<?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 Daniel Kronovet on Medium]]></title>
        <description><![CDATA[Stories by Daniel Kronovet on Medium]]></description>
        <link>https://medium.com/@kronosapiens?source=rss-bfe1f18e2a74------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Da3KgnILhILQbumzwojq5A.jpeg</url>
            <title>Stories by Daniel Kronovet on Medium</title>
            <link>https://medium.com/@kronosapiens?source=rss-bfe1f18e2a74------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Mon, 16 Nov 2026 22:51:12 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kronosapiens/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[Trie, Merkle, Patricia: A Blockchain Story]]></title>
            <link>https://medium.com/hackernoon/trie-merkle-patricia-a-blockchain-story-d8f20efc98d4?source=rss-bfe1f18e2a74------2</link>
            <guid isPermaLink="false">https://medium.com/p/d8f20efc98d4</guid>
            <category><![CDATA[humor]]></category>
            <category><![CDATA[introduction]]></category>
            <category><![CDATA[data-structures]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[blockchain-story]]></category>
            <dc:creator><![CDATA[Daniel Kronovet]]></dc:creator>
            <pubDate>Wed, 04 Jul 2018 00:00:00 GMT</pubDate>
            <atom:updated>2018-07-23T15:31:00.013Z</atom:updated>
            <content:encoded><![CDATA[<p>In which we tell the story of the Patricia Tree.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2XkmCRwKoxg9kh6PmN-Jjw.jpeg" /><figcaption>You’ll see.</figcaption></figure><h3>I. Introduction</h3><p>Spend a few days around blockchain engineers and certain words will start to sound familiar. “Merkle Tree” and “Patricia Tree” in particular will start to seem… important somehow. You’ll eventually gather that these are quite essential parts of this whole blockchain thing… but why? What problems, exactly, do they solve?</p><p>You might do a quick search and stumble upon more than a few peices of #content which explain these things, but retreat upon seeing the complicated-looking diagrams. Fear not, dear reader. Here we will explain these things, not with graphs, but with story.</p><p>Where to begin? The beginning, I suppose.</p><h3>II. The Hash Table</h3><p>In the beginning there was the <strong>computer,</strong> stretching infinitely in all directions. In fact, it’s hard to say that there even <em>was</em> the computer, since existence implies absence, and there was nothing that wasn’t the computer. So there was the computer, but the computer was inert. Nothing was <em>happening</em>. Boring. So the computer decided to create a <strong>programmer.</strong> Pop.</p><p>At first the programmer wasn’t very good, but over time she got better. There wasn’t much else going on at this time, so the programmer kept going, programming more and more things into the world. Animals and the like. After a while there were a lot of animals, which meant a lot of names to keep track of. This was a problem.</p><p>The programmer thought — “how can I keep track of all of the names of these animals? I want to be able to easily look up the name I gave to each species of animal. I could write all the names down in a big list, but eventually looking up the names will get really slow. If only I had the right <em>data structure</em>”.</p><p>And so the programmer created the <strong>hash table.</strong></p><p>What is a hash table? For starters, its the basis of everything else that’s going to happen, so we’re going to talk about it for a minute. Essentially, a hash table is a type of “key-value store”. This means that for a given “key” (i.e. an animal specie) you can save the “value” (i.e. the name of the animal). The main property of the hash table is that when you have a key, you can find the value <em>fast</em>, regardless of how many other items are in the hash table. In computer science terms, this is known as “constant-time lookup” and is very useful, which is why hashtables are <a href="http://steve-yegge.blogspot.com/2008/03/get-that-job-at-google.html">“arguably the single most important data structure known to mankind”</a>). Here’s an example:</p><pre>&gt;&gt;&gt; hashtable.set(&quot;dog&quot;, &quot;fido&quot;)<br>&gt;&gt;&gt; hashtable.get(&quot;dog&quot;)<br>&quot;fido&quot;</pre><p>How do they work? To understand the hash table, we have to digress for a moment and talk the <strong>hash function</strong>. Hash functions are a magical secret sauce which make some amazing things possible. Hash functions are the “cryptography” people talk about when they talk about blockchains. Hash functions are legit.</p><p>What is a hash function? Fortunately, hash functions are simple to understand. They are essentially tiny machines which take in some value, shake it around for a while (imagine a bartender shaking a cocktail), and output some other crazy-looking value (a big number). Their essential properties are:</p><ul><li>For a given input (like “cat”), you will always get the same output (like “0x52763589”)</li><li>Two similar inputs (like “cat” and “car”) should not have similar outputs. Put another way, given an output, you should not be able to guess the input.</li></ul><p>This makes hash functions extremely useful because they let us handle sensitive information safely. Have you ever wondered how <em>responsible</em> websites keep your passwords safe? They don’t store your password, they store a hash of your password. When you type in your password to log in, they take the hash of your password and compare that against what they have in their database. But if a hacker ever gets in, all they’ll know is the gibberish hash of your password — useless since they have no way of figuring out what your actual password was.</p><p>The other thing they’re useful for is making hash tables. Why? Remember that the output of the hash function is a <em>number</em>. So when you hash the key, you essentially get a number telling you where to find the value. Imagine the hash table as a cabinet with 100 drawers. You hash(&quot;dog&quot;) and get 34 – you go to cabinet 34 and get the name out. You hash(&quot;cat&quot;) and get 89 – you go to cabinet 89. No need to look through a whole list – you skip directly to the finish line.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*nHCDRkCPjzmzDZN_iukMKw.jpeg" /><figcaption>All hail Lord Kallax (src: Ikea)</figcaption></figure><p>Pretty cool right? Yes it is.</p><p>And so the programmer had the hash table, and for a while things were good. Great, even! But it couldn’t last. Eventually, the <strong>brogrammer</strong> appeared.</p><p>At first things were good between them: they shared ideas, they shared code, they shared space. But eventually dark clouds appeared on the horizon. They wanted different things. The programmer was fine with a little randomness thrown into things, but the brogrammer wanted certainty, and he wasn’t happy with hash tables anymore. They’re “not deterministic”, he said.</p><p>What did he mean? To understand this point, we’ll have to talk a little more about hash functions and hash tables. The first thing to note is that the “range” of the hash function (the possible values the output can take) is very large — depending on the computer, it can take as much as 2²⁵⁶, but more typically 2³² or 2⁶⁴ possible values. 2³² is 4,294,967,296 — and the others are much, much larger. Hash tables have to support this whole range, but we can’t make cabinets with that many drawers — there wouldn’t be room for anything else! So behind the scenes, we do a little trick: we take the hash value <em>modulo</em> the size of the cabinet. The modulo operation (%) is essentially division’s sidekick: it gives you the <em>remainder</em>. The nice thing about modulo is that the output (the remainder) is always between 0 and the base – so no matter how big the input, the output can only be so big.</p><p>So behind the scenes, we make a cabinet with 100 drawers, and when deciding where to put the name of &quot;dog&quot;, we look in drawer hash(&quot;dog&quot;) % 100. Because the hash value is random, the remainder will still be random, just smaller. This works great, but there’s a big downside: two animals might end up in the same drawer! Let’s say that hash(&quot;dog&quot;) is 1,000,034 and hash(&quot;shark&quot;) is 200,034. Different values, but both will be 34 after the modulo. So we put them in the same drawer, and we have to look through the drawer to find the dog’s name. It’s still fast, since there’s usually only one or two names in the drawer.</p><p>So it’s fine in practice, but the brogrammer’s point is that the spot you put the name in is not 100% determined by the hash function you’re using. Two more factors come in: the size of the cabinet, and the other animals! The size of the cabinet matters because a cabinet with 10 drawers will put both 72 and 182 in the same place (2), but a cabinet with 100 drawers will put them in different places (72 and 82). Also, you can’t tell in advance if a name will be alone in a drawer, or if it will have to share with other names.</p><p>The brogrammer wasn’t happy about this, but dealt with his feelings in a healthy way and went off into the mountains for a few weeks to think about alternatives. “A place for everything, and everything in its place,” he kept repeating in his head. When he eventually came down, he had a new idea.</p><h3>III. The Trie</h3><p>The problem, the brogrammer had realized, was that we were trying to put everything into a single huge cabinet, which could never be big enough. The solution, the brogrammer said, was to use a sequence of <em>smaller</em> cabinets. The first cabinet would give you the address of the second cabinet, the second cabinet the third, and eventually you would get to the cabinet which had the name you were looking for. You would need more cabinets (but not that many, as it turns out), and each cabinet could be quite small (maybe 16 drawers, or even 2!). Here’s an example, using an 8-drawer, 3-cabinet system (which gives us 8³ = 512 drawers total):</p><pre>&gt;&gt;&gt; hash3(&quot;dog&quot;)<br>0x237</pre><pre>&gt;&gt;&gt; firstCabinet = trie.find(firstCabinetLocation)</pre><pre>&gt;&gt;&gt; secondCabinetLocation = firstCabinet.drawer(7).contents<br>&gt;&gt;&gt; secondCabinet = trie.find(secondCabinetLocation)</pre><pre>&gt;&gt;&gt; thirdCabinetLocation = secondCabinet.drawer(3).contents<br>&gt;&gt;&gt; thirdCabinet = trie.find(thirdCabinetLocation)</pre><pre>&gt;&gt;&gt; thirdCabinet.drawer(2).contents<br>&quot;fido&quot;</pre><p>Note that each number tells us which drawer to open, and each number means one more cabinet. The brogrammer called this system a “<a href="https://en.wikipedia.org/wiki/Trie">Trie</a>” (as in re<strong>trie</strong>ve), and said that the beauty of it was that you didn’t need to build all the cabinets at once — you could start out with just one cabinet, and only build new cabinets the first time you needed them, wherever there was room. And while it means a little more work (opening more drawers), every name will have a dedicated drawer, always in the same place. And the brogrammer knew that no one would ever need <em>all</em> the drawers, and so most of the cabinets would never need to be built (although you can’t rule it out).</p><p>The programmer looked at the Trie and agreed it was a clever idea (although it involved quite a bit more walking), and there was harmony between them.</p><p>Years passed, and a new <strong>people</strong> started to appear in the nearby valley. Curious, the programmer and the brogrammer journeyed over to see these people and learn about their culture. They found the people intriguing, with a curious religion revolving around the worship of a particular arrangement of carved granite blocks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2XkmCRwKoxg9kh6PmN-Jjw.jpeg" /><figcaption>A particular arrangement of blocks (src: public domain)</figcaption></figure><p>The people were quite friendly, and after meeting with some of their priests, the programmer and brogrammer learned that these people had once been warlike, but after years of conflict developed a new system of “trust” which allowed them to co-exist in remarkable peace and prosperity. The computer, they said, was only as good as the programmer, and that humans could not be trusted to program alone. These people knew of the hash table and the trie, but they had found that people would cheat: sometimes people would come in the night and change the names in the drawers; there was no way to prove that the names in the drawers were the right names. For a while these people had a warrior class who guarded the cabinets, but found that this only led to more conflict.</p><p>Eventually a number of their most skilled artisans developed the technique of carving blocks of granite; these blocks, they realized, were very difficult to carve, and so things carved into these blocks could be trusted in away that the names in the cabinets could not. It was unfeasible to carve every name into the block, however, and to carve new blocks when the names changed. What they needed, they said, was some way to carve a <em>signature</em> of the names onto the block, such that if any one name changes, the signature would change; but if the names were the same, the signature would always be the same. Eventually, one of their scientists, Ralph, developed a <a href="https://en.wikipedia.org/wiki/Merkle_tree">solution</a>: the Merkle tree.</p><h3>IV. The Merkle Tree</h3><p>The Merkle tree behaves much like a Trie, but with a new rule: the drawers of each cabinet will not contain the location of the next cabinet, but rather the <em>hash</em> of all of the contents of the next cabinet. Separately, we keep track of the location of each cabinet (using, of all things, a simple hash table):</p><pre>&gt;&gt;&gt; hash3(&quot;dog&quot;)<br>0x237</pre><pre>&gt;&gt;&gt; firstCabinetLocation = hashtable.get(firstCabinetHash)<br>&gt;&gt;&gt; firstCabinet = trie.find(firstCabinetLocation)</pre><pre>&gt;&gt;&gt; secondCabinetHash = firstCabinet.drawer(7).contents<br>&gt;&gt;&gt; secondCabinetLocation = hashtable.get(secondCabinetHash)<br>&gt;&gt;&gt; secondCabinet = trie.find(secondCabinetLocation)</pre><pre>&gt;&gt;&gt; thirdCabinetHash = secondCabinet.drawer(3).contents<br>&gt;&gt;&gt; thirdCabinetLocation = hashtable.get(thirdCabinetHash)<br>&gt;&gt;&gt; thirdCabinet = trie.find(thirdCabinetLocation)</pre><pre>&gt;&gt;&gt; thirdCabinet.drawer(2).contents<br>&quot;fido&quot;</pre><p>Remember our hash function? Earlier we talked about hashing simple values like “dog” and “cat”, but in truth you can hash anything, including other hashes or sets of hashes. What Ralph realized was that by keeping the hashes in the cabinets, you can create a “hash trail” which will change whenever any value changes (remember how websites store your passwords? Same idea). Here is how you update a value:</p><pre>&gt;&gt;&gt; hash3(&quot;dog&quot;)<br>0x237</pre><pre>### Find cabinet same as before</pre><pre>&gt;&gt;&gt; thirdCabinet.drawer(2).contents = &quot;rover&quot;</pre><pre>### But then you start working backwards...</pre><pre>&gt;&gt;&gt; thirdCabinetHash = hash3(thirdCabinet.drawers)<br>&gt;&gt;&gt; hashtable.set(thirdCabinetHash, thirdCabinetLocation)</pre><pre>&gt;&gt;&gt; secondCabinet.drawer(3).contents = thirdCabinetHash<br>&gt;&gt;&gt; secondCabinetHash = hash3(secondCabinet.drawers)<br>&gt;&gt;&gt; hashtable.set(secondCabinetHash, secondCabinetLocation)</pre><pre>&gt;&gt;&gt; firstCabinet.drawer(7).contents = secondCabinetHash<br>&gt;&gt;&gt; firstCabinetHash = hash3(firstCabinet.drawers)<br>&gt;&gt;&gt; hashtable.set(firstCabinetHash, firstCabinetLocation)</pre><pre>&gt;&gt;&gt; firstCabinetHash<br>0x375</pre><p>Now the final value, 0x375, is a “fingerprint” of the entire Merkle tree. You can save this fingerprint (or engrave it into a granite block), and know that if anyone changes any of the names in the drawers, the process of making the hashes will give a different result – you’ll know something has changed. Notice that this adds more steps compared to a simple Trie: you need to have a separate hashtable to keep track of locations. But what you get is <em>security</em>.</p><p>The programmer and the brogrammer walked up to get a closer look at the granite blocks, and to their surprise, on them they saw engraved a series of hashes! 0x736, 0x264, 0x123, and so on, with 0x542 being the most recent. They were amazed! Nearby, they noticed some activity: one of this peculiar tribe wanted to prove that he had purchased a horse from another. He brought forward the name of the horse and his own name, set trie.set(horse, name) and through an elaborate ritual showed that his name, hashed with certain other names, with certain other names… voila! He arrived at 0x542, and thus all agreed that the horse was his.</p><p>What a remarkable society, the programmer and brogrammer agreed. There was something nagging at the programmer, though. This was a small tribe — only 512 members. As they grow, they will need a new hash function with a larger range — thousands, millions, billions. And so updating and verifying the values in the Merkle tree will become more and more costly — from three cabinets to five, to ten, to sixty and beyond! And for what? Most of these drawers will be empty. It seems like an expensive system, slow and costly. Surely there must be a better way? If only there was a Practical Algorithm To Retrieve Information Coded In Alphanumeric…</p><h3>V. The Patricia Tree</h3><p>To gather their thoughts, the programmer and the brogrammer took a walk into the hills above the valley. “There must be some way to optimize this tree!” they thought to themselves. The brogrammer suggested they look at a few random hashes, to build some intuition:</p><pre>&gt;&gt;&gt; hash8(&quot;cat&quot;)<br>0x14350235</pre><pre>&gt;&gt;&gt; hash8(&quot;dog&quot;)<br>0x14350762</pre><p>Then the brogrammer got excited — he noticed that both of these hashes happened to start with the same numbers: 14350. With just these two entries, getting to the final drawer should only need two cabinets: one for 14350, and one for whatever was left: 235 or 762. This would be much faster than using eight cabinets. You could always add more cabinets later, but why make more than you need? On each drawer we tape a little slip of paper, where we write down the common prefix for that drawer. Finally, the first cabinet is actually just a single drawer.</p><p>Looking up values would go like this:</p><pre>&gt;&gt;&gt; hash8(&quot;dog&quot;)<br>0x14350762</pre><pre>&gt;&gt;&gt; firstDrawerLocation = hashtable.get(firstDrawerHash)<br>&gt;&gt;&gt; firstDrawer = trie.find(firstDrawerLocation)<br>&gt;&gt;&gt; split(14350762, firstDrawer.commonPrefix)<br>(14350, 762)</pre><pre>&gt;&gt;&gt; secondCabinetHash = firstDrawer.contents<br>&gt;&gt;&gt; secondCabinetLocation = hashtable.get(secondCabinetHash)<br>&gt;&gt;&gt; secondCabinet = trie.find(secondCabinetLocation)<br>&gt;&gt;&gt; secondDrawer = secondCabinet.drawer(7)<br>&gt;&gt;&gt; split(62, secondDrawer.commonPrefix)<br>(62,)</pre><pre>&gt;&gt;&gt; secondDrawer.contents &quot;fido&quot;</pre><p>The programmer got excited — she felt pretty good about this. It would make the algorithm a little trickier, to make sure that cabinets were created appropriately and that common prefixes are kept up-to-date, but nothing they couldn’t figure out. A little more work at the beginning to set this all up would save the valley people a lot of time over the long run.</p><p>The pair sat down and worked out the details of this new system, which they called the <a href="https://github.com/ethereum/wiki/wiki/Patricia-Tree">“Patricia Tree”</a>. Satisfied, they descended to the valley and presented their work to the people there. They people were joyous; the slow Merkle tree had been a drag on their society. With the Patricia tree, they hoped, they would be able to advance their arts, sciences, and industry faster.</p><p>Satisfied, the programmer and the brogrammer left the valley. As they crested the ridge and began to make their way through the surrounding grassland, they heard a soft humming sound. Looking up, they saw a flying car sailing off into the horizon.</p><h3>VI. Summary</h3><p>What did we learn from this completely stylistically original story?</p><p><strong>First</strong>, that hash tables, tries, merkle trees, and patricia trees are all do essentially the same thing: they let you map <em>keys</em> to <em>values</em>. While there are differences between them, this is essentially what they do.</p><p><strong>Second</strong>, in computer science, nothing is free (but some things are cheap). Everything has a trade-off. Hash tables are fast, but have some randomness. Tries are fully determinstic, but slower. Merkle trees have nice security properties, but use a more complicated algorithm and are slower to update. Finally, Patricia trees are faster than Tries and Merkle trees, but require an even <em>more</em> complicated algorithm.</p><p><strong>Third</strong>, Patricia trees are useful for blockchains because they let you “prove” a potentially large amount of data is correct, without having to store all of that data. This is very convenient: you can have a big tree with a lot of data (such as all of the transactions in the last 24 hours), but you only have to store a few numbers (like 0x323757382) on the actual blockchain. You can keep the rest of the data on a regular database somewhere and know that no one will be able to tamper it and get away with it. Note that here the blockchain is only <em>part</em> of the system: it is co-dependent on other data stores to function.</p><p><strong>Fourth</strong>, the hash function is the magical machine that makes all of this possible. The design and implementation of hash functions has been the ongoing work of computer scientists for decades, and they are very hard to get right. You should take a moment and appreciate the years of work that made this magical technology possible.</p><p><em>Originally published at </em><a href="http://kronosapiens.github.io/blog/2018/07/04/patricia-trees.html"><em>kronosapiens.github.io</em></a><em> on July 4, 2018.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d8f20efc98d4" width="1" height="1" alt=""><hr><p><a href="https://medium.com/hackernoon/trie-merkle-patricia-a-blockchain-story-d8f20efc98d4">Trie, Merkle, Patricia: A Blockchain Story</a> was originally published in <a href="https://medium.com/hackernoon">HackerNoon.com</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Reputation Systems: Promise and Peril]]></title>
            <link>https://medium.com/hackernoon/reputation-systems-promise-and-peril-ae0af60595ea?source=rss-bfe1f18e2a74------2</link>
            <guid isPermaLink="false">https://medium.com/p/ae0af60595ea</guid>
            <category><![CDATA[economics]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[promise-and-peril]]></category>
            <category><![CDATA[reputation-system]]></category>
            <category><![CDATA[reputation-management]]></category>
            <dc:creator><![CDATA[Daniel Kronovet]]></dc:creator>
            <pubDate>Fri, 29 Jun 2018 00:00:00 GMT</pubDate>
            <atom:updated>2018-08-02T15:28:48.475Z</atom:updated>
            <content:encoded><![CDATA[<h3>I. Introduction</h3><p>When dealing with unfamiliar people, places, and companies, reputation can be powerful. Having a good reputation oneself will lead others to treat us more favorably, and we are more willing to deal with people and business which are themselves well-reputed. The role and importance of reputation in markets has been known by economists since at least the 1970’s, with George Akerlof famously defining the <a href="https://en.wikipedia.org/wiki/The_Market_for_Lemons">“Market for Lemons”</a>, describing how a market can fail when participants cannot properly gauge the quality of the offerings.</p><p>Historically, much of reputation has been implicit, qualitative, and socially embedded: we retain other’s reputations in our memories and learn them through our own experiences or by discussing with others. Companies embark on public relations campaigns to build goodwill with their customers.</p><p>Sometimes, reputation can been made explicit and quantitative: a notable example being the ubiquitous credit score, which condenses our financial reputation to a single number, bringing benefits to the financially stable and consequences to the rest. Other examples are the credit rating systems such as those offered by Moody’s and S&amp;P, which evaluate and rate securities (with some success and a number of <a href="https://en.wikipedia.org/wiki/Subprime_mortgage_crisis">high-profile failures</a>).</p><p>In the last two decades, firms have emerged which seek to leverage new technologies and data to more accurately rate and evaluate the business and people of the world. Companies like Foursquare and Yelp seek to establish reputations for businesses based on the wisdom of the crowd, while OnDeck, Lemonade, and the like aim to determine accurate credit worthiness by ingesting myriad traces of our digital identities. Sharing economy services like Uber and AirBnB encourage us to rate our interactions, helping us to find the best (and avoid the worst) cars to ride in and homes to stay in.</p><p>None of these methods are perfect and a quick internet search will <a href="https://www.inc.com/magazine/20100201/youve-been-yelped_pagen_3.html">surface</a> <a href="https://qz.com/1244155/good-luck-leaving-your-uber-driver-less-than-five-stars/">much</a> digital ink spilt in their critique. Yet this project seems likely to continue, and on an increasingly large scale: in the past year, China has begun to roll out a <a href="https://foreignpolicy.com/2018/04/03/life-inside-chinas-social-credit-laboratory/">“social credit system”</a> which extends the idea of a credit score to encompass more and more facets of life: our behavior in traffic, volunteer hours, and the like.</p><p>Many see dark clouds on the horizon. This mood is most famously captured by the now-classic Black Mirror episode <a href="https://en.wikipedia.org/wiki/Nosedive">“Nosedive”</a>. This episode, set in a futuristic society in which every personal interaction is rated from one to five, tells the story of a woman who experiences personal ruin while engaging in an ill-fated attempt to boost her rating.</p><p>The remainder of this essay will take up the question of reputation systems, their characteristics, and the problems they seek to solve in an attempt to understand the promise and peril of such systems, and to discern what kinds of reputation, in what contexts, may help us to better navigate and thrive in the world of the future.</p><h3>II. Peril</h3><p>Before turning to the promise of reputation systems, it is prudent to consider where other systems have failed. Understanding the failure of past systems will help us to discern design principles for the future.</p><p>Here, we will compare the fictitious world of “Nosedive” with the reality of Uber ratings, discuss the early experiments of China’s social credit system, and briefly mention the rating products of Foursquare and Yelp.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/401/1*f5DJBSf7u4GQKPRtZePr4A.jpeg" /><figcaption>Lacie from Nosedive (src: Wikipedia)</figcaption></figure><p><strong>Both “Nosedive” and Uber make prominent use of 5-star ratings.</strong> The key difference between them, however, is that Nosedive’s fictional app combines ratings for <em>every</em> interaction into a single score, while Uber’s ratings are highly specific, concerning the limited interaction between a rider and a driver. Human beings are deep and complex creatures, with different relationships to different people. To distill our value and worth into a single number is to both deny the complexity of the individual and guarantee shallow and invidious comparisons. In the case of Uber, however, such a score can be defended on the grounds that it is highly <em>specific</em>: someone can meaningfully be said to be a better or worse driver or passenger of a car.</p><p>The first thing, then, is to distinguish between some sort of <em>global</em> or <em>general</em> reputation and a task-specific one, with the latter being more meaningfully captured by a single number. Note that there are always shades of nuance and complexity: even the act of driving a car has many dimensions (reaction to traffic conditions, car maintenance, and so on), but a single number can get us most of the way there. Put another way, the <em>benefit</em> of the rating can be argued to outweigh the <em>cost</em> (there is always some cost to measuring anything).</p><p><strong>Now, what of China’s social credit system?</strong> It is an ambitious, country-wide effort being implemented jointly by many local governments and agencies. There is no “single” system: rather, different towns and cities are experimenting with slightly different systems. To hear the reports of citizens in these different locales is to receive valuable insights.</p><p>Listening to these citizens, we hear that social credit systems which seek to simply enforce the law (traffic violations, etc) are well received. This is in large part due to the lack of trust Chinese citizens have in the government courts. In a country with weak rule of law, a social credit system which steps in for the judiciary is welcome. In other cases, where officials have ambitions for social credit to encompass all facets of human character, the system is less welcome. Jurisdictions which attempt to track volunteering hours, charitable donations, and the like place distorted expectations on their citizens. A case study in <a href="https://en.wikipedia.org/wiki/Goodhart%27s_law">Goodhart’s law</a>, this type of system creates perverse incentives, replacing human relationships with rigid and quickly-counterproductive policy.</p><p>What can we take from the experiment in China? First is that people will welcome a reputation system which limits itself to objective, agreeable standards of conduct. To forego any reputation system out of fear of a dystopian outcome is to run into the arms of another darkness: fraud and corruption. The second is that while some simple behaviors can be measured numerically, higher level understandings of goodness and value need to be treated with more care, lest we create more problems than we solve. Complex phenomena require more complex representations and measurements: it may be that some things are best left unmeasured, rather than measured poorly.</p><p><strong>Finally, let’s consider the local search services provided by Foursquare and Yelp.</strong> These products allow people to submit numeric and text reviews of bars, restaurants, and the like, and determine ratings for these venues using various proprietary algorithms (full disclosure: I was previously a machine learning engineer at Foursquare). While these companies provide useful services, they are not without their drawbacks. First, the subjectivity of the reviews invites skew and bias: Yelp in particular is known for inviting quite hostile reviews. Second, and more subtly, in reality there is no “bar”, there is only “me at the bar”. <a href="https://foursquare.com/v/happyfun-hideaway/518fed68498e67e8b8c705b0">Happyfun Hideaway</a> is my favorite bar in New York, but the hipster-chic aesthetic isn’t for everyone. I don’t particularly like <a href="https://foursquare.com/v/employees-only/41e46880f964a520d41e1fe3">Employee’s Only</a>, but it’s “better”, ratings-wise. Assigning every venue a single number paints only a partial picture.</p><h3>III. Promise</h3><p>What then to do? The first thing to recognize is that different problems will call for different types of reputation systems.</p><p>For <strong>concrete, specific skills</strong>, a single numerical reputation could work well. <a href="https://colony.io/">Colony</a>, a blockchain startup, is taking this approach, developing a reputation system in which participants earn reputation within a taxonomy of skills and specialties. If I write a well-received blog post, I can receive reputation as a “writer”. If I write some code to speed up our testing, I earn reputation as an “engineer”. The incorporation of a skills taxonomy allows for striking a balance between specificity and generality and allowing for earned reputation to be put to use, for example to weight votes (full disclosure: I am currently employed by Colony as an engineer).</p><p>For <strong>higher-level aspects of our personality and character</strong>, however, numerical representations will fail. Numbers beg to be gamed, and so here we should prefer systems in which there is no way to “win”. To understand how we might approach this type of reputation, it will pay to take a leaf out of the psychologist’s book. Since the 1980’s, the <a href="https://en.wikipedia.org/wiki/Big_Five_personality_traits">“Big Five”</a> model of personality traits has been a common (if imperfect) tool to assess individual personality. In this system, each person’s personality is seen as a basket of five trait dimensions (specifically, “openness”, “neuroticism”, “conscientiousness”, “agreeableness”, and “extraversion”). As each trait is a dimension, or dialectic, there is nothing inherently “better” or “worse” about being highly agreeable, extroverted, and so on compared with the opposite. Conscientious individuals are more diligent and hardworking, but also stubborn: in contrast, people low in conscientiousness are more flexible and spontaneous, but can be unreliable. In one setting, one might be preferable, in another setting, the other. The essential difference, compared to a single 1–5 scale, is that no position on the scale is better or worse, only different.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xpunxhHHul_gBQzqnx_cag.png" /><figcaption>The “Big Five” personality traits (src: Wikipedia)</figcaption></figure><p>Personality and character are tricky issues, and the question is far from solved even today. To develop our sense of possibility, it is worth taking a quick tour of some “artistic” approaches to personality: specifically, the classic Dungeons &amp; Dragons <a href="https://en.wikipedia.org/wiki/Alignment_(Dungeons_%26_Dragons)">alignment system</a> and the <a href="https://en.wikipedia.org/wiki/Enneagram_of_Personality">Enneagram</a>.</p><p>Dungeon &amp; Dragons is a classic game in which players take the role of adventure heroes and explore various fantasy worlds. A major aspect of each character is their “alignment”, defined as a position on two axes: the “Lawful/Chaotic” axis and the “Good/Evil” axis. A character can be Lawful Good (a soldier, for example), Chaotic Good (a revolutionary leader), Chaotic Evil (a common criminal), Lawful Evil (a rapacious business leader), and so on. As alignment exists along two axes, or dialectics, no one position can be better or worse than another: only different. We have the freedom to decide, based on the situation (and ourselves), what type of alignment is most suitable.</p><p>The Enneagram is a popular personality system which divides people into nine “types”, each type being associated with a set of fears, desires, strengths, and weaknesses. As with the other personality systems described, the essential feature is that no type is “better” or “worse” than another, only different.</p><p>For <strong>inter-personal and inter-item compatibility</strong>, a third approach is called for. The challenge here is that different people like different things. Someone I find amusing you might find annoying. A restaurant you hate I could love. While you could argue that the flexible personality types described earlier could be used here (“put two extroverts together”, etc), it is worth thinking what other options exist. There is much research in the social science (and more recently, machine learning) literature on predicting the compatibility of a single person-item or person-person matching. A common theme in these approaches is to take a large dataset of interactions (for example, individual movie reviews, or whether two roommates like each other) and to extract a set of high-level traits which can be used to predict the success of new interactions (such as new movies or new roommates). While the specifics of these approaches is beyond the scope of this essay (you can learn more <a href="https://datajobs.com/data-science-repo/Recommender-Systems-[Netflix].pdf">here</a> and <a href="http://jmlr.csail.mit.edu/papers/volume9/airoldi08a/airoldi08a.pdf">here</a>), the essential takeaway is that no potential movie or roommates is inherently good or bad, just good or bad <em>for you</em>.</p><p>One could imagine building a roommate-matching system which asks participants to give a simple thumbs-up or thumbs-down as to whether they would live with a given roommate again, and use this to infer implicit community memberships (“partier”, “neat freak”, “chiller”, and the like), and match people based on whether they would likely be compatible. One advantage of this approach (as opposed to one where people self-evaluate on these categories) is that it provides a stronger signal (based on actual experience) and arguably avoids some self-perception biases.</p><h3>IV. Conclusions</h3><p>The aim of this essay was to review the idea of reputation and rating systems, and to argue that while such systems <em>can</em> be immensely useful, they must be developed with care. In particular:</p><ul><li>Numerical scores are appropriate only when they point to <strong>specific</strong> skills and competencies.</li><li>Personality is multifaceted and dialectic: there should not be “better” or “worse”, only “different”.</li><li>What one person likes someone else might dislike: the viability of a pairing should be a function of what is being paired.</li></ul><p>For those of us living in dense urban centers, it is common to hear people describe feeling “alone in a crowd”. While we are surrounded by people, they are strangers to us. Many of us lack the support infrastructure to navigate relationships in a transient world; this leads to loneliness. We must exercise caution in our personal and professional dealings to avoid being scammed, while people with quiet integrity run the risk of being overlooked in favor of the loud. All of this leads to friction, cost, and pain. Appropriate reputation and rating systems can help us navigate and thrive in our ever-expanding world, while poorly-designed systems can cause problems bigger than they sought to solve.</p><p>The continued development of such systems is inevitable; the task of building them well falls to us.</p><p><em>Originally published at </em><a href="http://kronosapiens.github.io/blog/2018/06/29/reputation-promise-peril.html"><em>kronosapiens.github.io</em></a><em> on June 29, 2018.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ae0af60595ea" width="1" height="1" alt=""><hr><p><a href="https://medium.com/hackernoon/reputation-systems-promise-and-peril-ae0af60595ea">Reputation Systems: Promise and Peril</a> was originally published in <a href="https://medium.com/hackernoon">HackerNoon.com</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Ritual Therapy]]></title>
            <link>https://medium.com/@kronosapiens/ritual-therapy-dfa1bf3d4a89?source=rss-bfe1f18e2a74------2</link>
            <guid isPermaLink="false">https://medium.com/p/dfa1bf3d4a89</guid>
            <category><![CDATA[hiking]]></category>
            <category><![CDATA[therapy]]></category>
            <category><![CDATA[ritual]]></category>
            <dc:creator><![CDATA[Daniel Kronovet]]></dc:creator>
            <pubDate>Sat, 09 Jun 2018 21:22:46 GMT</pubDate>
            <atom:updated>2018-06-10T06:19:56.635Z</atom:updated>
            <content:encoded><![CDATA[<p>I.</p><p>A few weeks ago I picked up Alejandro Jodorowsky’s <a href="https://www.innertraditions.com/psychomagic.html"><em>Psychomagic</em></a><em>, </em>essentially an exposition of his theories of ritual therapy. I’ve been a fan of Jodorowsky for a while, and it’s been great to get a closer look at the thinking which underlies his work.</p><p>The essential thesis is roughly this: while our conscious minds speak the language of words, our unconscious minds speak the the language of symbols. By concocting highly symbolic ritual acts, patients can communicate with their unconscious and heal themselves of past psychological trauma and the oft-associated psychosomatic symptoms.</p><p>Jodorowsky is first to admit that his method is more art than science, with only anecdotal evidence for the method’s efficacy. In one example, a woman feeling adrift and grappling with feelings of absence and abandonment following her father’s suicide is told to buy thirteen oranges, giving twelve to patients at a hospice and eating the thirteenth in exactly twelve minutes. She reports experiencing connection with the hospice patients that she had longed to have with her father, and while eating the final orange, experiences a new gratitude for her own life.</p><p>Some of the prescriptions seem a bit fantastic (one man, struggling in business, is told to tape gold coins to the soles of his shoes, so that all day he will be walking on gold), but I cannot help but see a certain logic to this theater. Someone lacking confidence cannot be given confidence simply by speaking affirmations in a mirror — rather, confidence is built piece by piece by <em>acting</em> in ways which build self-esteem. In a similar way, I can see why someone struggling with past traumas or failures may need to take some <em>actions</em> which allow them to overcome those experiences. The key idea is that words alone are not enough; it is meaningful action that matters: hence <em>ritual</em> therapy. Knowing exactly which actions to prescribe for a given ailment is, of course, the tricky bit, and the reason why Psychomagic is the most fringe of practices.</p><p>II.</p><p>I can relate a recent experience which I would describe as “ritual therapy”, concerning a four-day solo backpacking trip I took in the north of Israel.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/470/0*I70FQb7Euiu2emEp" /><figcaption>The terrain of the Galilee</figcaption></figure><p>If the trip was the therapy, what was the trauma? It was twofold, although admittedly to describe these experiences as “traumatic” is to be a bit, well, dramatic.</p><p>The first incident occurred during the Summer of 2010, when I went on a camping trip with four other friends. At one point during the trip, we decide to go exploring the hills around our campsite. The terrain was sparse woodland, with hills, cliffs, and the occasional pond. During our jaunt, we arrive at the base of a rocky cliff, and we decide to summit the face. We begin to ascend, with me in the rear of the group. About halfway up, I start getting nervous: we are climbing ad-hoc, with handholds and footholds being whatever seems to make sense. We have no guarantee of finding a path, and at this point we’ve set ourselves up for quite a drop. I’m feeling shaky, and I eventually tell the group that I’m turning back. They encourage me to continue, but I am resolved. I descend while they continue their climb. A short while after I reach the base, I see them complete their summit and reach the top of the cliff. By this point, I am exploring the wooded area adjacent to their hill, and telling myself I’m having a good time, but I’m starting to feel a sense of distress. I see them at the top of the hill, and start to think about the best way to rendezvous. Eventually I circle around and we are reunited; but there’s a faint disconnect. They went forward while I turned back. A faint disconnect which feels to me like a gaping chasm. Everyone is quite nice about it (one girl in the group took care to later emphasize the other activities which we “did together”), but what was done was done. A number of years later, I was talking to one of the friends who had gone on the trip, and he brought up that moment when I turned back. He asked if I ever thought about it, to which I responded I did. He posed the question of what might have been different had I stayed with the group. I’ll never know. But I do know that that moment of retreat has stayed with me ever since, much brighter in my memory than any other aspect of that trip.</p><p>The second incident occurred in the Fall of 2012, when I was living on the kibbutz in Israel. I had heard of the “Sea to Sea” hike, a four-day trip through northern Israel, starting at the Sea of Galilee and ending at the Mediterranean. I was interested in giving it a try, and recruited three of my classmates to come with me (two boys, two girls in total). We gathered food and equipment, I got a map, and we set out. In my mind trip is going well enough, but halfway through the other three tell me that they want to abandon the trail and take a bus back to the kibbutz. I’d like to keep going, but I’m not ready to finish the trip alone. We go back. While not as difficult a retreat as before (in large part because we abandoned the trail together), it was another lingering memory of turning back instead of going forward.</p><p>These incidents has left a mark on me; creating a narrative of someone who “turned back”, who didn’t have the skill/strength/fortitude/courage to press onward. And while I enjoyed many professional and academic successes in the years to come, I failed to push myself in the outdoors, and so nothing I did quite took away these feelings. Which takes us back to the trip I completed.</p><p>As you may have guess, this trip I finished was the same trip that I abandoned six years prior. I undertook the trip solo, not having found anyone with an aligned interest and schedule. The route involved hiking about 20km per day, through open country and rocky riverbeds, crossing rivers and going long stretches away from water, people, and civilization. While not the most dangerous of all possible routes, there were risks involved with hiking solo; in this case, the risks were the point. I needed to communicate to myself, not with affirmations but with actions, that I could complete a non-trivial undertaking in the outdoors. Completing the trip allowed me to release these long-held doubts and feelings of failure.</p><p>I returned from the trip with a certain feeling of lightness. I had done the thing. I wasn’t the person I was worried I was. I had redeemed my earlier failures. It was a ritual therapy: in a sense, an act of Psychomagic.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/470/0*qfeg7awlJkAcPZHz" /><figcaption>Finishing at the Sea of Galilee ;)</figcaption></figure><p><em>Originally published at </em><a href="https://kronosapiens.com/2018/06/09/ritual-therapy/"><em>kronosapiens.com</em></a><em> on June 9, 2018.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dfa1bf3d4a89" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Proof of Work vs Proof of Stake: a Mirror of History]]></title>
            <link>https://medium.com/hackernoon/proof-of-work-vs-proof-of-stake-a-mirror-of-history-f61c0934cd86?source=rss-bfe1f18e2a74------2</link>
            <guid isPermaLink="false">https://medium.com/p/f61c0934cd86</guid>
            <category><![CDATA[philosophy]]></category>
            <category><![CDATA[bitcoin]]></category>
            <category><![CDATA[blockchain]]></category>
            <category><![CDATA[proof-of-stake]]></category>
            <category><![CDATA[proof-of-work]]></category>
            <dc:creator><![CDATA[Daniel Kronovet]]></dc:creator>
            <pubDate>Fri, 16 Mar 2018 00:00:00 GMT</pubDate>
            <atom:updated>2018-07-27T18:44:10.106Z</atom:updated>
            <content:encoded><![CDATA[<h4>In which we observe that two things which appear quite different are in fact very much the same.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iMKDxeSDixvjTSz3q9SBww.jpeg" /><figcaption>Map of Europe, circa 1595 (src: <a href="https://www.pinterest.co.uk/pin/505388389411524578/">https://www.pinterest.co.uk/pin/505388389411524578/</a>)</figcaption></figure><h4>I. PoW vs. PoS</h4><p>I went to an amazing debate recently at <a href="https://www.digital.nyc/incubators/grand-central-tech">Grand Central Tech</a>.</p><p>Hosted by Crypto NYC, the event pit two prominent Proof of Work advocates against Proof of State advocate Nate Rush in an engaging debate over arguably the most important question facing the blockchain community.</p><p>For those unfamiliar, “Proof of Work” and “Proof of Stake” are types of what is known as a “consensus algorithm”: a set of rules for how a blockchain decides what transactions become “canonical” and which do not. A blockchain’s consensus algorithm is a little bit like its constitution: the fundamental rules which define it and determine its behavior.</p><p>Of the two, <strong>Proof of Work (PoW) </strong>is currently the de-facto standard, powering both Bitcoin and Ethereum, the world’s two largest cryptocurrencies. In essence, PoW requires participants to demonstrate that they have done a certain amount of computational work in order to place a transaction into the canon. By imposing this work requirement, Bitcoin and Ethereum make it economically inefficient to attempt to place fraudulent transactions into the canon. As long as the “good guys” have at least half of the computers, the “bad guys” will not be able to commit fraud, and the transactions in the blockchain can be trusted as being accurate.</p><p>Proof of Work was pioneered in 2008 in the original Bitcoin paper, written by the pseudonymous Satoshi Nakamoto. It was the first algorithm to solve the “double spend problem”, the achilles heel of prior cryptocurrencies. It is lauded for being effective, simple to understand, and simple to implement. Its essential drawback is its profound energy requirement: since the “work” of Proof of Work involves arbitrary (and unproductive) computation, participants in the network are essentially competing to waste the most electricity.</p><p><strong>Proof of Stake (PoS)</strong> is a more recent, experimental consensus algorithm often touted as a potential replacement for Proof of Work, and seeks to avoid PoW’s substantial energy cost. It does so by making the determination of which transactions enter the canon a function not of computing power, but rather of possessing and committing tokens. A participant commits some tokens (such as Bitcoin or Ether) in order to have a say in which transactions become canon. The more tokens they commit, the more likely they will get to decide. If you commit coins to try to include fraudulent transactions, there is a high chance those coins will be lost. Therefore, PoS creates an incentive to try and include honest transactions over fraudulent (assuming most of the token holders will prefer honest transactions). The challenge with PoS is that it is more vulnerable to certain types of attack, such as those coming from manipulations of the price of the token. There is much interest in the Ethereum community about moving to a Proof of Stake algorithm in the future, and it is an area of active research.</p><h4>II. Realpolitik vs. International Institutions</h4><p>Listening to the debate (and reflecting on reading I had done in the past on these protocols), I could not help but be reminded of a similar debate occurring in courses of international affairs. In international affairs, there is the notion of <strong>realpolitik</strong>, developed in Germany in the 19th century, is a political philosophy which holds that, in the absence of a hypothetical strong global government, practical realities like military strength determine the relationships among nations. They view nations as co-existing in a “state of nature”, in which ideologies, values, and norms are irrelevant or minimally impactful. As a philosophy, it is known to be effective, simple to understand, and simple to implement.</p><p>In absence of trusted diplomacy, states are incentivized to engage in arms’ race, competing to build stronger armies with more advanced weaponry. Resources are diverted from social goods like schools and hospitals, as the prosperous nation with a weaker military will inevitably be overrun. Periods of stability exist either when one nation is so powerful that there can be no challengers (a hegemony), or when multiple nations are similarly matched, and so the settle into a peaceful, if tense, stalemate (a multipolar world). Wars litter history like so many punctuation marks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*4iqqFtKorug52uO8xlvNug.jpeg" /><figcaption>Mount Washington Hotel, site of the Bretton Woods Conference (src: Booking.com)</figcaption></figure><p>Towards the end of World War II, the allied nations of the world came together in Bretton Woods, New Hampshire and set to work developing an alternative. This war of all against all had gone on for too long, they said. We will develop an <strong>international system</strong> of organizations and connect the nations together via a network of finance and trade, with mechanisms to debate, regulate, and resolve disputes. We will beat our swords into ploughshares, and learn war no more.</p><p>This system has proven to be very effective. While it has facilitated the intense concentration of global wealth into the hands of a few, it has also succeeded in preventing total war among the great powers, and in general is understood to have promoted science, technology, culture, and an improved quality of life for billions.</p><h4>III. Dialectics Rule Everything Around Me</h4><p>It is interesting to contrast these two systems of international affairs with the two consensus algorithms being presented, and to reflect on the essential similarity in the problems they solve. In both cases, we seek to achieve order among competing, possibly antagonistic parties. The first approach is to hinge the result on a show of force. This approach works in as much as force is a proxy for other desirable attributes: control over resources, organizational ability, conviction, and it is simple to understand and to implement. This approach is limited in that it necessitates hard-won energy being put towards no productive work. Inasmuch as the essential process of life is the acquisition and allocation of energy, this is necessarily suboptimal.</p><p>The second approach is to attempt to create a more nuanced system of mental relationships, and to hope that this framework is robust to the inevitable attempts at exploitation and capture by those seeking influence over the outcomes. Unlike the first approach, this approach is more vulnerable to being undermined, and in ways which can be difficult predict, detect, and respond. If we as a community — a cryptocurrency community, or a world community — have the skill and commitment to execute on the vision, then this system would allow participants to contend in a way that is fair and in which outcomes hinge on a display of desirable attributes, but also highly efficient, in as much as it consumes as little energy as possible to preserve energy for other tasks. Higher risk, but higher reward.</p><p>Looking at these two debates, which is essentially one debate, it is hard not to take a moment to look ever further back and draw a second connection, between these debates and our own development as animals, in which we began as simple bodies but over time developed sophisticated minds. And the same way, the tensions between these schools of thought is the same as the tension between our bodies and our minds: the mind carries risk, and often we wonder if the body was all we needed. On the other hand, the body is limited, and the mind holds such great promise. Do we have the skill, vision, and fortitude to make the leap, without losing touch with where we came?</p><p><em>Originally published at </em><a href="http://kronosapiens.github.io/blog/2018/03/16/pow-vs-pos.html"><em>kronosapiens.github.io</em></a><em> on March 16, 2018.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f61c0934cd86" width="1" height="1" alt=""><hr><p><a href="https://medium.com/hackernoon/proof-of-work-vs-proof-of-stake-a-mirror-of-history-f61c0934cd86">Proof of Work vs Proof of Stake: a Mirror of History</a> was originally published in <a href="https://medium.com/hackernoon">HackerNoon.com</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>