em>. We also revisit selfish mining and show that it can be made profitable for a miner with an arbitrarily low hash power share, and who is arbitrarily poorly connected within the network.”</blockquote><p>Thus, the economic equilibrium in a PoW network involves miners continuously undermining one another, leading to weaker network security and a centralization tendency. These perverse incentives are exacerbated at network maturity to the point that PoW networks will have seriously compromised security.</p><h4><strong>Hash renting minimizes the cost of a 51% attack:</strong></h4><p>These issues are made even worse by the growing ability of attackers to rent hash power rather than purchasing the equipment themselves. This means that the attacker’s exposure, however weak, to the correlation between mining equipment and network value, is eliminated. Combined with the tendency of miners to work against one another, resulting in lower effective network hash power, the cost of attacking a network via a 51% attack is much lower than typically thought.</p><p>Of course, as a whole, the owners of the mining equipment have some economic incentive to prevent such an attack. However, this is an economically irrational expectation in the absence of an enforcement mechanism (impossible in a pseudo-anonymous, decentralized network) due to the<a href="https://en.wikipedia.org/wiki/Free-rider_problem"> free rider</a> and<a href="https://en.wikipedia.org/wiki/Tragedy_of_the_commons"> Tragedy of the Commons</a> phenomena. The dominant <em>unilateral </em>strategy of any single miner is to rent out their hash rate for a better return, resulting in globally suboptimal Nash equilibrium.</p><p>Traditional security analysis of PoW assumes the attacker has fixed hash power, but, in reality, hash power can be acquired if economically rational. The resulting analysis shows that <em>transaction security against a 51% attack is far weaker than typically assumed</em>.</p><p>While the original Bitcoin white paper showed that the risk of a successful reorg attack for an attacker with less than 50% hash power declines <em>exponentially</em> with the number of blocks since validation, a<a href="https://www.bis.org/publ/work765.pdf"> recent BIS paper</a> shows that the game theoretic risk of an attack from <em>renting a majority of hash power</em> declines only <em>linearly</em> and depends highly on block fees/reward. This means that the required number of blocks for probabilistically secure transactions is much higher than anticipated, and oftentimes prohibitively long:</p><blockquote>“Equation (8) documents the high cost of decentralised payment security. For example, say that Bitcoin users are on average prepared to pay transaction costs of 1%, that rented hash power is twice as expensive as the underlying price of equipment and electricity for honest miners, and that bitcoin would lose one third of its value after a successful attack. Then, the required waiting time is 50 blocks (over eight hours). But if the average transaction cost is 0.1%, the required waiting time is 500 blocks, ie around three and a half days! … If users wait for six blocks or fewer (about one hour), the required average transaction costs (as a percentage of the transaction amount) are about 8.3%!”</blockquote><p>So the higher the transaction fee (or block reward), the better the security. Unfortunately, this has a perverse interaction with the selfish miner attacks, whereby miners’ incentive to undermine each other increases as miner fees increase!</p><p>Additionally, there is yet another free rider problem with how users decide on transaction fees that results, at equilibrium, in untenable finality times:</p><blockquote>“The key result is that the fee set on a decentralised basis is much lower than the optimal fee, resulting in extreme waiting times… With the above example of an attacker disadvantage of 1/2, 1,000 transactions, and a cost of waiting of 1% per block, the optimal fee is set at 7.07% of the payment … (resulting [waiting time is] around seven blocks), while it is set at 1%/1000=0.001% in the decentralised game. The resulting waiting time is 50,000 blocks, equivalent to almost a year!”</blockquote><p>The key implication of these attacks is that: (1) Unless the block reward dominates transaction fees (meaning inflation is high), rational economic miners will undermine the network security; (2) unless miner rewards are very high (and transactions are very expensive), then wait time to finality will be prohibitive; and (3) proof-of-stake systems with proper incentive structures will very easily exceed the security level of Bitcoin/proof-of-work. All of these tradeoffs render PoW prohibitively limited for store-of-value, payments, and other use cases</p><h4><strong>Empirical evidence of poor economic security: Successful 51% attacks</strong></h4><p>It is becoming increasingly easier to<a href="https://www.crypto51.app/"> rent</a><a href="https://medium.com/@HusamABBOUD/the-realistic-lucrative-case-of-ethereum-classic-attack-with-1mm-today-8fa0430a7c25"> hash power</a> to attack a PoW network, even in Bitcoin (less than $300k for a 1 hour attack). This has resulted in several successful 51% attacks over the past 18 months (and<a href="https://bitcointalk.org/index.php?topic=332584.0"> historically</a>).</p><p>Victims include:</p><p>— <a href="https://blog.coinbase.com/ethereum-classic-etc-is-currently-being-51-attacked-33be13ce32de">Ethereum Classic</a></p><p>— Verge<a href="https://www.ccn.com/1-1-million-malicious-miner-exploits-verge-network-for-seven-figure-payday/"> 1</a><a href="https://www.ccn.com/privacy-coin-verge-succumbs-to-51-attack-again/"> 2</a><a href="https://blockexplorer.com/news/third-times-a-charm-verge-suffers-51-attack-yet-again/"> 3</a></p><p>— <a href="http://fortune.com/2018/05/29/bitcoin-gold-hack/">Bitcoin Gold</a></p><p>— <a href="https://www.ccn.com/vertcoin-hit-by-51-attack-allegedly-lost-100000-in-double-spending/">Vertcoin</a></p><h4>1b. PoW has an inherent inclination toward centralization</h4><p>There are several strong factors that result in a heavy incentive to centralize in any proof-of-work consensus:</p><p>— High cost of mining and sensitivity to operating and input costs like electricity price lead to huge economies of scale. Geographic differences exacerbate this phenomenon.</p><p>— Validators are inherently in a race with one another, which has resulted in the development of specialized<a href="http://bitcoinfibre.org/"> relay networks</a> between cartels of miners.</p><p>— Selfish miner attacks incentivize economically rational miners to form large cartels that can undermine other miner groups.</p><p>— The lottery effect of miner rewards leads to high variance of payout, incentivizing miners to consolidate or form pools that are centrally directed. This idea can be quantified by noting that a miner’s risk adjusted returns (analogous to<a href="https://en.wikipedia.org/wiki/Sharpe_ratio"> Sharpe ratio</a>) are monotonically increasing with share of mining power: <em>x/sqrt(x*(1-x))</em>. This means that it is <em>always</em> better unilaterally to consolidate.</p><p>Economic analysis of these factors suggests that<a href="https://object.cato.org/sites/cato.org/files/serials/files/cato-journal/2015/5/cj-v35n2-12.pdf"> mining is a natural monopoly</a>.</p><p>Centralization tendencies have proven to be empirically true in<a href="https://btc.com/stats/pool?pool_mode=week"> Bitcoin</a>,<a href="https://www.etherchain.org/charts/topMiners"> Ethereum</a>, and other proof-of-work networks, where large pools dominate hash power. This has been confirmed by<a href="http://hackingdistributed.com/2018/01/15/decentralization-bitcoin-ethereum/"> academic</a><a href="https://arxiv.org/pdf/1801.03998.pdf"> researchers</a>:</p><blockquote>“Both Bitcoin and Ethereum mining are very centralized, with the top four miners in Bitcoin and the top three miners in Ethereum controlling more than 50% of the hash rate. The entire blockchain for both systems is determined by fewer than 20 mining entities. While traditional Byzantine quorum systems operate in a different model than Bitcoin and Ethereum, a Byzantine quorum system with 20 nodes would be more decentralized than Bitcoin or Ethereum with significantly fewer resource costs. Of course, the design of a quorum protocol that provides open participation, while fairly selecting 20 nodes to sequence transactions, is non-trivial.”¹</blockquote><p>This practical centralization has led to substantial agency and governance problems, such as in the Bitcoin Cash/Bitcoin SV fork.</p><p>Ultimately, centralization carries a high risk for any decentralized network whose security relies on assumptions of good validator behavior. Even though miners (for the most part) have acted relatively benignly, they are entirely unregulated and far less trustworthy than traditional institutions. Replacing an established, regulated, public system with an unregulated, anonymous one where the miners can arbitrarily attack the system is a poor tradeoff. This is particularly an issue due to the lack of economic incentives in place to police behavior.</p><p>The factors that result in a centralized equilibrium in proof-of-work either do not exist or are substantially attenuated in proof-of-stake.</p><h4>1c. Nakamoto Consensus is not fully Byzantine fault tolerant; even with modifications it can only provide probabilistic guarantees</h4><p>A basic requirement of a trustless network is that it provides a sufficient level of Byzantine fault tolerance (BFT); otherwise, how can you trust a network with potentially malicious participants? It has been known for several years that vanilla Nakamoto Consensus (as used by Bitcoin)<a href="https://eprint.iacr.org/2014/765.pdf"> is not truly BFT</a>, even <em>probabilistically</em>:</p><blockquote>“It is easy to see that Validity cannot be guaranteed with overwhelming probability unless the hashing power of the adversary is negligible compared to the honest players, i.e., t/n is negligible. This is because in case the adversary finds a solution first, then every honest player will extend the adversary’s solution and switch to the adversarial input hence abandoning the original input. While one can still show that Validity can be ensured with non-zero probability (and thus the protocol fails gracefully assuming honest majority), [Nakamoto Consensus] falls short from providing a solution to BA [Byzantine Agreement].”</blockquote><p>See also<a href="https://www.amazon.com/Science-Blockchain-Roger-Wattenhofer-ebook/dp/B01B5WMUT0/ref=tmm_kin_swatch_0?_encoding=UTF8&amp;qid=&amp;sr=&amp;dpID=51d7kmvFpzL&amp;preST=_SY445_QL70_&amp;dpSrc=detail"> Wattenhofer, R. The Science of the Blockchain</a>, which argues that Bitcoin’s security properties are quite weak.</p><p>It has been shown that Nakamoto Consensus can be modified to provide probabilistic BFT (albeit with a 33% BFT, rather than the oft-quoted 51%), but given the governance issues with Bitcoin, it seems unlikely that any attempt to majorly change the protocol will be unsuccessful.</p><p>While Bitcoin has not suffered major double spend attacks yet (that we know of), relying on the benevolence of unregulated and low trust miners is an exceedingly poor security model. Who is to say they won’t attack the system going forward?</p><h4>1d. Even ignoring the correlation issue, the economic power securing PoW is lower than in the equivalent in PoS</h4><p>While Bitcoin appears to have a huge lead in economic security based purely on the value of the mining equipment behind it, <em>in expectation</em> an equivalently valued PoS network would have much more security, even ignoring any incentive alignments.</p><p>This is most easily seen from an investment lens. A validator on a PoW and PoS network both need to make an economic investment to gain validating power: mining equipment in the former, stake in the latter. The aggregate investment by the validators on the network must correspond to an economically efficient return on investment (ROI) for a given total validator reward (inflation + transaction fees), taking into account the costs of validation and risk (including variance of income).</p><p>PoS comes out ahead across all of these factors, thus ensuring that a PoS network will have more economic value securing the network than a PoW network, <em>ceteris paribus</em>. PoW has substantially higher costs due to the depreciation of assets and PoW security: as better miners develop, the value of existing miners falls, and the electricity costs that contribute to the security of historical blocks also falls (because it costs less to replicate that same work). Conversely, PoS stake does not depreciate <em>per se</em> (although it does have market exposure). PoW also has substantially higher income variance compared to a well-structured PoS network that eliminates the lottery effect present in PoW.</p><p>It is also relatively easy for a new staking-based network to bootstrap quickly to very high economic security and overcome the security advantage of Bitcoin. Specifically, the very existence of network value (and any subsequent accrual) confers security immediately without additional capex. In other words, mining hash power is definitively <em>not</em> a “killer feature”, as some claim.</p><p>Putting all these factors together in an example, let’s say a new network with PoS + slashing + classical non-probabilistic consensus is worth $100mm (at market price). It is plausible that validators could stake $30mm. If the network is tolerant of up to 1/3 malicious validators by stake, then the economic security of a particular transaction is $10mm. Conversely, as stated previously, a double spend in Bitcoin (by far the most valuable PoW network) could be plausibly achieved for well less than $1mm by renting sufficient hash power for several hours. This is a <em>6,300x difference in security</em> when normalizing for network value!</p><h4>1e. Practical security issues: Simple Payment Verification</h4><p>The vast majority of users on any network cannot be expected to run full nodes and maintain complete local copies of the entire network state. Instead, such users will rely on light wallets that connect to third-party full nodes and ask for updated network information. When the user receives a payment and wishes to confirm that it was processed correctly, they can ask the node for the latest chain and look for the transaction in the appropriate block. This is called simple payment verification (SPV).</p><p>While a full node does not need to trust its peers on the network (assuming non-pathological connectivity) because it is monitoring the blockchain real-time, a SPV user in a PoW network does need to trust that the full node did, in fact, send the longest chain and not some false fork. This can be mitigated by asking several nodes for the longest chain and calculating which has the most accumulated work. While the probability of successful duping decreases exponentially with the number of nodes queried (unless attackers have overloaded the user’s peer table, which is far easier in a SPV context than in a full node), verifying the work is an expensive process on limited hardware.</p><p>There has been some research into making this process more efficient (see, for example,<a href="https://eprint.iacr.org/2017/963.pdf"> proofs-of-proof-of-work</a>), but the problem remains the same: unless you are running a full node, there is some trust required. This is a consequence of the unpredictability of who will add the next block to the chain and the probabilistic consensus. In a PoS context (when properly constructed), a recipient of a transaction can verify that the transaction is completely final and genuine with a simple certificate showing the appropriate validator signatures.² This problem cannot be fixed in PoW, only mitigated, while it can be solved in PoS.</p><h3>2. Severely limited scalability</h3><p><em>Note: here, using PoW and Nakamoto consensus interchangeably gets a bit awkward. There are some PoW-compatible consensus algorithms that mitigate some of the scalability issues with Nakamoto consensus, such as </em><a href="https://www.usenix.org/system/files/conference/usenixsecurity16/sec16_paper_kokoris-kogias.pdf"><em>ByzCoin</em></a><em>, </em><a href="https://eprint.iacr.org/2016/1159.pdf"><em>SPECTRE</em></a><em>, </em><a href="https://eprint.iacr.org/2018/992.pdf"><em>Prism</em></a><em>, and </em><a href="https://eprint.iacr.org/2016/917.pdf"><em>Hybrid Consensus</em></a><em>. Nevertheless, these systems all suffer from all the issues presented in other sections and still have suboptimal performance due to using PoW insofar as resources used to compute PoW (which dominates overall computations) could otherwise be used for direct transaction validation. Nakamoto consensus networks like Bitcoin and Litecoin are the primary subjects of this section.</em></p><h4>2a. Inherent mathematical limits on scalability due to block propagation before security guarantees break down</h4><p>Proof-of-work has mathematical limits on scalability that are substantially lower than hardware limits. Specifically, <em>the combination of block frequency and block size must be bounded to maintain security</em>. This bounds throughput to low hundreds of TPS, <strong>far below requirements for many practical applications</strong>. These bounds <em>cannot</em> be avoided in proof-of-work, but they can in proof-of-stake. Attempts to increase capacity by decreasing the block time or increasing the block size very quickly become unintentional but serious threats to security.</p><p>The fact that PoW is inherently bound by architecture rather than hardware was first rigorously demonstrated mathematically by a<a href="https://fc16.ifca.ai/bitcoin/papers/CDE+16.pdf"> consortium of researchers in 2016</a>:</p><blockquote>“We analyze how fundamental and circumstantial bottlenecks in Bitcoin limit the ability of its current peer-to-peer overlay network to support substantially higher throughputs and lower latencies. Our results suggest that reparameterization of block size and intervals should be viewed only as a first increment toward achieving next-generation, high-load blockchain protocols, and major advances will additionally require a basic rethinking of technical approaches.”</blockquote><p>The basic idea is that security in PoW relies on as many miners as possible working on the longest chain. Otherwise, they will be working on competing or stale chains (i.e., forks of the true longest chain) and not contributing to the network security. This, in turn, limits the network throughput according to the speed at which data can propagate sufficiently through the peer-to-peer gossip network. Based on empirical observations of full node hardware and propagation speeds, it is possible to estimate reasonable limits on maximum safe throughput and latency:</p><blockquote>“Observation 1 (Throughput limit) Given the current overlay network and today’s 10 minute average block interval, the block size should not exceed 4MB. A 4MB block size corresponds to a throughput of at most 27 transactions/sec.</blockquote><blockquote>Observation 2 (Latency limit) Given today’s overlay network, to retain at least 90% effective throughput and fully utilize the bandwidth of the network, the block interval should not be significantly smaller than 12s.”</blockquote><p>These results were also nicely summarized in the<a href="https://bloxroute.com/wp-content/uploads/2018/03/bloXroute-whitepaper.pdf"> bloxRoute whitepaper</a>:</p><blockquote>“An attempt to increase the block size (B) by a factor of 10, which would increase system capacity to ∼ 30 TPS, would increase the propagation time to t90th = 116 seconds. This in turn would increase the probability for a fork to occur to P(fork) = 17.58%, which is unacceptable for real-world usability. More importantly, it would increase the probability for a fork to remain unresolved for 6 blocks by a factor of 600,000, and users will have to wait for 14 blocks to be mined to maintain the same level of confidence. <em>Scaling the system to ∼ 300 TPS, which is at least one order of magnitude too small for wide real-world adoption, would keep the blockchain at a continuous state of fork</em>.”</blockquote><h4>2b. Nakamoto Consensus requires a strict synchronicity assumption that bounds performance</h4><p>To ensure basic security properties such as consistency, Nakamoto consensus requires confirmation times to be proportional to a conservative upper bound on network delay. In practice, this upper bound must be large, <strong>substantially limiting potential performance</strong>.</p><p><a href="http://drops.dagstuhl.de/opus/volltexte/2017/8004/pdf/LIPIcs-DISC-2017-39.pdf">Pass and Shi</a> summarized the issues that this presents in practice (emphasis added):</p><blockquote>“Remarkedly, the famous Nakamoto blockchain (that underlies Bitcoin) also works only in the synchronous model where the protocol must know an a-priori upper bound of the network delay (henceforth denoted ∆). Otherwise, if a delay upper bound ∆ is unknown, Pass et al. [<a href="https://eprint.iacr.org/2016/454.pdf">link</a>] shows that Nakamoto blockchain’s security can be broken. Pass et al. also show that the expected block interval of Nakamoto’s blockchain must be set to be, roughly speaking, a constant factor larger than ∆ for consistency.</blockquote><blockquote><em>Relying [on] a synchronous model, however, can be undesirable in practice</em>. Since the synchrony parameter ∆ must be set conservatively to leave a sufficient safety margin, in practice the actual network’s delay is typically much better than this pessimistic upper bound. <em>Unfortunately, any protocol that must wait for at least one synchronous round (or one blockinterval) for transaction confirmation cannot benefit from the network’s actual performance</em>.”</blockquote><p>This strong synchronicity requirement is not only an issue with performance, but also with security. <a href="https://eprint.iacr.org/2015/578.pdf">Gervais et al.</a>, for example, demonstrated several potential attack vectors related to the need for fast propagation of blocks in Bitcoin:</p><blockquote>“We show that current scalability measures adopted by Bitcoin come at odds with the security of the system. More specifically, we show that an adversary can exploit these measures in order to effectively delay the propagation of transactions and blocks to specific nodes — without causing a network partitioning in the system. <em>We show that this allows the adversary to easily mount Denial-of-Service attacks, considerably increase its mining advantage in the network, and double-spend transactions in spite of the current countermeasures adopted by Bitcoin</em>.”</blockquote><h4>2c. Practical incompatibility with scaling layers like Lightning and Sharding</h4><p>Given the impossibility of safely scaling the first layer, much excitement has been centered around second layers and payment channels like Lightning Network as a way of achieving scalability while using the first layer as primarily a settlement layer for opening and closing channels.</p><p>There are several fundamental issues with assuming Lightning and other second layers will be a silver bullet solution for all payments needs. These include channel drift towards merchants, centralization concerns,<a href="https://twitter.com/el33th4xor/status/1011296476637605890?s=21"> routing difficulties for non-trivial payment amounts</a>, and lockups of value in channels for extended periods of time. While these solutions are very promising for some targeted use cases (e.g. micropayments), in general they are a very incomplete solution for general payments, including some of the most compelling use cases like point-of-sale.</p><p>Since this result is generally well established, it is more important to note that first layer scalability is required for even these narrow use cases from both practical and security perspectives. That is, higher-layer throughput is a function of first layer capacity, not independent as many assume, and claims of “infinite scalability” fail to recognize this reality. As such, Bitcoin/proof-of-work is not <em>practically </em>compatible with second layer scalability.</p><p>I explained these practical and theoretical issues of second layers in a PoW context, where the first layer cannot provide adequate supporting capacity, in<a href="https://medium.com/logos-network/second-layers-are-not-the-scalability-silver-bullet-a9d262bc816c"> a previous article</a>, to which I refer readers for more detail.</p><p>First layer scalability solutions also present a challenge. Most notably, sharding has proven to be very difficult under the legacy crypto model (blockchain + proof-of-work consensus), and the outlook for viable sharding is bleak without drastic changes (e.g. proof-of-stake with a more sharding compatible data structure). Regardless, sharding can only<a href="https://medium.com/logos-network/sharding-how-many-shards-are-safe-bc361c487083"> safely scale a first layer 10–50x given statistical realities</a> (despite claims to the contrary that assume the wrong statistical distribution), and thus is inadequate in and of itself.</p><h3>3. Other Issues</h3><h4>3a. Energy inefficiency</h4><p>The environmental issues with proof-of-work are well known. Less appreciated is the practical usability impact it has. By requiring a large amount of energy per transaction, the fee per transaction (either direct or indirect via inflation) must be high enough to make up for that real cost. Lowering the energy per transaction of course commensurately lowers the security of each transaction. This makes proof-of-work far less efficient than proof-of-stake.</p><p>A<a href="https://www.cell.com/joule/pdf/S2542-4351(18)30177-6.pdf"> recent economic analysis</a> contextualized this energy inefficiency (emphasis added):</p><blockquote>“The Bitcoin network consumes at least 2.55 GW of electricity currently, and that it could reach a consumption of 7.67 GW in the future, <strong>making it comparable with countries such as Ireland (3.1 GW) and Austria (8.2 GW)</strong>. Additionally, economic models tell us that Bitcoin’s electricity consumption will gravitate toward the latter figure. A look at Bitcoin miner production estimates suggests that this figure could already be reached in 2018. With the Bitcoin network processing just 200,000 transactions per day, this means that the average electricity consumed per transaction equals at least 300 kWh, and could exceed 900 kWh per transaction by the end of 2018 … Bitcoin has a big problem, and it is growing fast.”</blockquote><h4>3b. Expense</h4><p>The inefficiency of the Bitcoin in terms of energy and other factors results in a significant cost: the average Bitcoin transaction costs about $25 in total validator reward. The bulk of this cost is paid indirectly via inflation, but it is a real cost nonetheless, as I discussed in a <a href="https://medium.com/logos-network/crypto-transaction-fee-economics-primer-948d1d93a4f5">recent article</a>. As many validators have recently halted their activity due to price declines, we can be pretty sure that the current cost is close to the marginal cost of validation, i.e. there is no economic rent paid to miners. This means that we would not expect, at a mature equilibrium, substantially lower fees.</p><p>This expense is prohibitive for the vast majority of potential use cases and severely limits the usability of Bitcoin or any other successful PoW network.</p><h4>3c. Confirmation latency</h4><p>PoW consensus is, by construction, probabilistic. That is, a transaction is not finalized when it is included in a block since there is always a chance that an alternative chain overtakes it. The rule of thumb in Bitcoin is to wait 6 blocks — waiting any less puts the recipient at serious risk of becoming a victim of a double spend attack. As stated previously, this analysis is faulty in that it assumes constant attacker power; more robust, game-theoretic analysis suggests that the appropriate waiting time (at a mature state of the network) is far longer.</p><p>Regardless, even at just 6 blocks, Bitcoin requires recipients to wait 1 hour for a transaction to be confirmed in expectation. Since accumulated hash power is linear in time, increasing the block time (as in Ethereum) does nothing to decrease the latency for a given security level.³</p><p>While latency on the order of 15 minutes to 1 hour is fine in some contexts (wires, remittances, etc), it is prohibitive in many practical payments contexts that could benefit the most from DLT. Furthermore, even for those use cases where such a wait time is acceptable, there is a potentially large cost of exposure to the volatility of Bitcoin or network token, which cannot be minimized below this confirmation window.</p><h3>Conclusion</h3><p>While Bitcoin and proof-of-work has proven invaluable in bootstrapping the public’s understanding and acceptance of the concept of cryptocurrencies, it is critically hamstrung as the backbone of a viable, long-term solution.</p><p>There are serious security issues inherent in proof-of-work, and the traditional framework for parameterizing risks has proven to be woefully inadequate given increased understanding of the game theory of crypto networks. This is exacerbated by poor economic incentive alignment and the rise of a robust hash power market, both of which make it relatively inexpensive to pull off a 51% attack. Even in the absence of malice, the variance of miner incentives and lack of a punishment mechanism make PoW unstable without significant, long-run inflation, calling into question its viability even just as a pure store of value. Other factors, such as centralization tendencies and questionable mathematical basis of Byzantine fault tolerance, further undermine security.</p><p>PoW also suffers from performance issues that are not correctable given our current understanding. The first layer is inherently limited by consensus rather than hardware at very low capacities of (generously) low hundreds of TPS or less, while first layer scaling solutions like sharding are largely incompatible with PoW. The limited capacity of the first layer in turn makes PoW networks unsuitable for second layers due to both practical and security considerations. Confirmation latency similarly is constrained at high levels of 15 minutes or more, without any way to meaningfully shorten it.</p><p>Finally, inefficiencies such as energy usage lead to high costs per transaction, even with relatively weak security.</p><p>Given the evidence, it is hard to conclude anything other than that Bitcoin and proof-of-work is not the long-term answer for the decentralized applications of tomorrow. An alternative solution that solves all of these issues is non-trivial to construct, but it is certainly possible. Currently, the most promising approach involves a proof-of-stake Sybil control mechanism coupled with a non-probabilistic, classical consensus algorithm with a strong economic incentive system that includes features like slashing and low reward variance. Such a system can dramatically improve capacity, reduce confirmation latency, and minimize fees and is compatible with additional scaling solutions like second layers and sharding.</p><p>Unfortunately, due to the value at stake, the crypto space has become increasingly ideological and loathe to accept results that challenge established views. It is critical that participants and stakeholders move beyond tribalism and reembrace the spirit of cooperative innovation that characterized the early days of crypto in order to deliver on the transformative potential that Bitcoin originally promised.</p><p>[1] This is what Logos achieves.</p><p>[2] For a concrete example, see the<a href="https://www.logos.network/whitepaper"> Logos white paper</a>.</p><p>[3] Increasing the block time does not decrease expected wait time in minutes at a given security parameter, but it does decrease variance.</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dd96d2775e99" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/why-proof-of-work-is-not-viable-in-the-long-term-dd96d2775e99">Why Proof-of-Work Is Not Viable in the Long-Term</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Crypto Transaction Fee Economics Primer]]></title>
            <link>https://medium.com/logos-network/crypto-transaction-fee-economics-primer-948d1d93a4f5?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/948d1d93a4f5</guid>
            <category><![CDATA[decentralization]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[blockchain-technology]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Wed, 06 Feb 2019 16:10:06 GMT</pubDate>
            <atom:updated>2019-02-08T01:18:30.603Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A 360 foundational explainer on crypto fees and the optimal fee model.</em></p><h4>Background:</h4><p>While much attention has been paid to the issue of capacity and scalability, arguably the larger hurdle to broad adoption of cryptocurrency is the steep transaction cost. Over the past 12 months, Bitcoin and Ethereum transaction fees have averaged more than $5 and $0.60, respectively, with peaks exceeding more than $20. As a result, a number of large merchants and payment service providers — <a href="https://www.cnbc.com/2017/12/07/valve-steam-drops-bitcoin-support-cites-high-fees-volatility.html">Steam</a>, <a href="https://stripe.com/blog/ending-bitcoin-support">Stripe</a>, <a href="https://www.ccn.com/microsoft-no-longer-accepts-bitcoin-customer-support-confirms/">Microsoft</a> — dropped Bitcoin payment integrations. While cryptocurrencies were originally envisioned as a leap forward in payments, they often cost substantially more than conventional payments rails. Along with the unacceptably slow speeds, this cost makes crypto a strictly inferior payment option for most use cases.</p><p>If these networks are not useful <em>now</em> due to cost, how can they possibly gain enough traction to make capacity matter?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/465/0*DZpOetzTYIK7zP9E" /><figcaption><em>LTM average transaction fees exceeded $5.00 and $0.60 for Bitcoin and Ethereum, respectively.</em></figcaption></figure><p>The economically astute reader will note that, in the traditional crypto model, capacity and cost are, in fact, one and the same. Bitcoin, Ethereum, and most legacy crypto networks have a dynamic, market-determined network fee. That is, users sending transactions choose the amount of fee that they wish to pay, and then, validators (miners, in the case of proof-of-work networks) choose which transactions they wish to include. Naturally, validators will tend to choose the transactions with the highest fees because that maximizes their profit.¹</p><p>This results in a classic supply-demand equilibrium, where the transaction fee is the market clearing rate where supply meets demand. There is a scarce resource — the space in the next block, or, more generally, the “slots” in the next set of transactions to be processed — and as demand for that resource increases, so does its price.</p><p>As this logic suggests, the increases in fees in Ethereum and Bitcoin occurred during periods of high transaction demand where capacity constraints were binding. Every big ICO or DApp launch (CryptoKitties being the most notable example) saw a commensurate spike in Ethereum transactions, and thus transaction fees, while the Bitcoin mempool <a href="https://jochen-hoenicke.de/queue/#0,all">spiked massively</a> during high excitement periods. The result was highly congested networks where transactions with “normal” fees often waited days or weeks for their transactions to be processed.</p><p>Alarmingly, this rise in prices occurred at effectively zero adoption of either Bitcoin or Ethereum. The fact that these networks cost so much with such small volume highlights their severe technical limitations vis-à-vis both capacity and cost. The fact that other networks have not seen a similar spike in fees is a testament to their utter lack of adoption, beneath even that of Bitcoin or Ethereum.</p><p>The issue of fees has not gone unnoticed. A number of newer networks have proposed a variety of alternative fee structures, each claiming to “solve” the fee problem at capacity. Most tantalizing, several networks, including EOS, IOTA, and Nano, claim to have <em>no fees </em>by design. At first blush, this sounds great — if the protocol prohibits fees, then they can’t rise! Like many claims in crypto, unfortunately, “zero fees” turns out to be too good to be true.</p><p>There has been some sparse academic research into fees — see, for example, section 7.3 of <a href="https://dtr.org/wp-content/uploads/2019/01/2019-01-16-Decentralized-Payment-Systems-Principles-and-Design.pdf">this paper</a> for a good review. Unfortunately, most research makes assumptions about conditions to optimize around, such as some concept of “fairness” or predictability, without justifying their objective functions. While such results may sound good at first blush, how can we know which outcome, among many plausibly good outcomes, is indeed the best? A more deductive and analytical approach is needed.</p><p>What then, is the answer to the existential problem of fees? And how can we go about thinking through the pros of cons of the various proposed fee models to figure out which one is optimal? Luckily, these questions are relatively straightforward to answer with basic economics.</p><h4>Fee Economics 101</h4><p>It turns out that crypto network costs are quite simple from a microeconomics standpoint.² There is a single, self-contained market for a uniform good — a transaction validation slot — with a set of effectively anonymous buyers and sellers. There are relatively few exogenous factors or externalities that need to be taken into account. <strong>While standard microeconomics models must be caveated heavily in normal markets, the simple, insulated structure of crypto fee markets results in a straightforward analysis.</strong></p><p>The way economics analyzes this type of simple market is through <a href="https://en.wikipedia.org/wiki/Supply_and_demand">supply and demand curves</a>. I won’t review the basics, but I will note several interesting considerations for crypto fee markets. I’ll assume you know how supply and demand curves are used to find market prices and what consumer and producer surplus and net welfare mean — if you don’t, you should review the Wikipedia article.</p><p>The standard crypto fee market involves a set of producers (validators) supplying validation slots that are bid for by consumers (the users of the network).</p><p>Interestingly, the crypto market differs from a simple supply-demand market in that not everyone gets the good at the same price. If you submit a transaction with a fee of $1 and the market clearing rate is $0.50, you will still pay $1. This ends up having no impact on the actual market dynamics — it just shifts consumer surplus to producer surplus.</p><p>The primary wrinkle for the crypto fee market is that supply is typically fixed — Bitcoin, for example, has one block of roughly 2,000 transactions every 10 minutes, or about 3 transactions per second. In general, there are only so many transactions per second that a given network can process. Typically, crypto observers assume this means that supply is <em>perfectly inelastic</em> — it can never increase or decrease.³</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*f0WJu6GZlmRjEN8j" /><figcaption><em>Most crypto networks have perfectly inelastic supply of validation slots</em></figcaption></figure><p>However, supply is only <em>perfectly inelastic</em> after a certain point. That is, supply is <em>capped</em>, but the producers in the market won’t necessarily supply up to that cap for zero cost. Specifically, the supply curve is upward sloping until it hits the cap (variable costs of validation, typically quite small) and has an initial jump (fixed costs of validation, often quite large). In Bitcoin, for example, fixed validation cost is very high, so validators would not process transactions for free.</p><p>In reality, fixed costs do not impact the marginal production but rather the decision whether to produce (validate) at all. The below chart roughly (but, strictly speaking, inaccurately) accounts for the effect of fixed costs by shifting up the supply curve. It <em>does</em> accurately show the net social welfare, which is our chief concern. As we will see below, due to a variety of other factors, these fixed costs can be abstracted away into a pure decision of whether or not to validate.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*eYE_4iMOVES5ZLwn" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*iHoFLPR1PWnJHsHx" /><figcaption><em>Updated supply and demand curves reflecting approximate fixed and variable costs of validation</em></figcaption></figure><p>The longtime Bitcoin enthusiast will object to this statement — for many years, transactions with no fee were processed quite regularly in Bitcoin. There are a couple of factors that impact this simple supply curve and explain this discrepancy.</p><p>First, traditional cryptocurrencies also include an inflationary validator payment (the block reward). This has the effect of shifting the supply curve down, just as a government subsidy shifts down a market supply curve. If the inflation is large enough, the initial sloped section of the supply curve all falls below zero, meaning that validators will process zero-fee transactions (typically, networks will not allow negative transaction fees). As we will see later, however, inflation comes at a cost.</p><p>Second is the more nebulous concept of crypto enthusiasm and benevolence. Many people derive utility or some uneconomic benefit from participating in and supporting the network. This is seen in the people who run full nodes on Bitcoin and Ethereum, incurring the cost of running a node without any economic benefit.⁴ Sometimes, they can actually derive some direct economic benefit — for example, a blockchain explorer might run a full node and serve ads on the site, or a merchant might run a full node to accept crypto payments and save on transaction fees. Regardless, the “enthusiasm” factor has the same effect of the inflation subsidy of shifting the supply curve down.⁵ There is a limit though — once the cost exceeds their utility, these “benevolent” participants will stop running their nodes.</p><p>An underappreciated reality is that <strong><em>there is a real economic cost of running a node</em></strong>. Generally, this cost scales with transaction volume⁶ CPU, data storage, electricity, and bandwidth are costly goods, and you need more of all of them for each additional transaction you process each second. Due to this increasing cost, <strong>it is economically unsound to rely on the goodwill of participants</strong>. We’ll explore this concept more in a bit.</p><p>This variable cost manifests itself as an upward slope in the supply curve. In a well-designed crypto network, the slope of the supply curve until it hits the capacity limit should be quite shallow, reflecting the low <em>marginal</em> cost of validation. That is, as long as you are already set up to validate, the incremental cost of validating one more transaction is quite low. <strong>However, the aggregate cost of processing a large number of transactions is non-negligible</strong>.</p><p>There are a few offsetting effects that shift the supply curve up. These include the upfront costs of both staking and mining, such as buying network tokens or mining equipment or the time and dollar cost of securing your operation, preventing DDoS attacks, and so on. Validators will demand a certain minimum payment to supply their service, which is reflected in the fixed costs mentioned previously.</p><p>Nevertheless, the net effect of these factors is that the entirety of the sloping part of the supply curve is well below zero in a network like Bitcoin or Ethereum.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*23XJMt0WApAX7LiH" /><figcaption><em>Net effect of inflation subsidy plus general enthusiasm shifts the supply curve down. Typically, this means we can use the basic perfectly inelastic model since the upward sloping part of the curve is all below zero.</em></figcaption></figure><p>All this applies to traditional crypto networks where there is a true fixed capacity. In a better-designed network where capacity is determined by validator hardware rather than by network limitations, there will actually be a somewhat upward sloping supply curve, since validators can upgrade their hardware if the fees rise commensurately. The upward sloping curve also can reflect the differing costs of the various validators — some may have cheaper access to electricity or hardware, for example.⁷</p><p>We’ll assume the fixed supply market in this article for simplicity. In general, though, the fixed supply is a conservative assumption in that it makes the arguments for the alternative (non-dynamic) fee models look better than under a variable supply assumption.⁸ <strong>That is, all the conclusions in this article will apply equally or more to a variable supply market</strong>.</p><p>Also, for simplicity, we will assume that transaction finality occurs concurrently with consensus. This is not the case with proof-of-work and other probabilistic consensus algorithms — in Bitcoin, for example, there is always a small chance that the transaction can be reversed, which declines exponentially with number of blocks since it was processed (at least probabilistically). There are several perverse dynamics in the fee market under a probabilistic consensus, such as those detailed in a recent <a href="https://www.bis.org/publ/work765.pdf">BIS paper</a>, that significantly complicate this analysis. Since these issues are primarily security rather than economic considerations, and more robust non-probabilistic consensus models offer strict improvements over the probabilistic ones, it is sufficient here to note that they do not materially impact the broader conclusions.</p><p>An interesting alternative approach is to model the fee market as <a href="https://en.wikipedia.org/wiki/Vickrey%E2%80%93Clarke%E2%80%93Groves_auction">an auction</a> (albeit a non-standard one where there are repeated opportunities), but we will leave such analysis for a future post. The conclusions are substantially the same as the basic supply-demand model.</p><p>This simple fixed supply model provides a basis that now allows us to dig into the various models for fees.</p><h4>Fee models:</h4><p>There are four main basic fee models:</p><ol><li>No fees</li><li>Fixed fees</li><li>Inflation fees</li><li>Dynamic market fees</li></ol><p>All other fee models boil down to either one or a combination of the four basic models.</p><p>From an economics standpoint, it is easiest to think of the dynamic market fees as the “base case” that matches the standard supply-demand analysis outlined above. The other three cases correspond to various shifts or changes in the supply and demand curves, with implications for overall efficiency and global utility or welfare.</p><p>We’ll start with the no-fee model, which will also serve to illustrate as the most extreme example of several problems that affect both the fixed and inflation fee models.</p><h4>Zero</h4><p>The fee model that has made the biggest splash recently is the zero-fee model, with IOTA, Nano, and others claiming to have no fees. The supposed value proposition is, of course, compelling: if there are no fees, then anyone can use the network, no matter the transaction size, and never have to worry about fees. <strong>Unfortunately, there are several flaws with the no-fee model that make it unviable in a long-term, healthy network</strong>.</p><h4><strong>The True Cost of Transacting</strong></h4><p>First, the <strong>true cost of transacting in a healthy network can never be zero</strong>. At a minimum, there is some computational (and thus electric) cost to sending out a transaction.</p><p>Of course, if a transaction only involves a cryptographic signature, then this computational cost is virtually zero. However, any pseudo-anonymous network with genuinely zero fees will inevitably be spammed until it becomes unusable (any zero-cost long-term mitigations require some permissioning that breaks the fundamental value proposition of a cryptocurrency). Thus, the network requires some economic spam-prevention measure. Commonly, this is in the form of a proof-of-work proof, but this is economically equivalent to a direct fee since computing the proof-of-work has a non-trivial cost (otherwise, it would be useless at preventing spam).</p><p>While dynamic fees are represented in the demand curve itself, these hidden fees can be represented by a downward shift in the demand curve since they decrease the net utility a consumer gets from sending a transaction. Regardless, from a consumer’s perspective, they can be viewed as equivalent costs to transacting, and all popular zero-fee networks are just direct fee networks with a different name. There is a significant difference from the validator’s perspective, however, since they are not receiving any compensation.</p><h4><strong>Tragedy of the Commons and Free Riders</strong></h4><p>The obvious question that arises in a network with no validator reward is why anyone will validate on the network if they do not receive any payment?</p><p>IOTA answers this by forcing end users to validate other transactions in their own transaction, which effectively ends up being an additional hidden fee. More importantly, the resulting consensus model is insecure,⁹ so it is not a valid example of a viable decentralized network fee model.</p><p>Other networks, notably Nano, claim that “Average Joe” network participants will run full validator nodes since they use the network and would like to maintain it, rather than simply relying on light wallets and third-party nodes. While this may seem plausible on the surface, this argument has long been refuted by the Tragedy of the Commons result.</p><p>The <a href="https://en.wikipedia.org/wiki/Tragedy_of_the_commons">Tragedy of the Commons</a> is a simple economic phenomenon that illustrates the issues with common ownership of an asset or good that needs to be maintained. In its classic formulation, a town has a common pasture area where its residents can let their animals graze. This is a good outcome for everyone involved as it is much more efficient than everyone keeping their own grazing grounds. However, the individual townspeople have little incentive to contribute to the upkeep of the commons, resulting in <a href="https://en.wikipedia.org/wiki/Free-rider_problem"><em>free riders</em></a>. These people observe that they can pay nothing to maintain the commons (and treat it more roughly than they would their own property) and still enjoy the benefits of the common resource.</p><p>Each person who decides to free ride increases the cost of upkeep for everyone who does pay since the same fixed upkeep cost must be met, resulting in a positive feedback loop. Of course, the logical end result is virtually no one pays for the upkeep, (anyone who does decide to pay is effectively left “holding the bag”) and the commons becomes unusable for everyone. Like in the related <a href="https://en.wikipedia.org/wiki/Prisoner%27s_dilemma#Multiplayer_dilemmas">Prisoner’s Dilemma</a>, the game theoretically dominant individual strategy of not paying leads to a worse outcome for everyone involved.</p><p>More generally, given economically rational actors and the absence of an enforcement mechanism, any common, shared resource will logically be abused by free riders. <strong>The key lesson of the Tragedy of the Commons is that altruism or enthusiasm alone is not enough to maintain a common good</strong>.</p><p>This is an empirically proven phenomenon that everyone should be able to identify in their own lives; it comes as a shock to no one that people abuse and <a href="https://en.wikipedia.org/wiki/Overexploitation">exploit</a> common resources like parks or public transportation more than their own goods.</p><p>There are a few mitigations to such an outcome. Often, a tax is levied on the entire community to pay for upkeep. Alternatively, a toll or fee is charged for use of the resource. Both of these approaches are inherently contrary to a zero-fee network. Social pressures play an important role in preventing the worst abuses, as free riders can be labeled as such and will have to deal with the consequent stigma, but this too is contrary to a pseudo-anonymous, open network.</p><p>As an open, trustless, peer-to-peer system, a crypto network can be viewed as a form of a commons. There is a certain amount of work (transaction validation) that needs to be done to maintain a network and make it usable to the benefit of all. The fewer people who do this work, the worse the network is for everyone — transactions may be slower, security would be lower, and reliability would be worse — until it is useless for everyone.</p><p>Unfortunately, as mentioned previously, this work is expensive, with both fixed (acquiring network stake, setting up DDoS protection, ensuring sufficient reliability) and variable (data, bandwidth, CPU, electricity, etc.) costs. A network that relies on people doing this work out of genuine altruism or even based on the realization that everyone is better off if the resource is maintained will inevitably have a Tragedy of the Commons result, as economically rational actors will free ride off the other participants. In a pseudo-anonymous crypto network with a non-trivial number of transactions (and thus a non-trivial cost of validation or “upkeep”), this result is economically inevitable.</p><p>There are some trivial cases where a Tragedy of the Commons may not be the end outcome, particularly when the network is immature, as there are users who generally derive utility from participating in the network, even for free. A network probably is able to rely on enthusiasts if it only processes a few transactions per second, but once it starts processing thousands of transactions per second, the large cost of running a full node will likely soon outweigh the “soft” benefit those users get from running the node. How many people are willing to pay hundreds of dollars per month to run a node when they can free ride off others?</p><p>In existing crypto networks, the vast majority of users use light wallets that rely on other nodes. In a network with a well-designed consensus, this does not increase their risk of being a victim of a double spend or other attacks,¹⁰ so free riding is a dominant strategy.</p><p>Nano is the most prominent proponent of this “validation by desire to participate” strategy, but it ironically serves as an early example of the free rider phenomenon. Virtually no true third parties run network validators, with almost all voting power held by the development team or closely affiliated light wallets, applications, and exchanges. The largest validator on the network, Binance, often does not even bother voting in consensus, which has led to some<a href="https://medium.com/@ajohnson_83800/how-to-secure-your-nano-do-it-today-aee4c3b093bd"> practical issues and concerns</a>.</p><p>As with all aspects of network design, the focus should not be on what works now but what will work at maturity. The zero fee approach boils down to relying on user’s altruism or enthusiasm, which is not a sustainable growth model.</p><h4><strong>Validator Complexity and Security</strong></h4><p>Even if validators participated solely out of altruism, it is doubtful that “Average Joe” merchants and users would be able to maintain a robust network. A mature network processing thousands of TPS will require substantial expertise in building and maintaining a node, even in AWS or some other cloud service. Only so much can be optimized and streamlined, as threat models evolve, security demands following best practices, and networking itself is non-trivial. A regular company will only devote so many technical resources and <strong>unproductive working capital</strong> (in the case of a staking network) to run a secure, robust node on the network before giving up and just using a third party instead.</p><p>The fact that many networks do not have the basic protections that result in this complexity is not a signal that they are unneeded but rather a reflection that those networks are poorly designed, not significant enough, or not sufficiently decentralized to bother attacking. A poorly set up node that is subject to DDoS attacks and frequent downtime does not contribute to the network and may, in fact, decrease its efficiency.</p><h4><strong>Deadweight Loss in a Perfectly Inelastic Market</strong></h4><p>There is yet another issue at play: <a href="https://en.wikipedia.org/wiki/Deadweight_loss">deadweight loss</a>. Mandating zero fees is effectively equivalent to imposing a <a href="https://en.wikipedia.org/wiki/Price_ceiling">price ceiling</a> of zero on the market.</p><p>In a normal supply-demand market, price ceilings result in deadweight loss, which means total welfare (consumer surplus plus producer surplus) decreases. In other words, the economy overall is worse off. The same basic analysis has been demonstrated in <a href="https://web.stanford.edu/~diamondr/DMQ.pdf">rent control</a>, which results in fewer housing units overall and decreased mobility. The end result benefits the few people who get rent controlled apartments but hurts the rest of the population. Unlike rent control, however, where there are several complicating factors, the transaction market is extremely simple and self-contained, meaning that the deadweight loss analysis is straightforward.</p><p>The recent Econ 101 student will object by pointing out that the transaction market differs from normal markets due to having a perfectly inelastic (fixed) supply (by our previously stated assumption), which means that there would be no deadweight loss — only a transfer of surplus from producer to consumer.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/558/0*cDu_xURL2ngzrjh_" /><figcaption><em>In a classic economic model, there is no deadweight loss due to a price ceiling in a market with perfectly inelastic supply. However, this fails to recognize the “lottery effect,” where those that value the good the most are not necessarily the ones that receive the good. </em><a href="http://econ101help.com/consumer-surplus-producer-surplus-and-dead-weight-loss-with-inelastic-supply-curve/"><em>Source</em></a><em>.</em></figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*s_-vnXwJ_3geRQOb" /><figcaption><em>Once the “lottery effect” is taken into account, there are many consumers who receive the good who value it lower than the consumers without the price ceiling. This results in deadweight loss, which has practical usability issues in the zero-fee model. </em><a href="http://econ101help.com/consumer-surplus-producer-surplus-and-dead-weight-loss-with-inelastic-supply-curve/"><em>Source</em></a><em>.</em></figcaption></figure><p>However, this common economics result has a key flaw — it assumes that the consumers that get the good are those that value it most. In reality, price ceiling markets typically act as a lottery amongst all consumers whose reservation price exceeds the price ceiling. The result is that the consumers that actually receive the good have, as a whole, lower aggregate consumer surplus than the consumers that receive the good in a free market. Thus, there is a true, overall deadweight loss even if supply is fixed. This insight is reflected in the more sophisticated <a href="https://en.wikipedia.org/wiki/Deadweight_loss#Hicks_vs._Marshall">Marshallian approach</a>.¹¹</p><p>By way of example, imagine that the network can process 10 transactions per second, but there is demand for 100 transactions per second. In a zero-fee network, none of the demand will be priced out by a fee, so all 100 transactions will be competing for 10 slots each second. Some of these transactions may be very important and thus highly valued, such as paying a highway toll (a long delay could cause a huge traffic jam). Other transactions will have barely any value at all to the participants, such as a $0.0001 transfer. And yet, under the zero-fee system, both transactions will be equally valued and likely to get one of the slots. If all of the slots are taken by low utility transactions and many high utility transactions are excluded or delayed, then the network will fail to maximize overall utility. And since any distribution of transactions is almost certainly to have a large right tail, there are many more low utility transactions than high utility transactions, all equally likely to get a spot.</p><p><strong>This deadweight loss has real, practical implications — participants need a network that will reliably prioritize their most important transactions, and one that cannot reliably process your transaction due to this “lottery effect” will not be very useful. </strong>For example, a merchant whose transaction may or may not be processed and who has no discretion to increase the fee to prioritize the transaction may be left pending indefinitely — hardly an acceptable outcome!</p><p>Note that <em>high utility/value</em> should not be confused with <em>high amount</em>. It is very possible, for example, for a $1,000 transaction to only result in a $1 economic surplus between the participants, as is very common in low-margin businesses. Conversely, a $10 transaction might result in $9 of surplus. Any prioritization rule, such as give preference to larger transactions, is, at best, a first-order approximation with non-trivial loss.</p><p>In the best case, the system will result in an obfuscated, back-room fee market where validators are paid under the table for prioritizing specific users’ transactions. While this would prevent the “lottery effect,” it is a strictly worse version of the transparent, dynamic fee market and subject to abuse.</p><p>This is the essence of deadweight loss — a few lucky people benefit while everyone else has a suboptimal outcome.</p><p>As a result of these issues, the zero-fee model is critically flawed and cannot support a viable, long-term crypto network.</p><h4>Inflation subsidy</h4><p>A couple of the issues introduced by zero-fees boil down to the validators having no incentive to run and maintain the network. Several networks, most notably <a href="https://github.com/EOSIO/Documentation/blob/master/TechnicalWhitePaper.md#block-rewards">EOS</a>, have tried to maintain the benefits of zero-fees to consumers while paying the validators with newly-minted, inflationary tokens. Superficially, this seems to enable transactions of any size and provide a legitimate incentive to validators.</p><p>While it does help solve the free rider and Tragedy of the Commons issues, the inflation model too has several issues that make it unattractive for a sustainable network.</p><p>The first issue again is one of deadweight loss and the “lottery effect.” Since the entire cost of transacting is covered by the inflation subsidy, there is no ability for the market to differentiate between high and low utility transactions. Just as in the zero-fee model, this results in low utility transactions being given equal weight as high utility transactions in competing for limited spots that are lotteried off by the validators.</p><p>This “lottery effect” severely hampers the overall usefulness of the network at maturity. In the absence of any filter, extreme low-value transactions will dominate and take most of the available slots, leaving legitimate, higher value transactions waiting potentially indefinitely. As seen in Bitcoin and Ethereum over various points in 2017 when mempools ballooned, this effectively renders the network unusable to average users. Although they allow dynamic fees, their capacity is so constrained that these periods were equivalent to a lottery between high fee transactions.</p><p>The deadweight loss imposed by the inflation model can be viewed alternatively as <a href="https://www.thoughtco.com/analysis-of-a-subsidy-1147899">subsidy deadweight loss</a>. While our idealized fee market is different due to the fixed (perfectly inelastic) supply, the implication is the same: the market is being externally subsidized, and those funds have to come from somewhere.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/649/0*ZluGBrirCR0JRWdm" /><figcaption><em>Deadweight loss from a subsidy in a standard market (without perfectly inelastic supply). A subsidy costs the ecosystem as a whole more than it receives as benefit. </em><a href="https://www.thoughtco.com/analysis-of-a-subsidy-1147899"><em>Source</em></a></figcaption></figure><p>In a decentralized crypto network, the subsidy is not coming from the government but rather inflation in the token supply. This inflation subsidy is like a government subsidy in that the cost is ultimately borne by “society” as a whole — the holders of the tokens. Each additional token created erodes the value of the existing tokens since there is no change in total network value.</p><p>Thus, the inflation subsidy can be viewed as a proportional tax on existing token holders that is redistributed to the validators on the network. If a network is worth $1bn in aggregate, for example, a 10% inflation is equivalent to a 9.1% tax on token holders that distributes $91mm to validators.¹² This tax is independent of the actual utility the user derived from the network. In fact, a user who sends comparatively few but high-value transactions may even derive <em>less </em>overall utility from the network than a user with many low value (think a fraction of a cent) transactions who is able to grab the vast majority of available transaction slots. This perverse dynamic discourages users from holding tokens, with significant implications for network security and growth.</p><p>It is generally considered far preferable to directly charge individuals in proportion to their use of a shared, limited resource than subsidize via broad base taxation. An analogous scenario is road usage, where funding via tolls (whether direct or indirect via fuel taxes) is far more efficient than funding via general taxes.</p><p>So while the inflation-only model fixes several of the issues with the zero-fee model, it still results in a suboptimal outcome that severely impairs network usefulness at maturity.</p><h4>Fixed</h4><p>The third major alternative fee model that new networks have adopted is fixed fees, where every transaction pays the same fee that is set by the protocol. This has notably been implemented by <a href="https://www.stellar.org/developers/guides/concepts/fees.html">Stellar</a>, in which each network version fixes the fee and upgrades are required to modify that fee.</p><p>The reasoning behind fixed fees is admirable but ultimately flawed. At the surface, fixed fees appear to be more equitable than dynamic fees, which ostensibly allow a wealthier person to pay more fees to get their transaction confirmed faster. Unfortunately, this logic does not quite hold, since it makes all <em>transactions </em>rather than all <em>users</em> equitable. For example, a wealthy user who wants to send many small, low utility transactions is advantaged versus a poorer user who has a few but relatively higher utility transactions. Instead, fixed fees result in a worse outcome for the network users as a whole compared to dynamic fees.</p><p>Under the economic framework, fixed fees can be viewed as a “top down” management approach that results in deadweight loss relative to the “bottoms up”/”invisible hand” of the dynamic fee model.</p><p>There are three possible outcomes for the fee charged under the fixed fee model, all of which are strictly worse than the dynamic fee model.</p><p>First, the fee may be <em>higher</em> than the market clearing rate. This corresponds to a <a href="https://en.wikipedia.org/wiki/Price_floor">price floor</a>, which results in a deadweight loss. Practically speaking, it means that users of the network are paying too much for the service provided, to the benefit of the validators. Many socially valuable transactions that could use the network at the dynamic market rate are unable to use the network under the higher fixed rate, reducing overall social welfare. This is analogous to the purported impacts of minimum wage, where the people who hold on to their jobs benefit but a proportionally greater number of people lose their jobs, resulting in an overall negative outcome. While the minimum wage case is controversial due to confounding factors and complexity, the fee market, as we have observed previously, is extremely simple and self-contained, and thus, the conclusions are straightforward.</p><p>Alternatively, the fee may be <em>lower</em> than the market clearing rate. This corresponds to a <a href="https://en.wikipedia.org/wiki/Price_ceiling">price ceiling</a>, also with deadweight loss. The zero-fee model is merely the most extreme version of this price ceiling, and the same issues carry through with a non-zero price ceiling. If the fixed fee is below the cost of validation, the network will likely fail or become significantly impaired due to free riders causing a Tragedy of the Commons.¹³ Regardless of the level of fixed price, treating all transactions equally results in an adverse “lottery effect” that inefficiently allocates the scarce, valuable resource of a transaction space. As with both the zero-fee and inflation-only models, this effect at its most extreme can render the network unusable for many high-value users like merchants.</p><p>The third possible outcome is that the fixed fee happens to <em>exactly equal</em> the market clearing rate. In this case, the result is identical to the dynamic fee market. However, it is objectively worse since exact equality is a degenerate case that is unlikely to hold for long. Demand for transaction space naturally varies over time. One might expect, for example, higher demand during the normal 9 to 5 business hours than on a Sunday at 5am. As demand shifts, the dynamic fee that optimizes overall social welfare will shift, while the fixed fee, by definition, remains constant. At best, the “top down” upgrade system would dynamically shift the fixed fee to match the market clearing rate, but this of course just results in a more centralized version of dynamic fees.</p><p>There are a few points that supporters of fixed fees might raise to push back on this logic. One potential benefit is that everyone pays the same rate, which may be more equitable provided that the fixed fee is close to the dynamic fee. However, this does not improve overall social welfare — it merely shifts surplus from producers to consumers. Since users will not pay higher fees than the utility they derive from their transactions, variations in transaction fees do not reflect any true economic inequality in the system. Confusion on this point often arises over conflating transactions with users. A quick counterexample is a poor user who wishes to pay many bills on the network versus a rich user who only needs to send a single network transaction. In this case, the poor user ends up unfairly paying more in the fixed fee market than the rich user.</p><p>Incidentally, there are some cognitive costs of forcing everyone to choose their own fee. It is simple, however, to instead have everyone choose their maximum fee, and then set the fee for everyone to the minimum fee paid in the batch of transactions, which is the market-clearing fee. This largely eliminates the game played by users, and thus the cognitive costs, when selecting fees.¹⁴</p><p>Second, fixed fees could help avoid short-term spikes in fees that have been observed in both Bitcoin and Ethereum. While this is true by definition, this is actually not a desirable outcome. The spikes in fees represent genuine spikes in demand for high-value transactions. Why should a substantially lower value transaction be artificially advantaged (relative to the dynamic fee outcome)? Forcing the fees to be consistent and smooth does not eliminate these swings in demand — it merely makes it so transaction confirmation behavior is a lot less certain and well behaved. Opaqueness is a far cry from equity.</p><p>The third and most plausible, but still ultimately unconvincing, argument for fixed fees is it simplifies the user experience and eliminates the possibility of <a href="https://news.bitcoin.com/bitcoin-miner-repays-customer-who-accidentally-paid-2-5-bitcoins-transaction-fee/">“fat finger” fees</a>. This phenomenon appears to be relatively rare even in the remarkably user-unfriendly world of crypto and is much better solved at the user application level rather than at the protocol level. Simple checks like limits on fees, suggested fees, and additional abstraction of the user away from the underlying protocol can eliminate the vast majority of these occurrences. Levying the requirement on the base protocol to make up for poorly designed user experiences at the expense of network efficiency is a poor development approach.</p><h4>Dynamic</h4><p>The best basic fee model is dynamic fees. As stated in the model definition, allowing users to set their own fee and letting validators select the transactions with the highest fees for processing results in an efficient market for fees. This market incurs no deadweight loss and maximizes overall social utility and network efficiency.</p><p>The dynamic fee market automatically adjusts to changes in capacity and demand and efficiently allocates space to the highest utility transactions. This is a far better outcome than the effective lottery imposed by the other models, where many low-value transactions will take space from higher-value transactions.</p><p>The merits of this model have been touched on in the discussion of the other fee models, but there are a few additional features that are worth mentioning.</p><p>Dynamic user defined fees allow users to effectively place “limit orders,” where they can submit a low value transaction with a commensurately low fee. If the current market clearing fee exceeds this set fee, the transaction can wait until a period of less demand, when it will automatically be processed.</p><p>Most of the benefits of dynamic fees are observable at capacity. Before capacity is reached, the equilibrium fee level should be less than or equal to the fixed fees or implied inflationary fees (assuming that the latter two fees exceed the cost of validation; otherwise the network is not sustainable). This means that the user of the dynamic fee network is never worse off than users of other networks in the long run.</p><p>To the extent that the network capacity can vary with validator hardware, the rise in fees when capacity is reached will encourage existing validators to upgrade their hardware and better validators to join the network. The result is a higher performance network that benefits all users.</p><p>A common critique of non-zero direct fee models is that they complicate token transactions. For example, a stablecoin token user on Ethereum must buy Ether in addition to the stablecoin to pay the gas fees. Once again, this is a user experience rather than protocol issue, and making a quick fix at the protocol layer that erodes its efficiency and overall usefulness is the wrong approach. As user experience continues to improve, many of these issues can be easily abstracted away. Few users have difficulties navigating the use of credits in mobile games or online services, and network tokens held to pay fees can be viewed much the same way.</p><p>Ultimately, increasing capacity, rather than fiddling with fee models, is the best approach to making fees equitable and accessible. <strong>Of the four basic fee models, only dynamic fees do not have any serious long-term drawbacks.</strong></p><h4>When is inflation good?</h4><p>While the dynamic fees model is the clear winner of the basic fee models, it is possible that some combination of the fee models results in a better overall outcome. As we saw in the discussion of the various models, the effects of each can be easily incorporated into the microeconomic analysis. We can boil down each of the alternative models into some form of price control, subsidy, or tax, all of which result in deadweight loss in the self-contained fee market. This would seem to suggest that there is no place for inflation in an economically rational design.</p><p>However, there are some additional macro considerations that impact this analysis and suggest that <strong>there is a place for inflation in conjunction with a direct fee</strong>.</p><p>Inflation can help bootstrap an immature network to broader adoption in several ways. First, it subsidizes early adopters at the expense of existing token holders. This is equivalent to the founders and early investors in the network providing an incentive to early adopters, which is a very common business strategy.</p><p>Second, inflation that exceeds the cost of validation can enable early users to transact with zero fees for themselves, which can streamline the user experience. While it is a poor long-term strategy to bend the protocol to deal with UX issues, the reality is that the state of crypto infrastructure remains quite poor and justifies accommodating initial users to get the network off the ground.</p><p>Third, inflation can help distribute tokens to early adopters who are meaningfully contributing to the network. By providing a compelling value proposition at an early stage, the network can attract strong validators that increase network decentralization and robustness. Note that all of these benefits are<em> short-term</em>; at maturity, inflation for the sake of users is harder to justify.</p><p>Nevertheless, there are compelling reasons to keep some inflation in the network long-term: macroeconomic stability. The consensus in central bank monetary policy is that <a href="https://www.nber.org/papers/w16054.pdf">optimal inflation</a> in an economy is around 2% per annum. While antithetical to the early crypto ethos that is highly critical of inflationary policy, a low amount of inflation has clear benefits to an economic system by reducing frictions and promoting overall growth.</p><p>The dangers of the alternative — <a href="https://en.wikipedia.org/wiki/Deflation#Effects">deflation</a> due to a growing economy with constant monetary supply, like Bitcoin — are incontrovertible. By promoting hoarding of money, deflation introduces frictions into the economy, which can result in a severely adverse positive feedback loop called a <a href="https://en.wikipedia.org/wiki/Deflation#Deflationary_spiral">deflationary spiral</a>.</p><p>Before the introduction of modern central bank policy, deflation was at the center of many of the most severe economic crashes. By maintaining some modest inflation, a crypto network can grease the wheels of its economy and mitigate several existential risks.</p><p>Finally, there is a plausible reason to broadly tax the token holders on a network via inflation. All users of the network derive a genuine benefit from the network’s existence, whether they are currently using it or not. The very option of being able to use an efficient, cheap, secure, and decentralized network going forward is a positive externality — for example, as a hedge against potential censorship or infrastructure to build a future business. While it is imperfect in its impact, an inflation tax does accurately represent that not only the direct users of the network benefit from its existence.</p><h4>The Logos Approach</h4><p>So what, then, is the optimal fee policy?</p><p>At Logos, we have concluded that the best approach is dynamic, user-paid fees plus modest inflation. This combines the lessons of both the microeconomic and macroeconomic models to promote sustainable growth across Logos’ life cycle.</p><p>By permitting users to choose their own fees and validators to prioritize transactions accordingly, we can maximize overall network effectiveness and usefulness. The dynamic fee model yields the best microeconomic outcome by responding instantly to changes in demand and encouraging sustainable network growth.</p><p>The inflation component helps jumpstart the initial growth of the network in the short term. In the long term, a reasonable level of inflation keeps the network economy running smoothly and avoids major macroeconomic risks. By design, Logos concentrates the most expensive validation tasks in the hands of a relatively small number of delegates (while ensuring that these delegates are strictly accountable to the rest of the network, and, thus, that the network has a high level of security). As a result, a small amount of inflation can be a very large reward for the delegates, ensuring healthy competition that maximizes network performance. Such a level of inflation imparts very little deadweight loss on the overall network (provided that there is a dynamic fee to eliminate the “lottery effect”), making this a good compromise.</p><p>Logos’ biggest fee design is indirect: maximizing capacity. By providing optimal infrastructure for payments that pushes capacity to the limits of hardware and beyond, Logos ensures that direct fees will remain close to the marginal cost of validation, even at global levels of adoption.</p><h4>Conclusion</h4><p>Ultimately, there is no single right economic fee model, and even top economists will disagree over what is the proper level of inflation. Economic models are inherently simplifying and can bely much of the complexity present in a market. While simple in most respects, crypto has several unique properties that confuse this analysis in the short term, particularly the emotionally charged ideologies that are inseparable from the users of many networks.</p><p>Nevertheless, this does not invalidate the model conclusions. By applying simple principles that are logically true at a mature, rational economic equilibrium, we are able to see that the biggest buzz fee models — zero fees, fixed fees, and inflation only — have severe limitations that will significantly impair networks that use them. While alternative models may promote short-term growth, it is critical for projects and communities to be realistic about what will work in the long term if they wish to succeed.</p><p>Thanks to <a href="https://medium.com/logos-network/team-member-spotlight-david-smalling-lead-strategic-advisor-805f1be84697">David Smalling</a> for providing feedback on this article.</p><p>[1] Note that this isn’t strictly consistent with the empirical evidence from Bitcoin, but uneconomic choice of transactions is declining as miners improve and legacy selection rules that emphasized other traits are phased out. See Section 3.3 of <a href="https://arxiv.org/pdf/1709.02489.pdf">this paper</a> for more info.</p><p>[2] This simplicity contrasts with the relative complexity of the <a href="https://www.minneapolisfed.org/publications/the-region/the-interchange-fee-debate-issues-and-economics">payment interchange fee market</a>, where multiple intermediaries must be induced to participate in the network.</p><p>[3] We are ignoring big changes in hash power in a PoW network for simplicity.</p><p>[4] A related phenomenon is when people accept a lower salary to work in a field that they are passionate about or strongly believe in.</p><p>[5] In reality, each participant will have a different “enthusiasm” utility, but we can assume a constant shift down in the supply curve for simplicity.</p><p>[6] An interesting exception to this rule is sharding, but <a href="https://medium.com/logos-network/sharding-how-many-shards-are-safe-bc361c487083">sharding itself is limited</a>, and the prevailing wisdom that sharding can scale a network without limit is provably false. In general, sharding just pushes out this scaling problem to each shard.</p><p>[7] The fixed capacity should not be confused with the ability of the development team to increase that capacity. The fixed capacity results in a vertical supply curve, and an increase in that capacity would shift that curve right. This is very different than an upward sloping supply curve, as the former requires an exogenous network upgrade, while the latter occurs dynamically at a given network configuration.</p><p>[8] Spoiler alert — the dynamic fee model is the economically best fee model.</p><p>[9] Without getting too technical, IOTA lacks any Byzantine fault tolerance and is susceptible to double spend attacks.</p><p>[10] This is not true in proof-of-work networks where simple payment verification is a non-trivial problem, but more advanced consensus models make SPV simple and secure.</p><p>[11] for a good discussion of this issue, see the <a href="http://econ101help.com/consumer-surplus-producer-surplus-and-dead-weight-loss-with-inelastic-supply-curve/">comments of this article</a>.</p><p>[12] There is no endogenous increase in total network value with inflation. The previous holders now own 90% of the total token supply, while the validators now own 10%.</p><p>[13] This logic also holds for an inflation only model that does not sufficiently reward validators.</p><p>[14] See <a href="https://poseidon01.ssrn.com/delivery.php?ID=318115065115117123005078068003017087017069064045069066075089120069073004022122118018022103030104056120001028117082089002118095117022094008013000021018101117104070058087029001003075008031001115010088086092108065123077005112119113101031012106024095&amp;EXT=pdf">Basu et al.</a> for an analysis of this simple fix.</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=948d1d93a4f5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/crypto-transaction-fee-economics-primer-948d1d93a4f5">Crypto Transaction Fee Economics Primer</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Delegated Does Not Mean Centralized or Less Secure]]></title>
            <link>https://medium.com/logos-network/delegated-does-not-mean-centralized-or-less-secure-434ae4660f2?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/434ae4660f2</guid>
            <category><![CDATA[crypto]]></category>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[payments]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Fri, 04 Jan 2019 09:12:16 GMT</pubDate>
            <atom:updated>2019-02-08T01:21:06.132Z</atom:updated>
            <content:encoded><![CDATA[<p>An oft-touted rule of thumb is that delegated consensus mechanisms trade decentralization for speed. While seemingly plausible, it is actually false. In reality, delegation and decentralization are two orthogonal traits: a non-delegated network can be centralized, while a heavily delegated network can be highly decentralized.</p><p>In this article, we will clarify this misconception and explore what it really means to be decentralized.</p><h3>Defining Decentralization</h3><p>The biggest hangup most people have is understanding what decentralization really means. The “delegated = centralized” argument effectively equates the degree of decentralization to the <em>number</em> of validators or the hurdle to becoming a validator. But this is the shallowest way of looking at things.</p><p>As an example, take two networks: network A uses delegated proof-of-stake with 10 validators, while network B uses vanilla proof-of-stake with 1,000 validators. At first blush, network B appears to be more decentralized. But if just 4 validators control a super-majority of voting power in network B, then the remaining 996 validators are not contributing meaningfully to B’s decentralization. If voting power is evenly distributed amongst the validators in A, then it is substantially more decentralized than B.</p><p>Alternatively, let’s assume that network A has a strict rewards system that incentivizes validators to behave correctly, while network B lacks any such structure. Then network B is much more susceptible to the influence of a single validator or group of validators working together!</p><p>These examples should make it clear that the number of validators has very little to do with decentralization in reality, despite prevailing wisdom that suggests the contrary.</p><p>In an article on the <a href="http://bit.ly/2s8AiOx">Scalability Trilemma</a>, I defined decentralization much more generally:</p><p><em>“Any network that sufficiently guarantees that no single user or cartel of users can control the network can be said to be decentralized, while an architecture that fails to provide this distribution of power is centralized.”</em></p><p>I believe this gets at the heart of decentralization. As much as we would like to have an easy litmus test that we can use to say a particular network is decentralized, it’s not that simple! It doesn’t boil down to a single pithy statement like the <a href="http://bit.ly/2s8AiOx">Scalability Trilemma</a>, or the ability to run a full node on your laptop, or the number of validators. Rather, decentralization is a complex continuum that is influenced by many factors, and whether consensus involves delegation has little to no influence on this determination.</p><h3>Delegation is not the same as DPoS</h3><p>It is important to make a distinction — when I talk about delegated consensus, I am NOT referring to <a href="http://bytemaster.github.io/bitshares/2015/01/04/Delegated-Proof-of-Stake-vs-Proof-of-Work/">DPoS</a>, the specific consensus used by EOS, BitShares, and Steem. While DPoS is a type of delegated consensus, it is not the same as all delegated consensus algorithms. <a href="https://www.logos.network/whitepaper">Logos</a>, for instance, uses a very different delegated consensus model. Unfortunately, it seems that much of the confusion around delegation and decentralization stems from the misconception that DPoS is equivalent to delegated consensus.</p><p>In many ways, DPoS networks are more centralized than alternatives. They are not Byzantine fault tolerant (and make no effort to <em>mathematically</em> prove any fault tolerance), which means that a small number of validators could compromise the security of the system. The networks, as implemented, also lack any robust incentive structure to ensure validator compliance beyond the very weak threat of potentially losing a future election. While this property is independent of consensus, it makes the network very susceptible to attack.</p><p>However, DPoS networks are hardly representative of all delegated consensus networks, and delegated networks, if designed correctly, can actually be <em>more</em> decentralized than alternatives.</p><h3>Delegation Can Increase Decentralization</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*sL5lfi7FxA16Pksk" /><figcaption><em>Share of total validator power of top validators for </em><a href="https://blockchain.info/pools"><em>Bitcoin</em></a><em>, </em><a href="https://www.etherchain.org/charts/topMiners"><em>Ethereum</em></a><em>, and Logos.</em></figcaption></figure><p>All else equal — meaning that incentive structures, Byzantine fault tolerance, and all other non-delegation related factors are held constant — a delegated consensus model is likely more decentralized than vanilla proof-of-stake or proof-of-work.</p><p>How can that be? While, in theory, an open validator set would seemingly increase decentralization, if a small number of validators control most of the voting power, as we saw in the first example above, then the network will be effectively at the mercy of those few validators.</p><p>For a number of reasons, <em>vanilla proof-of-stake or proof-of-work networks would concentrate power in fewer hands in expectation</em>. Any secure, open, trustless network must have some mechanism of assigning votes that reflects an economic investment, such as computers in proof-of-work or tokens in proof-of-stake. Regardless of mechanism, there is inevitably an economies of scale, where larger validators are disproportionately advantaged compared to smaller validators. This is particularly exacerbated in proof-of-work, where large ASIC mining operations inevitably dominate the retail miners. However, it is also true in proof-of-stake, as there are unavoidable fixed costs in validation, such as setting up and maintaining a secure and performant node. Even in the absence of any economies of scale, we’d expect some natural <a href="https://en.wikipedia.org/wiki/Pareto_distribution#Relation_to_the_%22Pareto_principle%22">Pareto distribution</a> of economic power at equilibrium that results in a few major validators and many smaller validators. Combined, these two effects mean that any network that is profitable to validate will tend towards consolidation of power in a small number of hands, and vanilla PoS and PoW both lack any explicit mitigation.</p><p>This outcome is readily observable empirically. The chart above shows the actual distribution of validator power for Bitcoin and Ethereum compared to the distribution for Logos’ delegated model with 50 delegates. The top three validators for both Bitcoin and Ethereum control a critical majority of mining power (over 50%), while 33 validators would need to work together to control Logos’ consensus provided sufficient connectivity.</p><p>Admittedly, we are abstracting some details here. A PoS proponent may argue that features like staking pools could decrease the impact of economies of scale. Even assuming optimal design, <em>on an apples-to-apples basis, a delegated network should not be any less centralized than a vanilla PoS</em>.</p><p>Under a fair, well-designed, Byzantine fault tolerant election system, a node with 10% of the voting power should be able to elect 10% of delegates, and thus have 10% of the total representation in consensus. By comparison, a 10% validator in a good PoS system should also control 10% of consensus. Any other outcome would be poor design!</p><p>Of course, designing such an election system is non-trivial (so too is designing a good vanilla PoS), and many prominent networks like EOS fall short. But provided that care is taken, there is no technological, economic, or practical reason why delegation is necessarily less decentralized than alternatives.</p><h3>What Actually Matters — Incentive Alignment</h3><p>We’ve established that delegation, in practice, has little to do with decentralization. While there are a number of factors, including several foundational like Byzantine fault tolerance, what matters perhaps more than any other is the network’s incentive system.</p><p>What many people latch on to when they equate delegation with centralization is the potential for a single validator to abuse their position and act against the interests of the users that elected it. However, this is not much different than a huge mining pool acting maliciously. In fact, regular elections and greater distribution of power likely make the former less risky than the latter!</p><p>But this isn’t really getting at the heart of the issue, which is why a validator would act maliciously in the first place. This is fundamentally a game theory question.</p><p><strong>In order for a network to be robustly decentralized, the validators of the network — whether delegates, mining pools, or staking pools — must be held strictly accountable to the rest of the network via a game theoretically secure incentive system.</strong></p><p>In Logos, for example, there are a number of mechanisms to keep delegated validators honest. These include regular elections and loss of future income (rather weak), a wholesale recall that replaces the entire validator set (somewhat stronger), and slashing conditions (strongest).</p><p>Without delving too deeply into the game theory behind these mechanisms, the key result is that <em>a validator cannot attack the network without directly forfeiting a large amount of money</em>. This result can be extended to directly calculate the total dollar cost to pull off any attack on the network, which means that <strong>the network has a strong, quantitative, economic security</strong>. Note that this high-level feature is completely orthogonal to consensus and delegation.</p><p><strong>This means that a delegated system like Logos has a significantly higher security level (and a correspondingly high level of decentralization) than all proof-of-work networks and any proof-of-stake network without slashing.</strong></p><p>In a proof-of-work network like Bitcoin, for example, there are numerous <a href="https://bitcoinmagazine.com/articles/selfish-mining-a-25-attack-against-the-bitcoin-network-1383578440/">attack</a> <a href="http://randomwalker.info/publications/mining_CCS.pdf">vectors</a> beyond the more famous 51% attack that result from a poor incentive structure. The 51% attack itself is <a href="https://medium.com/@HusamABBOUD/the-realistic-lucrative-case-of-ethereum-classic-attack-with-1mm-today-8fa0430a7c25">surprisingly cheap to pull off</a> (by renting rather than buying hash power) and has occurred on <a href="https://www.ccn.com/privacy-coin-verge-succumbs-to-51-attack-again/">several</a> <a href="https://www.ccn.com/bitcoin-gold-hit-by-double-spend-attack-exchanges-lose-millions/">occasions</a>. This is possible because PoW incentives are grossly misaligned — the value of hardware has very little correlation with the value of the underlying network tokens, and there is no ability for the network to slash attackers’ mining power.</p><p>Bitcoin is surprisingly centralized for a network that is often touted as the gold standard in decentralization. Conversely, delegated consensus networks, while often maligned and mischaracterized, can have drastically higher security and decentralization at the equivalent network value. However, this has much more to do with the network’s incentive system than with the consensus itself. So while reductionist heuristics like “Delegated consensus is more centralized” are enticing, they are usually far from the truth.</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=434ae4660f2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/delegated-does-not-mean-centralized-or-less-secure-434ae4660f2">Delegated Does Not Mean Centralized or Less Secure</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Cryptoasset Valuation #3: The Winner-Take-All Myth and Skinnyfat Protocols]]></title>
            <link>https://medium.com/logos-network/cryptoasset-valuation-3-the-winner-take-all-myth-and-skinnyfat-protocols-47e9a6f41f04?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/47e9a6f41f04</guid>
            <category><![CDATA[investing]]></category>
            <category><![CDATA[valuation]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[cryptocurrency]]></category>
            <category><![CDATA[crypto]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Wed, 19 Dec 2018 15:15:08 GMT</pubDate>
            <atom:updated>2019-02-08T01:21:55.548Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Which side are you on?</em></p><p><em>This is the third of a series of articles exploring valuation of cryptoassets. This article assumes an understanding of the Equation of Exchange valuation model, which was explained in the </em><a href="https://medium.com/logos-network/cryptoasset-valuation-1-improving-the-equation-of-exchange-model-6525aa748d28"><em>first article</em></a><em>.</em></p><h3>Thin vs Fat Protocols</h3><p>The past two years have seen an unprecedented explosion of new decentralized protocols, spurred by numerous technical advances, practical improvements, and increased investor interest.</p><p>The vast majority of the genuinely innovative protocols are what I call <strong>general-purpose smart contract platforms (GSCPs)</strong>. These projects effectively aim to be a better, more scalable version of Ethereum, with full virtual machines and smart contract support. Their declared goals are to <em>build the optimal platform for most, if not all, decentralized applications</em>, from payments to computation to gaming to decentralized legal contracts. GSCPs include Dfinity, Algorand, EOS, Tezos, NEO, Hashgraph, IOTA, Ethereum, and many more.</p><p>This winner-take-all approach is commonly known as the <strong>Fat Protocol thesis</strong>, which was coined by Joel Monegro in a namesake <a href="http://www.usv.com/blog/fat-protocols">USV blog post</a>. The original post mostly argued that protocols will accrue more value than their applications,¹ but it has since matured into the belief that the vast majority of value and usage will accrue to a <em>single</em> platform. Proponents argue that protocol’s network effect will overwhelm any limitations or deficiencies in specific use cases compared to its competitors.</p><p>A small minority of new protocols, including Logos, have much more specialized designs tailored to a specific use case. These protocols aim to be the optimal solution for that single area of application, but they do not have broader aspirations.</p><p>These projects align with the Thin Protocol thesis, which was most notably proposed by Teemu Paivinen in a <a href="https://blog.zeppelin.solutions/thin-protocols-cc872258379f">Zeppelin blog post</a> as a direct response to Fat Protocols. The specialized protocols expect that their advantages in their target use case will allow them to dominate their niche.</p><p>Teemu takes it one step further, arguing that the ability to cheaply fork networks and the rise of interoperability technologies will result in hyperspecialization where no single protocol accrues significant value.²</p><p>Clearly, these two approaches are incompatible. <em>So far, prevailing wisdom has largely sided with the Fat Protocol thesis, resulting in the GSCPs grabbing most of the public’s mindshare. Empirical evidence, however, has so far sided with the Thin Protocol camp, with a huge proliferation of derivative networks and forks with only incremental improvements.</em></p><p>So which one is correct? This question has massive implications for cryptoasset valuation and investment — should you bet the farm on a single general platform winning out, or do you invest in many niche, specialized networks with compelling use cases?</p><p>For a number of reasons, <em>I believe that</em> <em>neither Fat nor Thin Protocols are entirely correct</em>. There are real benefits of specialization that cannot be ignored, as well as numerous economic and technological reasons why no network will be the single dominant, fat protocol. Nevertheless, network effects are real and networks cannot always be arbitrarily forked, which calls into question the assumptions of Thin Protocols.</p><p><strong>Instead, the most likely outcome balancing these effects is something in between, where there are many specialized DLT verticals, each of which is dominated by no more than a few networks.</strong> Sticking with the established jargon and, for lack of a better term, we can call these networks <em>Skinnyfat Protocols</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Z_gRpMmhvoIgkUU7" /><figcaption><em>Skinnyfat Protocols: not quite fat, not quite thin</em></figcaption></figure><h3>Specialized Dominates General Purpose</h3><p>Of the two leading theories, Fat Protocols is far less likely than Thin Protocols. The argument was well articulated in the original Thin Protocols article, so I will focus on some more concrete details. In short,<strong> specialized protocols can massively outperform general protocols in their target use case, and increasingly mature interoperability technologies erode any universal network effect.</strong></p><h3>No Free Lunch</h3><p>By definition, a specialized protocol is tailor-made for a specific use case, while a general purpose network necessarily takes a jack-of-all-trades approach. The question is whether it is possible for a general purpose network to be equally well suited, or at least competitive, with a specialized network in a specific use case. Most GSCP proponents, for example, operate under the assumption that their smart contract platform can be both the best payments network and the best decentralized computation network.</p><p>It has become increasingly clear that the answer to this question is emphatically, “<em>No</em>.” This is easily seen empirically in the context of payments, once you cut through the highly-contrived marketing benchmarks. Two of the most advanced smart contract platforms, <a href="https://people.csail.mit.edu/nickolai/papers/gilad-algorand-eprint.pdf">Algorand</a>³ and <a href="https://dfinity.org/tech">Dfinity</a>⁴, claim less than 1,000 TPS capacity, which is far below what centralized payments networks today can handle. By comparison, <a href="https://www.logos.network/">Logos</a>, the leading payments focused network, is the exception and processes over<a href="https://medium.com/logos-network/alpha-implementation-benchmarking-and-demo-d1c4d9dbba8f"> 17,000 transactions per second (TPS) on its initial test net</a> under realistic conditions and will have capacity for more than 150,000 TPS when it is fully featured. A similar performance gap between Logos on the one hand and Algorand and Dfinity on the other is observable in confirmation latency, and this gap is consistent across advanced general protocols.</p><p>The fact that general protocols are not competitive with specialized protocols makes sense from first principles. At a high level, this can be deduced from a No Free Lunch principle — you can’t get something for nothing. After a few strict improvements (e.g. moving from proof-of-work to proof-of-stake), you very quickly get to a point where there are fundamental tradeoffs in DLT architecture that have a real impact.</p><p>For example, a general purpose network must include smart contracts — otherwise, it is not very general at all. Smart contracts impose <em>significant</em> overhead. There is no getting around this — it simply takes a lot more computational time and effort to validate a smart contract operation than a simple send.</p><p>As we have detailed in a <a href="https://medium.com/logos-network/second-layers-are-not-the-scalability-silver-bullet-a9d262bc816c">previous article</a>, off-chain solutions are not a silver bullet or even a good solution for many high performance use cases like payments, and a full, Turing-complete smart contract system will always be at a huge disadvantage to a specialized, limited network.</p><p><strong>This performance gap is exacerbated by the fact that general networks need to optimize over all use cases, while a specialized network is able to tailor its design to a specific use case. </strong>This specialization is often suboptimal for other use cases but can result in a significant boost in performance for the use case the network cares about.</p><p><strong>The performance gap has a real impact not only on the convenience and efficiency of use of the protocol but also the cost.</strong> If designed economically optimally, then capacity is a free market, and lower capacity directly translates to higher fees. Even if there are “no fees,” there is some economic deadweight loss — someone who valued their transaction did not get it through. Moreover, execution time has a real cost in itself. These costs mean that non-specialized networks are at a significant competitive disadvantage.⁵</p><p>Unfortunately for general purpose networks, a jack of all blockchain trades is necessarily a master of none.</p><h3>Interoperability Means Competitive Pressures</h3><p>The general purpose performance gap would not be the Achilles’ heel for the Fat Protocol thesis if blockchain were a normal market. In the payments space, for instance, many systems are outdated and inefficient but persist due to the oligopolistic market conditions. A new centralized payment rail would need to jump through countless regulatory hoops and technical integrations before it is useful, a process involving many years and likely billions of dollars of expenses. These barriers to entry protect incumbents, insulating them from competitive pressures and stifling competition, thereby allowing them to accrue huge value.</p><p>The world of blockchain is vastly different. In particular, there are a growing number of increasingly mature interoperability technologies that ratchet up the competitiveness of potential markets.</p><p>Centralized exchanges like Coinbase or Binance are the primary means of interoperability currently, and they are increasingly accessible, usable, and compliant from a regulatory perspective. Decentralized exchanges like 0x are further democratizing and automating this exchange market. Atomic swaps, interchain transfers, sidechains, and several other technologies look increasingly promising at further decentralizing efficient exchange.</p><p>As it becomes easier to swap to and from different networks, the competitive advantages of one protocol over another will be increasingly realizable. Since the difference between specialized and general protocols is likely to be very large, both in terms of performance and cost, users will have a commensurately large incentive to use a network tailored to the use case of interest rather, than an inferior general network.</p><p>Interoperability between networks is only one side of the coin, the other side being software interoperability. Here, too, the blockchain space is highly conducive to competition. Once one cryptocurrency is integrated in a particular project, it is very easy to integrate additional cryptocurrencies. The open source nature of both protocols and decentralized infrastructure makes it virtually impossible for it to be any other way. As a result, if a merchant and consumer can save money by switching from a general purpose network to a far more efficient payment network, it will be fairly easy for them to do so.</p><p><strong>Given the magnitude of differences between general purpose and specialized networks, as well as unprecedented interoperability, a single fat protocol is highly unlikely.</strong></p><h3>There is a Network Effect — But It is Limited</h3><p>So far, our arguments have closely followed the Thin Protocols canon. However, the Fat Protocol view does have some merit. Specifically, there certainly is a network effect that has an undeniable impact on the value of a network to prospective users.</p><p>Take payments, for example. A payment rail is definitely more useful if more merchants and consumers accept it. There is a reason why people favor Visa over Diners Club.</p><p>Moreover, the Thin Protocol argument has several faulty assumptions. First, forking a network does not result in a perfect economic substitute. There are real differences even if the codebase is entirely the same, particularly vis-à-vis network security, reliability, decentralization, and usefulness.</p><p>Bitcoin Gold is a good example — it forked Bitcoin without any meaningful changes for the end user,⁶ and it is a strictly inferior network from a usability standpoint.</p><p>Second, no swap or transaction is completely frictionless, and there is only so much you can automate. Exchanging between two networks <em>always</em> comes at some cost. At a minimum, there is inevitably bid-ask spread plus any slippage, which reflects the real risk associated with market making. There are additionally cognitive and implementation costs that may be even larger than the direct dollar costs. This can lead to some counterintuitive outcomes. For example, if you can save $0.01 by transacting in networks A and B for two very related use cases compared to just using network C, but if you need to spend $0.02 to swap between them, then network C will dominate.</p><p>Third, there are components of networks that are not forkable or replicable at all. For example, network funding can have a big impact on the prospects of the network. Similarly, many services built on top of a network are not open source and may be resistant to supporting a fork. More broadly, this concept is reflected in the somewhat cliche adage, “The best tech doesn’t always win.”</p><p>Together, these observations mean that interoperability can do so much to attenuate network effects. <strong>Unless there is a real, substantive value difference for using one network over the other that is <em>greater than the cost of swapping</em>, users will not have an incentive to move between networks</strong>. In general, this incentive will be present and impossible to ignore for a specialized network taking on a general network, but it will likely not exist for a specialized network that is only marginally better than another specialized network.</p><p>In other words, <em>networks will become quite sticky once the marginal benefit of specialization no longer outweighs the marginal cost of additional swapping</em>. This convergence will likely be quite rapid — once you get close to the limits of hardware for a particular use case, there isn’t much that can be improved.</p><p>Of course, I don’t think Thin Protocol proponents would say that networks are infinitely forkable and every minute advantage for each tiny niche use case will be eked out. But I think they greatly overestimate how much networks can specialize and underestimate the value that can accrue to a particular network.</p><h3>Skinnyfat Protocols and Blockchain Verticals</h3><p>The blockchain space at maturity, then, is unlikely to be dominated by a single fat protocol or spread thin across many small, niche protocols. Given the economic and technological phenomena described above, the much more likely outcome is the intermediate “Skinnyfat” Protocol, where there are one or two dominant but specialized protocols within each use case vertical.</p><p>The logic is pretty simple: <strong>a network specialized for a specific vertical will have a huge competitive advantage over general networks, but after a while, the incremental improvements will not be able to outweigh the real costs of using yet another network. </strong>Swapping between verticals will be relatively seamless and low frequency, such that the overall interoperability cost will be low.</p><p>In contrast to the Thin Protocols argument, these verticals will be relatively wide due to network stickiness. For example, a general decentralized data storage network will probably dominate a network focused on photograph storage and another focused on document storage, even if there is a slight advantage the hyper-specialized networks offer.</p><p>Exogenous, structural factors will define some verticals. In the data storage example, a network highly specialized around HIPAA compliance likely has room to exist due to regulatory rather than technological reasons.</p><p>Some verticals will be very broad as a result of a strictly dominant set of properties that a single network can offer, leaving no room for further specialization. In payments, for example, the quickest, cheapest, and most secure network would likely be equally optimal for B2C payments as P2P.</p><p>To the extent that a single vertical has multiple viable networks, we’d expect them to fall into some sort of <a href="https://en.wikipedia.org/wiki/Pareto_distribution">Pareto distribution</a> where the largest network has the bulk of market share, and each successive network only has a fraction of the previous network’s size. <strong>A likely cause of coexistence is geography: while DLT can erode borders, it probably will not erase them completely</strong>, at least not in any near future.</p><h3>Value Accrual in a Skinnyfat World</h3><p>Unlike Thin Protocols, the Skinnyfat outcome suggests that the dominant specialized protocols <em>will</em> accrue significant value that will not be forked away to oblivion. This partly reflects the entrenched network effects. However, as discussed in a <a href="https://medium.com/logos-network/cryptoasset-valuation-2-the-velocity-problem-8bbb4111c9c7">previous article</a>, accrued value does not necessarily correspond to economic rent. More importantly, it reflects the genuinely large market that a single dominant network within a vertical can capture.</p><p>While the value that a general purpose network might capture is probably significantly less than their supporters hope, this is not to say they will not capture any value. In fact, they all are specialized in their own way — for smart contracts! Decentralized computation itself is likely several verticals, such as GPU intensive computation vs CPU intensive computation, so I would not be surprised if several smart contract networks are very successful.</p><p>Right now, we are far away from any state close to maturity. No protocol has a network effect since nothing has any meaningful adoption. Even Ethereum has essentially no network effect since the Ethereum Virtual Machine (or WASM), Solidity, and their developer ecosystem can easily be ported to another protocol. As a result, the race is still wide open. From a valuation perspective, then, it makes sense to focus on the highest value verticals and the best protocols within those verticals.</p><p>We designed Logos from the ground up according to this thesis. Payments is the most compelling specialized vertical, and Logos is built to be the <a href="https://medium.com/logos-network/do-we-even-need-another-payments-solution-477137ea6a27">optimal payments network</a>.</p><p>[1] I will not elaborate on this point in this article, but I agree with Joel that the vast majority of value will accrue at the protocol rather than application layer in blockchain.</p><p>[2] From the post: “A lot of users/applications indicates a ‘fat’ protocol and likely means the protocol has a large market capitalisation, creating an incentive for competition either by forking or specialised network creation.”</p><p>[3] Algorand claims 125 times the capacity of Bitcoin, roughly 500 TPS.</p><p>[4] Dfinity claims they will run at roughly 50x gas limit of Ethereum and 2x the block time, roughly 800–1000 TPS.</p><p>[5] The oft cited (by Bitcoin maximalists) <a href="https://en.wikipedia.org/wiki/Focal_point_(game_theory)#Schelling&#39;s_example">Schelling effect</a> does not come into play here in real-world applications, where there are informed participants and repeated games. The <a href="https://en.wikipedia.org/wiki/Lindy_effect">Lindy Effect</a> may come into play in the form of network stickiness, which we discuss later.</p><p>[6] Bitcoin Gold’s only innovation — switching hashing algorithms — has virtually no impact from the perspective of an end user.</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=47e9a6f41f04" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/cryptoasset-valuation-3-the-winner-take-all-myth-and-skinnyfat-protocols-47e9a6f41f04">Cryptoasset Valuation #3: The Winner-Take-All Myth and Skinnyfat Protocols</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Introducing the Logos Test Net]]></title>
            <link>https://medium.com/logos-network/introducing-the-logos-wallet-test-net-2d2e9cfd72e2?source=rss-f9bef90711a5------2</link>
            <guid isPermaLink="false">https://medium.com/p/2d2e9cfd72e2</guid>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[techupdates]]></category>
            <category><![CDATA[bitcoin]]></category>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[announcements]]></category>
            <dc:creator><![CDATA[Michael Zochowski]]></dc:creator>
            <pubDate>Mon, 17 Dec 2018 23:49:29 GMT</pubDate>
            <atom:updated>2019-02-14T18:06:04.390Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Be the first to try it out — application instructions for first access below!</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*29KbVN0mM-Y2P2xZhfs4kQ.jpeg" /></figure><p>Here at Logos, we were in heads down mode for the past few months working to bring the payment platform of the future, today, into the hands of our very first users. Now, we are absolutely thrilled to announce a significant milestone and the product of our efforts — <strong>the launch of our permissioned test net last Friday</strong>.</p><p>We welcome you to try it out — please apply to use it <a href="https://docs.google.com/forms/d/e/1FAIpQLSfJlQx7ajWcyDGZMUwwlyvgS9waHjzGQhI0EhFnwSu70D_Uvw/viewform">here</a>.</p><p>Once you’ve been accepted, you will get the full user instructions on how to download and play around with the Logos iOS Wallet.</p><p>As we share this moment with you, on behalf of the whole Logos team, we thank you again for being an integral part of the growing Logos ecosystem as we continue to build out the premium hyper-scalable, low-cost and real-time payment platform.</p><p>After less than a year since Logos’ inception, we are very pleased with the progress the team is making and the momentum we have to meet our current and upcoming technological milestones, including the open source launch in the summer of 2019.</p><p>So, without further ado — download the app, grab some Logos coins, and check it out!</p><p>As always, we welcome and appreciate your feedback and thoughts — which you can send to <strong>testnet@logos.network</strong>.</p><p>Onward,</p><p><a href="https://medium.com/u/f9bef90711a5">Michael Zochowski</a>, Founder of Logos</p><p><a href="https://medium.com/u/300ab04b5655">Carl Hua</a>, Head of Engineering</p><p><strong>If you’d like to keep up with what we are doing:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/142/1*SLSkqLl-9Rb04qw_NMD53Q.png" /></figure><p><strong>Follow us</strong>: <a href="https://twitter.com/LogosPayments">Twitter</a> | <a href="https://discord.gg/BB9CQwn">Discord</a> | <a href="https://web.telegram.org/#/im?p=@logosofficial">Telegram</a> | <a href="https://www.reddit.com/r/LogosPayments/">Reddit</a></p><p><strong>Read</strong>: <a href="https://www.logos.network/whitepaper">White paper</a> | <a href="https://www.logos.network/">Logos website</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2d2e9cfd72e2" width="1" height="1" alt=""><hr><p><a href="https://medium.com/logos-network/introducing-the-logos-wallet-test-net-2d2e9cfd72e2">Introducing the Logos Test Net</a> was originally published in <a href="https://medium.com/logos-network">Logos Network</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>