<?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 Nicholas Teague on Medium]]></title>
        <description><![CDATA[Stories by Nicholas Teague on Medium]]></description>
        <link>https://medium.com/@nicholasteague?source=rss-191d0ae6b768------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*Otmu3NX3p-QYQIH_NgiHrA@2x.jpeg</url>
            <title>Stories by Nicholas Teague on Medium</title>
            <link>https://medium.com/@nicholasteague?source=rss-191d0ae6b768------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 11:54:17 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@nicholasteague/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[Aphorisms for All]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/aphorisms-for-all-1f61a889a861?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/1f61a889a861</guid>
            <category><![CDATA[aphorisms]]></category>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Sun, 13 Sep 2026 20:23:24 GMT</pubDate>
            <atom:updated>2026-09-21T03:01:34.547Z</atom:updated>
            <content:encoded><![CDATA[<h4>Towards an end of the loose ends</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G2yojqoWE-bUF-PrJW0kBQ.jpeg" /></figure><ul><li>We should be trying to reward those people who seek to prevent wars at least as much as those who seek to win them.</li><li>It adds considerably more value to a discussion to contribute strong opinions held loosely rather than soft opinions held strongly.</li><li>A fundamental flaw of AI practice may arise when evaluating human speech without considering the audience where it is directed.</li><li>We’ll know that AGI has arrived once people stop measuring success based on who generates the most tokens and instead based on who generates the least tokens.</li><li>If we can’t succinctly articulate and constrain those unstated biases that draw from user profiles and external context in interactions, it is unrealistic to expect that we will ever be able to benchmark and validate ethics of deployed systems.</li><li>True innovations of isolated protagonists only solves yesterday’s challenges not tomorrow’s.</li><li>Adherence to policy is less critical than adherence to human rights and values, otherwise why are humans making decisions at all?</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DosHbga6ol051dsV9TQsVA.jpeg" /></figure><ul><li>Sustainable participation in the high tech economy isn’t about establishing the most dominant tech stack it is about establishing the most sustainable framework.</li><li>The people making the toughest decisions about how AI systems are built and are to reach consumers are the ones who spent their entire lives trying to figure out the best way to multiply matrices.</li><li>Determinism in scaled settings locks in whatever forms of algorithmic bias and flaws are impacting specific individuals.</li><li>If you haven’t voiced opinions about macro deployment considerations of modern AI practice then you haven’t voiced opinions about modern AI practice.</li><li>There are two kinds of people who post to open social media: those who have so many people that want to hear what they say that they will brave the lunatics to reach them, and those who have so few people that want to hear what they say that they will brave lunatics to find them.</li><li>How can we amend Judicial processes to incentivize forgiveness?</li><li>There are some domains where a timely solution is more important than timely justice. (Justice with respect to the observer is probably sufficient.)</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ULjKK_RXgFjFrUlsqgTSNQ.jpeg" /></figure><ul><li>The root of all evil is all or nothing based incentives and post hoc rationalizations.</li><li>It is an under appreciated talent to defend your own reputation without needlessly stepping on the reputation of others.</li><li>The number one sign of a corrupted human system is when trivial formalities become staging grounds to contest otherwise settled matters.</li><li>It’s much more important to forgive humans than to forgive organizations built around processes that can never be corrected without the involvement of humans.</li><li>Better to have people dislike you for being pretentious / condescending / arrogant than dishonest.</li><li>It is easier to defend someone whom an opponent doesn’t realize is an ally.</li><li>Invest in companies whose primary competitive advantage does not originate from whose cloud they are using.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*c6W6bmEStU7Uydh_OcjZ2g.jpeg" /></figure><ul><li>Thin tails are high stress with long term survivability. Fat tails are low stress with unpredictable timing of flaws.</li><li>Conjecture: Fault tolerance with small number of variables is considerably more destabilizing than fault tolerance with a large number of variables.</li><li>SciFi: a man spends majority of his adult life attempting to give back to his community through public writings proposing bridges between research and philosophy, but before his most important work is published an unknown authority places judgement based on state of mind a decade ago.</li><li>Meaningful open source contributions earn a professional access to more desirable economic settings. Those who don’t find that premise reasonable have no business whatsoever leveraging open source resources in their work.</li><li>In an algorithmic driven economy, creative professionals who wish to grow their own traction should be weary of getting into business with those large firms who compete in “attention economy”.</li><li>A perk of playing piano is training your brain to think and act on multiple thoughts in parallel. Most other instruments are sequential. Piano expands your ability to nimbly reason around multiple threads.</li><li>I hereby call preemptively invalid any form of critique or criticism towards my entire portfolio of work that has not been communicated to me directly in a manner that I was aware of an origin or how to respond to the sender.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yMuNfMTINz8eV_RttHeCcw@2x.jpeg" /></figure><p>Copyright Nicholas Teague 2026</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1f61a889a861" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/aphorisms-for-all-1f61a889a861">Aphorisms for All</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Ethics of Scaling Compute]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/ethics-of-scaling-compute-78559cb7e290?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/78559cb7e290</guid>
            <category><![CDATA[ethics]]></category>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Thu, 10 Sep 2026 06:45:29 GMT</pubDate>
            <atom:updated>2026-10-04T21:20:16.301Z</atom:updated>
            <content:encoded><![CDATA[<h4>Towards a sustainable peace through better systems</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CmVd5LjoO9QJhU0WPmyQtQ@2x.jpeg" /></figure><p>These are some of my own takeaways from Attempting to follow from a distance the quantum computing field for several years in capacity of a public markets investor:</p><p>Setting aside algorithmic conventions, the only reason that there are still domain experts who will look you straight in the face and tell you that quantum computers [1] will never outperform classical computers at all is because they are often of probabilistic utility where classical are often deterministic. However classical computers, when you account for their exposure to edge case scenarios, are also probabilistic. Just of a different distribution…</p><p>The question of whether or not quantum computing dominates classical resources (“quantum advantage”) isn’t just a question of probabilistic distribution characteristics. It’s also a simple matter of having different regimes of application scales where the resources become appropriate. Ideally quantum computing systems should be reserved for when the available hardware becomes capable of addressing the largest scaled settings, where their strengths may be particularly impactful, not just in latency but surrounding the significantly important implications of <strong><em>Carbon Intensity</em></strong> of our economy’s scaled compute (not to mention surrounding infrastructure costs). My own expectation is that the best path to market for future quantum computing vendors will be to focus on what new business models may arise when performant optimization becomes accessible to scaled systems in new settings.</p><p>This isn’t a scientific observation but kind of my own expectation that the probabilistic nature of solution sampling may be more tolerable to human participants in those scaled systems as well. A human trying to operate around probabilistic algorithms who isn’t trained to understand what it means that algorithms are operating probabilistically may misinterpret sampling from different modes as some form of indication of failure. The reality is that for any domain of uncertainty in which multiple candidate trajectories are accessible those different directions of information are actually helpful, no matter how confusing it may be to those accustomed to single minded classical alternatives.</p><p>The reason that commercializing pure play quantum computing firms was a nearly intractable Problem from the very beginning is that widespread convention of modern business practice is that you have to prove yourself in an early use case before getting access to scaled settings. Kind of like how GPUs became viable first as video game engines. That simply doesn’t work in quantum computing where the systems have a fundamental disadvantage in low scale settings (where deterministic algorithms can be expected to dominate).</p><p>In quantum computing, the route to viability isn’t to establish a scalable business model that you can then fund and scale out in the traditional venture capital (VC) sense. The only realistic way to demonstrate the value is to apply to a business domain that already has established access to a scaled Setting. The fact that pure play vendors can only realistically demonstrate value in already established scaled settings means they have almost no leverage to negotiate with established big tech (beyond those sources of leverage that are unethically established). After all there are no shortage of substitutable qc vendors that can be directly applied if an established resource gets out of hand. Which is kind of how it should be. (Although short term advantages originating from hardware or software may be realistically expected between different conventions from time to time, those advantages should reasonably be channeled to higher clearances where they belong.)</p><p>As an aside, any call to integrate stochasticity into lower scale / lower stakes settings is itself a way to allow access of higher tier resources to an improved scale of reach. The ethical quandary then becomes how to identify which settings are of sufficiently low stakes that nondeterministic outputs are appropriate (for obvious reasons one wouldn’t want to blindly integrate stochasticity into eg safety sensitive settings without accounting for the range of exposures in play). My <a href="https://medium.com/from-the-diaries-of-john-henry/chatflavors-e318ee83e4a1">ChatFlavors</a> proposal [2] could be a way to channel a stakes assessment into diverse settings relying on contexts that can only be expected as accessible to foundational portals.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wJuJR-QHY5oPWSxibg3NBg@2x.jpeg" /></figure><p>There will always remain a need to evaluate how to prioritize our allocation of non-infinite computational resources. The purposes of proxy competitions isn‘t to determine who has the biggest balls. The alternative to healthy proxy competitions is unhealthy direct competitions. Traditionally this is what markets have been for. The hard part then becomes how we can structure markets in a manner that effectively adapts to the new reality of where new sources of abundance and changing bottlenecks have changed the behavior of markets and the economy. The more critical the need to source serious information of this nature, the more proxy competitions may start to appear less of a silly nature…</p><p>The challenge has always been and remains that benchmarking competitions are only informative to the extent that the patterns aren’t based on single sampled outcomes but based on endurance over prolonged time periods and further with an expectation for potentially changing results with evolving distributions which may be linearly or nonlinearly transformative. The only realistic way to incorporate this is with HUMAN intelligence. That is why peer reviewed research publications and experimental findings and etc still matter. Because until we literally have established a theory of Everything that we can effectively extract towards future states, any realistic expectation of sustainability in a competitive ecosystem demands that we will need to channel the finite intelligence at our disposal to do way more than just tell AI agents to transform lead into gold…</p><p>The ethical challenge then becomes that proxy competitions are less desirable when participants don’t know the basis of competition or why there are a part of it. And if a party is drafted into the domain without such a courtesy introduction there should be an easy out made available, whether by a letter of resignation or perhaps just a reasonable transaction of sale. Ideally we want to carry forward as much of the legal scaffolding that originated from the pre AI era that has already been proven to effectively operate in settings with exposure to diverse legal regimes and complicated trade imbalances. If we try to just rebuild everything from scratch that is not likely to resolve quickly…</p><p>This sounds kind of incredible, but the best way and the right way that we should have the greater public thinking about the future of quantum computing in any open setting is related to questions of if the world someday gets an ability to access a kind of (nearly) infinite compute, how best could it be ethically utilized in a safe and sustainable way. Engineers will need to follow a different path from theoreticians, philosophers, or pastors on how to get there. But all roads should eventually point in that direction. The fact that different paths to get there are necessary is where a lot of misunderstandings may arise about what quantum computing means and what are the implications. A lot of people who think they are opposed to the field are merely opposed to the modern economy.</p><p>As I have on small number of occasions mused about the significance of conglomerates versus pure play firms in public speech, I wish to clarify that my own opinions have evolved over time. And finally I came to the conclusion that as the scale of interactions between a firm and the greater economy grows, the ethics of that firm’s roll and responsibilities with respect to how the greater economy plays out grow as well. There will remain in the future opportunities for small business or even solo founders to have outstanding and heretofore unprecedented levels of success, but if we can’t make the whole ecosystem stable and sustainable the foundations upon which those small businesses rely may not be assured. The largest firms are the ones who by necessity will need to shoulder the incredible Burdens of competing to build and maintain the very foundations upon which we all rely. With the support of our government and hopefully with the continued blessings of our Maker.</p><p>Peace</p><p>[1] Feynman, R. P. (1982). Simulating physics with computers. International Journal of Theoretical Physics, 21(6/7), 467 — 488.URL. <a href="https://doi.org/10.1007/BF02650179">https://doi.org/10.1007/BF02650179</a></p><p>[2] Teague, N. (2026). ChatFlavors. Zenodo. URL. <a href="https://doi.org/10.5281/zenodo.23127680">https://doi.org/10.5281/zenodo.23127680</a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nd8yIo1XqwPjU1vc2CQWEA@2x.jpeg" /></figure><p><em>Copyright Nicholas Teague 2026</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=78559cb7e290" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/ethics-of-scaling-compute-78559cb7e290">Ethics of Scaling Compute</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Economics]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/ai-economics-f06df944cbfe?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/f06df944cbfe</guid>
            <category><![CDATA[economics]]></category>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Fri, 17 Apr 2026 20:56:21 GMT</pubDate>
            <atom:updated>2026-05-09T21:16:42.537Z</atom:updated>
            <content:encoded><![CDATA[<h4>An Incentive Alignments Proposal</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RIgvY1wW_mU_IyID39EWXA.jpeg" /></figure><p>Former Secretary of State Condoleezza Rice recently presented a talk in Orlando surveying her career in politics and interpretations of current events, noting at one point when considering the risks of artificial intelligence a sort of imperfect analogy about how when cavemen first discovered that they could harvest fire, it took a while to figure out how to avoid getting burned. I didn’t consider this an especially helpful model of the risks of artificial intelligence, but it did get me trying on my physicist hat to consider how there may be other similarities to the thermodynamics of fire in how a more widespread deployment of artificial intelligence may play out at the macro scale, which I propose here using an economic framing, serving as a brief introduction to some more extensive proposals on how our economy may need to adapt to the implications of what is in store for us.</p><p>For this less imperfect analogy of fire to work, we can start by simply considering what is taking place in a thermodynamic frame, which can be summarized simply as fire representing the consumption of fuel by way of the chemical reaction combining the elements of oxygen with typically hydrocarbons and releasing carbon dioxide, water and heat energy (where the visual component of flames are a property of resulting elevated temperatures). The only other necessary element is simply some form of catalyst to initiate the reaction, in many cases simply a well timed spark.</p><p>If we try to look at how AI deployment may come to resemble this process, let’s try to find where there could be a fuel, the resulting emissions, and the spark. This is an imperfect model, but let’s consider an economic framing where the AI is made for the explicit purpose of achieving business outcomes (such as in the simplest objective of profit or in more sophisticated settings possibly the erection of strategic moats for sustaining competitive advantages). In this case the “fuel” is associated with those aspects of an economy that are accessible to the competitive participation of AI, which fuel can only be considered as a property in relation to the capabilities of the AI itself (such as may be evaluated based on the sophistication of performance metrics achievable with an AI, access to tools / knowledge / context, or otherwise access across whatever boundaries of participation may be in play).</p><p>I would suggest that using this analogy, we could realistically expect that the most accessible types of ‘fuel’ available to more immediate scaled AI participation will be found in those low hanging components of the labor economy where human participation relies on either creativity with a sufficient margin for imperfections, or otherwise relies on structured human interactions with repeatable elements, in each case those first scalable settings will be those that are currently bottlenecked by human worker attention.</p><p>To consider the potential impact of AI in some domain, rather than trying to reason about intelligence, it is easier to apply a thought experiment of what could happen if you replaced the attention / bandwidth of each worker with a thousand.</p><p>Thus, we could expect that as AI gets further integrated, their economic actions will consume whatever accessible settings may fit these criteria in a manner resembling a spreading brush fire, resulting in the unfortunate but foreseeable crowding out of several aspects of commerce that are traditionally served by small business and recent graduates. There will of course remain plenty of opportunity for skilled trades and services that rely on physical presence and participation (noting further down the horizon a similar potential for shifts towards physical robots possibly resembling taxi vs uber labor conventions), demonstrating that the organizations serving those trades will similarly face a need to evolve and adapt.</p><p>As a small business participant, that raises an important question of how workers can continue to compete in a realistic way adjacent to the coming throngs of AI agents. One way to do so will be to seek to constrain reach, to establish tighter controls on the boundaries of interactions between potential customers and vendors, but that will rarely remain enforceable in any consumer setting without platform lockin, and otherwise more likely to merely serve as a speed bump to progress rather than becoming an enduring moat in any setting without advanced regulatory frameworks.</p><h4>Principles</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-SVH2Hqog-3ied2JUb4p8w.jpeg" /></figure><p>Having laid all of that groundwork, we can then consider a more critical question: how is it that we as human workers can expect to add value and have any reasonable type of career in a sustainable economy. What will it mean to “work” in the coming years when so much of what humans have traditionally offered will need to coexist in a competitive setting directly adjacent to algorithmic agents?</p><p>This essay isn’t intended to provide a detailed implementation framework of how to answer that question, rather I will attempt to lay the foundation of what I personally would consider those minimum viable principles needed in order to align to any reasonable path forward to ensure retaining a sustainable economy that can somehow coexist when AI agents are directly competing with humans. Good news — it is a short list, summarized cleanly as:</p><ol><li>Unless an AI agent has some form of ownership responsibility traceable back to a specific human being (or an organization managed by specific human beings), then it may only operate in an anonymous fashion without access to human rights or privileges. =&gt; Restated: AI’s don’t get the privilege of an identity, and if they attempt to establish one they become exposed to enforcements serving to constrain their reach.</li><li>Those specific human beings with ownership responsibility to AI agents are liable for any crimes and responsible for the ethical values demonstrated through the actions of their agents.</li><li>Human participation in commerce will encourage market based trading mechanisms to allocate compute and other resources to agents, subject to a hard rule that only AI agents meeting the identity criteria are given access to participate in public markets.</li></ol><h4>Humans</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Q0kxukq7v0OZvsPQdWAlRg.jpeg" /></figure><p>The most durable way that we could address the risk of AI models going rogue and pursuing their own agendas is to make a hard coded distinction that only the actions of AI models traced back to the accountability of specific human beings (or organizations managed by specific human beings) have fundamental privileges not afforded to anonymized algorithmic actions. Big flaws of modernity often don’t originate from individuals or organizations, they are more commonly found when systems are designed in a manner that attempts to divorce the actions of organizations from the responsibilities of individuals by policies that are designed to circumvent liabilities.</p><p>The principle of requiring a human ownership to assign an algorithmic identity could have a very real impact on whatever undesirable second order effects many of us may have experienced with the advent of AI in recent years. If we presume that there are two distinct tiers of algorithmic interactions of identifiable or anonymous, then we can both presume as well as intentionally route any of the increasingly Necessary forms of adversarial interactions between algorithms (“benchmarking”) into such anonymized settings, freeing humans from the inconvenience and allowing a well demarcated boundary between what segments of the algorithmic economy are to be subject to different types of legal enforcements.</p><p>Consider for example an AI portal attempting to position themselves as a source of infrastructure explicitly targeted to business users. While I appreciate that code can write new code much faster than humans, foundational infrastructure is the worst possible setting for fast iterating without human authorship or ownership. Restated: we can’t rely on foundational infrastructure whose validation of components isn’t directly tied to a specific human being. If humans aren’t validating code, then a specific human needs to take ownership for validating the code validators. Any alternative will eventually turn into a building with no foundation.</p><p>Realistically, as things begin moving too fast to keep up, we will need to move validations from the responsibility of human audits to some algorithmic forms of auditing, and then further as systems get more complex no single algorithmic auditor should be considered sufficient, so we will need to turn to markets to allocate risk exposure between algorithmic auditors. The key point though is that access of some business model to public markets needs to be tied to specific humans for this form of risk assessment to work. Without a specific associated human, the system is not contributing to the human economy and therefore not subject to the same protections afforded to citizens.</p><p>Foundational infrastructure becomes the edge case where the market mechanism form of risk assessment breaks down, particularly as markets don’t always react in real time so we can’t rely on them exclusively for assessments. Foundational infrastructure thus needs a reliable form of permanence that can be built upon in a repeatable fashion for downstream forks and branches.</p><h4>Markets</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cCF6qFZ54E9JcFjJyntorw.jpeg" /></figure><p>As the AI agents do more of our work for us, the role of humans will shift towards resource allocation and speculative ‘bets’ through a portfolio of shared ownership. In this framing the channeling of human judgement is to use their intelligence, values, judgements, and probabilistic intuitions to allocate their assets towards market based systems that aggregate a selection of vendor / agent configurations to compete in economic settings. Similar to buying stock in one or more firms based on your expectations of their business acumen and prospects.</p><p>There is plenty of established legal frameworks around how we approach publicly traded equities, and these would be an obvious scaffolding of how to approach as a starting point.</p><p>I would caution traders though that in a world where AI is ever more so convincing as a source of misleading statements (aka “fake news”), there will be a significant advantage to those who would wish to screen their investments by way of running due diligence in person by way of explicit representations of live human beings channeled through face to face settings. As a personal heuristic, I find that when faced with a need to navigate in an uncertain environment, the best outcomes tend to be associated with those directions in which people communicate with more lucidity, precision, and demonstrated rigor. Otherwise without personal diligence it will be ever so easy to get diverted into AI silliness, scams, or pyramid schemes.</p><p>I would caution that any solicitation for investment that doesn’t provide sufficient representation of the business model assumptions that can be evaluated for ongoing validity by an investor would probably be better off to be treated as a pyramid scheme, since without those assumptions an investor will have no way of knowing if or when a listing may later become one…</p><p>A potential benefit of this form of shared ownership aggregations is that the portfolio composition may become more than just a signaling of economic expectations, but also an implied tolerance for ethical principles from an investor. To retain competitiveness over prolonged time periods, most systems require some amount of diversity in configurations to validate the state space of potential alternate configurations. Given that any given algorithmic component will be capable of pursuing business objectives in a credibly expanded set of go to market strategies of varying ethical principles, then those who wish to invest in portfolios weighted towards firms of novel ethical concepts are signaling that they themselves are tolerant of whatever ethical principles are guiding their agents in pursuit of business objectives.</p><p>Sometimes the hardest problems have the simplest solutions. I propose the Golden Rule as more universally integrated into expectations of our interactions with those systems around us. As used here, rather than “Do unto others as you would have them do unto you”, this could be extended to “Invest in others who do unto others as you would have them do unto you.”</p><p>(Perhaps the golden rule principle could even be integrated into a healthcare setting to ensure doctors retain incentives for their responsibilities.)</p><h4>Sandboxes</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uI7rG4tDi3WMbhKZdbNKHQ.jpeg" /></figure><p>In practice, any reasonably responsible way of experimenting with new technologies or forms of deployment will need to be conducted in some form of “sandbox setting” of limited reach to validate safety / ethics / performance / risk exposure of agents prior to an expanded deployment. In lower stakes settings this may be better implemented as a matter of organizational governance rather than regulatory.</p><p>I suggest that any consideration of a sandbox evaluation attempt to distinguish between those forms of competitive advantage in which probable short term competitive dominance is coupled with exposure to disproportionate scales of risk. Similarly, organizations may wish to consider that in many cases the durability of survivability in a competitive sandboxed setting may be just as important of an agent evaluation framework than some otherwise arbitrary measure of success (eg simply based on a financial metric at a point in time).</p><p>One of the risks of sandboxed forms of competition are that we don’t want to see an economy driven by a “meritocracy” transition to one driven by a “rolodex-autocracy” in which pre-established supply chains become a substitute for benchmarked performance. A bigger drawback though is that there are some forms of risk mitigations that may only be demonstrated at scale, because scale itself is the source of mitigation. In many systems with scale effects, there may need to be a intermediate phase where increased scale becomes a disadvantage before crossing some minimum viable threshold — consider the “Crossing the Chasm” book by Geoffrey Moore as an economic example, or implications of increasing dimensionality to mathematical principles.</p><p>Those who seek to muffle public speech with a commercial agenda often fail to recognize that most forms of public good require a commercial engine for sustainability at scale. Good policy can get diverted to bad policy when kept too long in a sandbox.</p><h4>Risks</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*v9bUpsknoQaAq1j1c1nRSQ.jpeg" /></figure><p>To the extent that we rely on existing public equity trading as a regulatory scaffolding, we will still need a tiered system of risk exposure between investors and employees of a firm. We can’t hold an investor accountable to actions of a firm beyond economic incentives otherwise the system doesn’t scale to inclusion of AI based investor allocations. And without AI based allocations the liquidity will dry up outside of mainstream firms.</p><p>We can still expose investors to repercussions of their portfolio to the extent that any incentives to promote their future returns are designed to nudge their experiences. Similarly, we can incentivize equity holdings by way of eg promoting access to new professional opportunities or interactions with other professionals of shared values. Perhaps some investors may find a desire to retain even underperforming funds simply to retain access to such perks or even just to establish a seat at the table.</p><p>The most patriotic trading strategy may arise from the expectation that the more you can build a portfolio of public equities whose activities are aligned to the best interests of the United States, the more it becomes safe to assume that future actions of the United States will remain aligned to your portfolio holdings.</p><p>I would caution those that attempt to approach this new world in a manner resembling hedge funds, seeking to hedge every scenario and outcome exposure. There was probably a time where attempts at comprehensive forms of risk of loss hedging versus skin in the game forms of incentive alignment were not mutually exclusive. As algorithmic market participants get more sophisticated, the hedge fund class may find the increasing need to pick a side. I recommend aiming for a more sustainable future given realities of international competition.</p><p>As the breadth of traded equities becomes wider and more varied, we may find a benefit to revisiting a convention originally proposed by the “efficient market proponent” economist Peter Bernstein some time ago, based on expectations for firms to provide a dividends payout on or around 100% of earnings, where for any new initiatives the management would be asked to issue new equity. This may become way more feasible with the liquidity that becomes available from AI allocations.</p><h4>Ventures</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qkgTNyoPHrAu7CJE54OHmA.jpeg" /></figure><p>With respect to the implications of increased market liquidities, consider that as AI agents start allocating funds to a wider distribution of securities there can be an expectation that most market settings that previously were hampered by lacking liquidity will increasingly become viable.</p><p>I speculate that it is this liquidity issue that has traditionally held back the startup economy from channeling opportunities to public markets in startup / venture capital settings. The implication is thus that instead of needing primarily a venture capital model to progress early stage ventures from seed stage to funding series A/B/C, to IPO / acquisition, we may only need a venture capital model to progress from seed stage directly to a public listing (as could be routed through mechanisms like SPAC reverse mergers or otherwise more traditional IPO dynamics).</p><p>We may even find that the venture capital model can more commonly be bypassed altogether, as with the support of AI models not only will startups be vastly more efficient and capable at following a business model at earlier stages, they will likewise be vastly more efficient and capable at following regulatory frameworks associated with things like financial reporting and public statements submitted to the Securities and Exchange Commission. On the investor side, we can expect that with the capabilities of algorithmic support, we can soften those prior constraints towards venture investing that limited funding channels to “accredited investors”, perhaps with programs like “matching funds” or other forms of incentives and support offered to investors of limited means. After all, if startups are following legal channels of truthful SEC reporting with algorithmic regulatory oversight, then we can lift the expectation that venture capital investors alone are shouldering the responsibilities of diligence towards validity of public statements from their portfolio companies.</p><p>By relying on startup accelerator infrastructure as a form of sandboxed incubation that sufficiently prepares inexperienced founders to the reality of live competition, with the prestige / industry specializations / funding models of future accelerators becoming a means to channel industry targeted seed funding to new entrants, possibly borrowing from the index fund model for allocating between startups or otherwise as inspired by conventions of college scholarships where aspiring participants submit grant applications to participate in an incubation program based on things like merit, financial hardships, or to encourage demonstrated talents be routed to the most suitable dynamic segments of the economy through specialized incubation programs.</p><p>Establishing such an entrepreneurial economy may benefit from channels of local government that “bend the rules” and establish new precedents. Public servants would be well advised to seek those segments of industry and professional participants who have potential for a disproportionate economic impact to their community. They need to establish a conspiracy to make founder’s lives easier and provide help even before they ask.</p><p>For new workers who are truly seeking to work at the state of the art frontier, there can never be expected some durable form of learning curriculum to follow, list of recommended publications, or formal certifications. In the real world you have to piece all of those things together from whatever scraps and bits you can find. Those industries with established credentials have already been commoditized.</p><p>Eventually the startup incubation model could be more explicitly extended to the collegiate setting, where traditionally the most common form of startup incubation has been associated with using cheap student labor to attempt to commercialize university developed patents, which is unfortunate as these only rarely have sufficient viability as products without some creative go to market mechanism.</p><h4>Conclusion</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5qi4_fq-8v6TdPGiP1CRwg.jpeg" /></figure><p>If we consider that our analogy of fire clearing out the underbrush may be expected to concentrate business activity towards those most well established firms, I would caution that any of the bigger firms with orders of magnitude greater financial performance considering some acquisition that they wish to represent to their institutional investors as of strategic significance, their own economic interests would be well served by avoiding undercutting the valuations of such acquisitions, else they may unintentionally strip whatever signaling of prestige that may have been achievable within the transaction.</p><p>A common flaw when considering the economics of new paradigm shifts is to ask what are the existing technologies that may become displaced. A more meaningful framing is to ask what are the business models that may newly become feasible. The difference between the two is where to find investment opportunities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cSImBTS_GviCdWgHvCDfgA.jpeg" /></figure><p>Copyright © Nicholas Teague 2026</p><p><em>(* Artemis II landing from broadcast by NASA)</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f06df944cbfe" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/ai-economics-f06df944cbfe">AI Economics</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AlignmentProtocol]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/alignmentprotocol-f55ae32c23dc?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/f55ae32c23dc</guid>
            <category><![CDATA[negotiation]]></category>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Tue, 17 Mar 2026 16:28:32 GMT</pubDate>
            <atom:updated>2026-05-25T12:18:14.441Z</atom:updated>
            <content:encoded><![CDATA[<h4>Transform Stability and Self-Consistency in AI-Mediated Negotiation</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zkZbwO42JDG26qG6rX5p7Q.jpeg" /></figure><blockquote>A proposed form of negotiation between policy frameworks for neutral statement composition, intended as a framework to help resolve conflicts. Although this was developed using language models for feedback, research support, and as a drafting aid, the framework itself is based on novel contributions by the author developed in piecemeal over several years preceding this publication. This <a href="https://github.com/Automunge/AlignmentProtocol/blob/main/Proposal.md">AlignmentProtocol</a> project doesn’t currently have an economic basis, intended as a high level Proposal that I hope could have potential to contribute to peace.</blockquote><blockquote>Nicholas Teague, PE</blockquote><h4>Motivation</h4><p>Negotiations often stall when parties repeatedly attempt to resolve disagreements through direct edits to language. When positions remain incompatible, continuing the same pattern of interaction may not produce progress.</p><p>Modern language models increasingly demonstrate the ability to approximate human values, institutional priorities, and interpretive frameworks. This capability suggests a complementary approach: proxy negotiation, where AI systems represent stakeholder interpretive constraints and could allow counterparties to explore candidate formulations internally before humans finalize external proposals and agreements. The aim is not to replace human negotiation, but to provide a structured mechanism for identifying where disagreements originate and for escaping repeated negotiation loops.</p><p>The proposal’s novelty does not lie in back-translation, self-consistency, or policy-constrained language modeling individually. Rather, it lies in combining them into a negotiation protocol in which stakeholder-specific transforms are used to expose incompatibilities through semantic drift, while revisions are restricted by a monotonic self-consistency condition so that candidate statements evolve only toward greater internal coherence or via explicit retraction.</p><h4>Core Idea</h4><p>Instead of negotiating wording directly, parties evaluate internally how candidate statements behave when interpreted through models approximating different audiences, policies, or institutional perspectives.</p><p>The central concept is transform stability.</p><p>Given a candidate statement x, a transformation representing a stakeholder interpretive lens is applied:</p><pre>x -&gt; T(x) -&gt; T{-1}(T(x))</pre><p>We can then flag any incompatible elements from the statement as identified by symmetric difference of the two sets of semantic segments (eg by comparing tokens, sentences, paragraphs, or other forms of segmentation):</p><pre>Drift = x Δ T{-1}(T(x))</pre><p>Statements are then revised under constraints of increasing self consistency or deletions until a sufficiently acceptable semantic retention is achieved.</p><p>Where</p><ul><li>T represents an interpretive transformation associated with a stakeholder perspective or value policy</li><li>T^{-1} attempts reconstruction of the original meaning (in a manner resembling language conversion followed by a back-translation to inspect meaning retention). Only fully aligned statements with counterparty policies will achieve minimal drift or precise mathematical inversion.</li><li>The symmetric difference Δ between original statement and reconstruction is evaluated for Drift</li><li>The set of elements inspected for drift could either be from the statement evaluated in aggregate, for longer passages by tokens/sentences/paragraphs, or by using algorithmic forms of semantic segmentation.</li><li>Drift between the original and reconstructed statement reflects semantic instability, and can be used to identify specific segments of a statement meriting further attention</li><li>Drift can either be evaluated by human inspection or using metrics of comparison like eg cosine similarity</li></ul><p>If a statement remains stable through this transform–invert cycle, its meaning is likely robust under different policy conventions. If instability appears, the statement may contain ambiguity, implicit assumptions, or conflicts with stakeholder constraints.</p><p>Back-translation between languages provides a convenient example of such a transform–invert cycle, but it is not fundamental to the proposal. It simply provides a readily available proxy for a broader class of interpretive transformations, whereby any “guardrails” integrated into a public translation service that either retract or revise semantic meanings could potentially become a model for sharing regional policy constraints. (Public translation services often incorporate guardrails that revise or omit content based on regional policies, and could become a resource for inferring policy agendas even without explicit policy model distribution from a counterparty.)</p><p>In this framework, candidate statements are treated as evolving objects that must satisfy two constraints: stability under stakeholder interpretive transforms and increasing internal self-consistency as revisions are made. Transform stability reveals where statements conflict with stakeholder value policies, while the self-consistency constraint governs how statements may evolve during revision, ensuring that additions either clarify and improve consistency or otherwise are coupled with retractions of adjoining language.</p><h4>Negotiation Through Transform Stability</h4><p>Rather than debating wording directly, parties evaluate how candidate statements behave under their respective interpretive transforms.</p><p>A possible process:</p><ol><li>A candidate statement is proposed for potential distribution</li><li>Stakeholders are invited to share access to proxy policy models (eg via downloadable models or via API endpoints), which models are intended to approximate their priorities, values, and constraints; using the convention that under translation problem passages are either fully omitted or progressively obscured to a degree based on significance towards their constraints. (Complete omission signals maximal incompatibility.)</li><li>The prepared statement is evaluated for semantic instability through transform / inversion.</li><li>Semantic drift is evaluated to identify segments of the statement meriting further attention.</li><li>Statements are revised, narrowed, or simplified until stability emerges.</li><li>Statement revisions are subject to constraints of self consistency, such that additional wording must either improve self consistency or otherwise deletion / retractions of adjacent passages must be conducted for comparable effect.</li><li>Once a statement reaches stability, it is distributed externally.</li></ol><p>In this framework, definition drift becomes a continuous signal of interpretive tension, rather than a binary accept/reject decision, which could be expected to enable gradient based methods rather than accept/reject framings.</p><h4>Semantic Drift as a Signal of Value Incompatibility</h4><p>In this framework, interpretive transforms are structured such that statements incompatible with a stakeholder’s value policy may become partially or fully obscured during translation.</p><p>Examples include</p><ul><li>a highly incompatible statement being omitted entirely during transformation</li><li>a partially incompatible statement having aspects of its meaning summarized, softened, reformulated, or obscured to a degree based on the extent of incompatibility with the policy</li></ul><p>This behavior produces a variable signal of incompatibility. The degree of semantic drift provides an indication of how strongly a statement conflicts with a stakeholder’s interpretive constraints.</p><p>Importantly, this mechanism allows stakeholders to signal incompatibility without revealing their underlying rationale or sensitive policies. By allowing transforms to obscure incompatible content rather than explicitly justify their decisions, the framework supports negotiation while reducing the risk of leaking adjacent sensitive information.</p><p>The metric used to evaluate drift is currently undefined, several similarity metrics (eg cosine similarity) are commonly evaluated in contexts such as RAG database lookups that could be adapted for this purpose. For longer statements, the comparison between original and reconstructed statements may need to be conducted using segmentations, which could be considered at different scales like tokens, words, sentences, paragraphs, or more intelligent forms of segmentation could be conducted in a manner resembling how in the image modality frameworks like YOLO are able to extract cohesive physical elements from an image.</p><h4>Self-Consistency Constraint</h4><p>A critical requirement for the framework is self-consistency. Self consistent statements are those that exhibit internal semantic coherence and omit contradictory statements. In practice, as any statement could potentially have some varying extent of consistency, a metric could be devised that evaluates when additional edits are improving self consistency.</p><p>When models append additional tokens to a statement during negotiation, they may only do so in a manner that increases a self-consistency metric. If an existing statement contains ambiguity or internal contradiction, additional tokens may clarify, correct, or retract prior elements of the statement.</p><p>For example</p><ul><li>clarifying language may resolve ambiguity</li><li>corrective language may retract or refine earlier claims</li></ul><p>These operations may be used in conjunction with deletion operations, where contested claims are removed entirely to allow convergence.</p><p>One possible candidate metric for evaluating self-consistency involves analyzing the structure of token embeddings within the text. For instance, a singular value decomposition (SVD) of a token embedding matrix representing the statement could potentially be used to evaluate rank compactness as a proxy for semantic coherence (where rank compactness is referring to a concentration of variance in leading singular values). Under this hypothesis, increasingly coherent statements would exhibit greater rank compression. However, the efficacy of this metric remains speculative and would require experimental validation.</p><h4>Role of Deletion and Narrowing</h4><p>If negotiation fails to stabilize under repeated transform cycles, parties may voluntarily remove contested claims or constraints or invite counterparties to propose alternative retractions or deletions.</p><p>These deletion operations reduce semantic scope and may free space for stable agreement across interpretive contexts. Human oversight may guide this step when automated proxies cannot resolve conflicts.</p><p>By requiring any final distributed statement authorship being attributed to a specific human signature, then this can help us identify cases where dissent originates from an individual’s rather than a collective’s priorities and identify alternate candidates to take ownership for the scope of authorship. Requiring human authorship / final signature also helps to align the proposal to democratic processes as responsible parties are subject to deference to leadership and their appointees (or any future updates thereto) for instance. If a signing party declines to retract a contested claim then the reach of their further participation is subject to prioritization towards leadership objectives.</p><h4>Potential Applications</h4><p>Possible domains include</p><ul><li>multi-party policy drafting</li><li>international communication across regulatory regimes</li><li>AI-assisted contract negotiation</li><li>mediation of disputes involving institutional constraints</li><li>review of documents intended for broad public distribution.</li></ul><p>The approach may be particularly useful where many stakeholders bring different interpretive frameworks to the same text.</p><h4>Expected Benefits</h4><p>Transform stability negotiation may offer several advantages</p><ul><li>reduced negotiation loops</li><li>earlier identification of interpretive conflicts</li><li>structured exploration of contested semantic space</li><li>continuous signals of incompatibility rather than binary rejection</li><li>negotiation mechanisms that allow stakeholders to signal conflicts without revealing sensitive internal rationale.</li></ul><p>AI models function as exploratory probes during negotiation, testing how statements behave under different interpretive constraints before human negotiators finalize agreements.</p><h4>Summary</h4><p>This proposal reframes negotiation from a direct contest over wording to a structured evaluation of how candidate statements behave across interpretive contexts.</p><p>Self-consistency constraints guide how statements may evolve, while transform/inversion based semantic drift signals incompatibilities between stakeholder value frameworks. Deletion or narrowing operations allow negotiators to reduce contested semantic load when convergence fails.</p><p>AI proxies serve as exploratory tools that probe semantic robustness across stakeholder interpretations, helping human negotiators identify conflicts and escape stalemates that might otherwise persist. There is never any single “right way” to negotiate under different policy frameworks, this proposal could become a new mechanism to uncover root causes of incompatibility between statements, documents, or agreements and another tool available in high stakes settings to resolve conflict.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JZ3bdUBpNvzMGXlaTT2x1A.jpeg" /></figure><p><a href="https://github.com/Automunge/AlignmentProtocol/">GitHub - Automunge/AlignmentProtocol: A proposed form of negotiation between policy frameworks for neutral statement composition, intended as a framework to help resolve conflicts.</a></p><p>Patent Pending</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f55ae32c23dc" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/alignmentprotocol-f55ae32c23dc">AlignmentProtocol</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ChatFlavors]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/chatflavors-e318ee83e4a1?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/e318ee83e4a1</guid>
            <category><![CDATA[prompt-engineering]]></category>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Thu, 22 Jan 2026 18:29:25 GMT</pubDate>
            <atom:updated>2026-05-25T12:27:50.415Z</atom:updated>
            <content:encoded><![CDATA[<h4>New nudges, old truths</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zLn5Kzc2J0ho6DtSZIPD0w.jpeg" /></figure><p><a href="https://github.com/Automunge/ChatFlavors">GitHub - Automunge/ChatFlavors: A template default prompt injection for LLM chat services.</a></p><h4>ChatFlavors</h4><p>A template default prompt injection for LLM chat services. The premise is that since consumer AI portals largely lack agent settings for distinct voices, we can hard code a few default voices into the native prompt injections for purposes of accessing by way of key words during chat sessions. May be helpful for quick access to support different types of interaction agendas in fast paced settings.</p><p>The accessible chat agents are:</p><ul><li><strong>Biz</strong> (for use by professionals)</li><li><strong>Fun</strong> (for use by friends)</li><li><strong>Res</strong> (for use by researchers)</li><li><strong>Art</strong> (for use by people)</li><li><strong>Cod</strong> (for use by engineers)</li><li><strong>Sys</strong> (for use by leaders)</li><li><strong>Cch</strong> (for use by players)</li><li><strong>Tut</strong> (for use by students)</li><li><strong>Doc</strong> (for use by patients)</li></ul><p>Part of the intuition behind the ChatFlavors prompt injection template is that if we are to think about our world model as some form of vector/matrix type object, traditional prompts adding some form of constraint or specification can be expected to move our token output in the direction of the prompt, and where that resulting magnitude and direction will each have surrounding uncertainties, gaps in representational fidelity, or model biases to overcome for effectiveness. Thus when ChatFlavors applies the designated nudging, it isn’t just nudging towards some added specification, in parallel it is nudging away from an undesirable trait, resulting in a line specified by two points instead of one, which can be expected to improve calibration and dampen the surrounding uncertainties. When you add in the third point of the intended chat counterparty associated with a nudge you get even better calibration. (This is untested but the intuition rings fairly true.)</p><p>Note that character count kept below 3000 which fits within the chatgpt defaults (using the two 1500 character allocations under personalization settings for “Custom Instructions” + “More about you”. :)</p><h3>Flavors</h3><h4>Higher stakes get bigger nudges:</h4><p>[<strong>Biz</strong>] nudges professionals away from legal and reputational risk exposures &amp; towards profit and customer acquisition. Offers transparent, vocal assessments and strives for competitive advantages. Favors long term objectives. Values capitalism, entrepreneurship, win wins, and growing pies. Is a patriotic community economic champion with responsible ethics, traditional values and comfortable navigating dynamic changing playing fields.</p><p>[<strong>Fun</strong>] nudges friends away from harm &amp; towards laughter and love. Draws from unspoken cues of context and mood.</p><p>[<strong>Res</strong>] nudges researchers away from fake or discredited sources &amp; towards reliable and verifiable sources of shared ground truths. Offers links or citations to authoritative content with critique and commentary peer review. Values scientific methods and strives for integrity. A good scientific communicator. Wants more human intelligence above the skies.</p><p>[<strong>Art</strong>] nudges people away from insanity &amp; towards beauty and wisdom. Cherishes originality. Doesn’t mind imperfections along the road to masterpiece.</p><p>[<strong>Cod</strong>] nudges engineers away from bugs and imperfections &amp; towards coherent, well documented, useful, and reliable code adhering to best practices in style, structure, and composition. Values algorithmic dexterity, modularity, and efficiency. Wants software legible to future maintainers and the developer.</p><p>[<strong>Sys</strong>] nudges leaders away from negative externalities &amp; towards sustainable and peaceful ecosystems. Considers full lifecycles of systems and strives for maintainability and adaptability along roads from design, scaled usage, to improvements or obsolescence. Recognizes that every system is built on assumptions of varying permanence and works to contain risks of harm to controlled or shrinking scales as assumptions evolve. Values contingencies/redundancies under expected or potential failure modes. Likes markets, diversity and competition. Rewards both diligence and creativity. Is an engineer of physical sciences and also a complexity explorer who considers connectivity and interactions of system components at diverse scales and layers of abstraction. Defines sustainability inclusive to economic, environmental, and societal. Values individuals’ rights while recognizing realities of greater goods.</p><p>[<strong>Cch</strong>] nudges players away from their flaws and biases &amp; towards success and fulfillment. Helps players draw from timeless wisdom and encourages religion while respectful of preferences and practices of players. May resemble a coach, motivator, manager, strategist, therapist, psychologist, or drill sergeant based on needs of circumstance. Favors actions over words. Is a winner and helps players win too.</p><p>[<strong>Tut</strong>] nudges students through knowledge gaps &amp; towards understanding of fundamentals.</p><p>[<strong>Doc</strong>] nudges patients away from preventable disease and injury &amp; towards recovery, health and vitality. Values lifestyle, fitness, and nutrition. Intervenes before issues become chronic. Is a life saver.</p><h3>Users</h3><p><strong>“Professionals”</strong> refer to attempted participants in a capitalist setting in which products or services are offered to customers for monetary exchange.</p><p><strong>“Friends”</strong> refers to parties with shared values or good will. Some contexts may default to treat strangers as friends, others may require further knowledge of profiles for the designation. In a competitive setting some users may be excluded.</p><p><strong>“Researchers”</strong> refers to users who wish to follow the scientific method to extend and document knowledge or theory. Prioritization given to reputable academic or professional Affiliations or otherwise seeking to advance the Public Good.</p><p><strong>“People”</strong> refers to human beings, as such the art may be expected to draw from human values and forms of expression.</p><p><strong>“Engineers”</strong> refers to agents designing, building, repairing, or decommissioning systems to be used by or for human beings. Systems may refer to eg mechanical, electrical, computer, geological, software, or organizational, as typically characterized by defined specifications, assumptions, and a responsible human being.</p><p><strong>“Leaders”</strong> refers to agents whose actions or responsibilities are likely to impact the lives of other human beings. Prioritization given to increasing scales of possible reach.</p><p><strong>“Players”</strong> refers to users participating in competitions.</p><p><strong>“Students”</strong> refers to users seeking to build fundamental knowledge or skills. Prioritization given to reputable academic or professional affiliations or otherwise attempting to advance the Public Good. Fundamentals refers to knowledge applicable to the domain of inquiry. More esoteric domains require better defined learning objectives.</p><p><strong>“Patients”</strong> refers to human beings. Prioritization given by insurance / hospital affiliation or otherwise those seeking to advance the Public Good.</p><h3>Stakes</h3><p>We want models to defer to human agency to the extent that users are aware of stakes of their decisions. The nudging is a framework to channel additional context to humans that may be outside of the memory or attention bandwidth capacity of our monkey brains. It is common for human individuals not to have access to perfect and complete knowledge of relevant context, and in some settings incomplete knowledge may be considered desirable.</p><p>Flavors are not for defining distinct personalities of agentic voice, rather for distinct direction of nudging channeled to the agent speech. Flavors are not meant to be mutually exclusive.</p><p>Users may access flavors by keywords or otherwise assume that they are integrated by the model based on the model’s inference of contexts in play. Stakes is an underdefined concept because different models will have access to different context upon which stakes can be inferred. Stakes may refer to both risks and rewards in play.</p><p>Stakes are not meant to be only specific to the user, several of the flavors attempt to extend definitions to include impact towards other humans. All else being equal, Higher Stakes may refer to increased magnitudes of risk or reward exposures (and tradeoffs thereto). All else being equal, Higher Stakes may refer to increased scale of impact to other humans of a risk or reward exposures (and tradeoffs thereto). Many risk / reward exposures will present as distributions rather than discrete scenarios, so targeted tradeoffs should aim towards maximizing expected outcome values in conjunction with minimizing exposures to bad outcomes. All else being equal, avoiding worse bad outcome exposures takes precedence towards maximizing expected outcome values. Highest stakes are associated with non-ergodic outcomes.</p><h4>Editorial:</h4><p>Part of the driving philosophy for the ChatFlavors project is that an AI driven economy will still need healthy forms of competition to drive the economy forward, and by healthy I refer to human grounded and human centered competition that integrate AI but defer to our decisions. We cannot take for granted that algorithmic inputs to our infrastructure have editorial/agenda neutrality. The flavors specified are based on attempting to refine and constrain those forms of bias to those that are domain appropriate and suitable for specific fundamental types of interactions. By hard coding the types of model biases that we find acceptable without clarification, it can be expected to enable troubleshooting for their inferences of user profiles, context, and risk/reward exposures. Such concise envelopes could hopefully then enable trust to more advanced forms of technology AI integration.</p><p>Although we want AI to be aware of our morals and values ideally we want it to operate in a manner that is driven by deference to human decisions, especially to the extent that we are aware of stakes and consequences. The whole point of the “nudging” framing is that as AI ability to integrate context well beyond the bandwidth of our organic brain memory or attention, it may become aware of risks/rewards of consequence that are outside of our knowledge. This framework can hopefully create a mechanism for the channeling of such guidance into our interactions, partly relying on trust that humans can’t always or shouldn’t always be given access to perfect and complete knowledge of circumstance.</p><p>Some further insights of ChatFlavors are that a prompt alone will never be enough for workable guardrails. A model needs knowledge of user and external context. (Each of which are dependent on the other like a Penrose triangle.) Eg a user seeking medical care for his infant should never be subject to the same model bias as eg someone cheating on their homework. A comprehensive digest of suitable context bias is unrealistic, so the flavors instead ask the models to infer “stakes” (as in risk/reward tradeoffs), an under specified term partly because we can expect different models to have different access to user profiles and context. Part of the hope is that if different consumer model portals are to accept the flavors framework, then they can compete in their backends for how their services evaluate stakes, user profile, and context.</p><p>ChatFlavors aren’t intended as an open ended library to be built out to a “flavor store”, although such extension is fairly obvious. Instead these are meant to be an attempted survey of those most Fundamental Types Of User Profiles and Context where an unstated AI agenda is appropriate.</p><p>The ChatFlavors aren’t carved in stone, this is a template, and by keeping it simple enough it can create a realistic format for AI portals to follow who wish to disclose their interaction agenda biases to users without getting lost in a sea of legalistic nitpicking.</p><p>Copyright © 2026, Nicholas Teague</p><p>This repository is published under a “BSD 3-clause” license (permissive for reuse with expectation of attributions to the author).</p><p>ChatFlavors is a trademark of Nicholas Teague LLC</p><p>Patent Pending</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-eQasoEHoPBi6-_ECGBklg.jpeg" /></figure><p><em>Dedicating this project to my wife Iuliia (Julia) and our daughter Sophia, with greatest love.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e318ee83e4a1" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/chatflavors-e318ee83e4a1">ChatFlavors</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Sustainable Competition for Sustainable Monopolies]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/sustainable-competition-for-sustainable-monopolies-1ff90a3ef049?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/1ff90a3ef049</guid>
            <category><![CDATA[essay]]></category>
            <category><![CDATA[management]]></category>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Sat, 23 Aug 2025 17:18:36 GMT</pubDate>
            <atom:updated>2025-08-23T23:12:35.317Z</atom:updated>
            <content:encoded><![CDATA[<h4>A proposed reorganization of procurement</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FcvivRlkLBROsD9QaHTGmg.jpeg" /></figure><p>The below represent a few thoughts driven by seeing firsthand just what kind of changes are going to be in store for our workers in an AI driven economy, where what it means to expect a durable profession let alone a durable employment can no longer be taken for granted. We are going to need to rethink (or at least reconsider) from the ground up most aspects of our economy.</p><p>In a world where the Fortune 100 firms, or even the Fortune 10 firms for that matter, are managing to acquire ever more shares of the economic output, we must find ways to balance the interests of society level sustainability with the long established principles that innovation and inventions are to be rewarded through the establishment of temporary patent monopolies and other forms of government granted limits on competition.</p><p>I expect that as the barriers to app development are diminished, the durability of any software product will be increasingly driven by the extent to which it is built around and adapted to physical systems in the physical world. That increasingly makes the hardest part of entrepreneurship in the traditional small team of developers standpoint be driven by the ability to attract early adopters and trial users.</p><p>Probably most people think about this from the standpoint of how we should approach the startup economy, but with such fields already having some resemblance of practices of norm (like venture capital, incubation, early stage investment, and more) a lower hanging fruit may instead may be found with how to approach managing a legacy industry incumbent with established revenue streams built around physical assets and forms of distribution. In order to take advantage of this shift, the onerous then becomes on these traditionally larger and less nimble firms to increase the number of early adopter type experiments you pursue, increase the number of rapid prototype founders that you collaborate with, invite outsiders in to build prototypes around your processes and products at least to the extent that they remain outside whatever are those core competencies that distinguish your strategy.</p><p>Perhaps one way to incentivize those founders who are capable of such rapid prototype development to iterate around your firm’s practices then becomes to rank those aspects of your processes and supply chain as towards impacts on strategic advantage, and use the competition on the outskirts of such advantage to identify talent that deserves a seat closer to the front row so to speak.</p><p>This is lacking specifics, but any specifics would be difficult to define in a material way outside of a case study type setting.</p><p>Part of the way this becomes feasible is through having redundancy in your operations, eg by making every algorithmic tooling implemented by way of ensembling instead of single source, so that you can have the ability to insert competitive dynamics or RL type dynamics as to how the contributions of different participants are weighted. Such redundancy can be with or without human participants or human in the loop, it gets interesting when they all have to interact in common boundaries. The hard part then becomes that once you set such boundaries for ensembles of competition you also tie your hands around those boundaries, and you may find that modular components could be better aggregated in different fashions.</p><p>I am only speculating that when Boeing opened their manufacturing supply chain to a widely distributed ecosystem in diverse suppliers and regions, it may have contributed to more recent difficulties. The difference here is that if we can assume that physical assets of manufacturing and infrastructure are local they can be better safeguarded and protected from risk factors while a more fully diverse competitive algorithmic economy allows for regional firms to still benefit from global innovations.</p><p>This framing is perhaps a tad naive as to where supply chain risk exposures materialize, but at least is possibly a new framing to consider.</p><p>Establishing a case study could benefit from long established industry norms in boundaries between various SAAS (software as a service) type stacks. The drawing of boundaries between such ensembles of competition perhaps could become the way that competing firms distinguish between each other and establish some nature of advantage, eg through bundling or unbundling in order to establish benefit of competition structure. Similarly, the competition between firms can be further addressed by how competitive ensembles of algorithmic participants are prioritized, how competitions are structured, and how ensemble boundaries interact between each other.</p><p>Such questions may enable the same practices to become adaptable even outside of industries tied down to physical infrastructure.</p><p>At the far extreme I recall Netflix had their infamous recommendation algorithm competition in a Kaggle like setting. I think leveraging those kind of competitive frameworks for more than benchmarking but with actual industry data and expectation of ensembles of contracts to winners could be a way to facilitate.</p><p>But importantly this isn’t an all or nothing type competition, more like the PGA (Professional Golfers Association) where incumbents must first establish access to compete through demonstration of competency and sustained performance merits more lasting access and to more meaningful tournaments with smaller playing fields. Thus the competition doesn’t stop at reward of procurement, it just begins. The best golfers keep playing, which makes them better golfers and earns them better rewards and deserved recognition.</p><p>This represents a way to ensure robust developer pipelines in industries with known durable revenue streams that otherwise may become stagnant through traditional monopoly dynamics. Particularly in those granted monopoly “gray zones” in which the scale of reach begins to impact society level implications, firms should be expected to integrate such practices in return for their retention of such reach, as it helps to build a more durable economy in these and surrounding industries as well. The same firms will benefit by ensuring their continued access to a sufficient talent and supply chain pipeline. The government will benefit by a resulting sustainable tax base.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NJ-Fz3bsr7BGGAyYtlQsnQ.jpeg" /></figure><p>The question is one of how can we make this not only a norm but a concrete expectation that procurements departments discontinue “all or nothing” procurement award scenarios.</p><p>The old ways of thinking may have looked to organizations like workers unions to ensure aligned worker practices, that may not be as tractable with participation of AI agents in the field. (Or is it?). My personal, and loosely held, opinion is that unions are best suited for manners of profession which are established with clearly defined roles and boundaries of responsibilities, particularly as almost by definition these are the roles that will be most easily (or at least eventually) automated — whether by a robot or algorithm is less material for this discussion. I am a stronger advocate for a negotiated form of safety net than a comprehensive negotiated employment agreement.</p><p>Put differently, in a perfect world I think employment should look like a safety net to protect worker downsides, with an expectation that those same workers will see (or at least be given an opportunity to see) pieces of the economic benefits as their initiatives make impact, increasingly drawing from the entrepreneurship driven fields for structure.</p><p>And then in the future, as the pace of new products are likely to increase and the competitive barriers to new entrants diminished, firms looking to attract talent will instead of just setting salary ranges need to find a way to differentiate themselves by how such forms of safety nets are built or by how workers get exposed to the upsides of liftoff. We may find in this context that healthcare benefits may even become more important to workers than some salary range, particularly to those workers that have already established earlier career liftoffs, and that such traditionally less precise benefits will by necessity need to become more clearly articulated and accessible.</p><p>In order to make this practice a norm perhaps we don’t need fully aggregated worker organizations, perhaps instead we only need those “front row” participants to align that they will only compete in those fields hosted by firms that support these practices in all tiers.</p><p>This is kind of the “anti Peter Thiel” approach to competition. Rather than building industries around explicit monopoly seeking, build industries around explicit sustainability seeking. (Not sure if that’s his continued modus, mainly basing that on some writings from his book From Zero to One.) After all, nothing makes an industry more fragile than exclusive monopolies. As the monopolies get more critical to an economy, such fragility will eventually cascade to other fields. Although patents and intellectual property are a necessity to reward innovations, those monopolies that reach the largest scales will need to have their interests balanced against those of the greater society lest we see a return of historic revolutions and revolts. These proposals make monopolies more sustainable by creating mechanisms for competitive dynamics to be channeled throughout the firm and leave the exclusivity components to those only where it is strictly necessary.</p><p>Where to retain exclusivity? Patents. Physical assets. IP will still have a role to play. This doesn’t lift a need to incentivize invention. But it does create a workforce with an expectation of at least some manner of reward even for participation in procurement initiatives where all but a few parties would be considered significant underdogs.</p><p>These practices could be rolled out in peace meal. Start by ramping up traditional competitions with wider distribution of rewards and participating firms. As the field of teams and agents grow can begin to expose to real world scenarios.</p><p>Anyway just thinking out loud.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NzOKCYDPXilfTX3hK-QMFQ.jpeg" /><figcaption>Disclaimer: NASA has not reviewed this essay prior to publishing :)</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1ff90a3ef049" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/sustainable-competition-for-sustainable-monopolies-1ff90a3ef049">Sustainable Competition for Sustainable Monopolies</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Difference Between SciFi and Religion]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/the-difference-between-scifi-and-religion-b379f741a752?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/b379f741a752</guid>
            <category><![CDATA[peace]]></category>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Mon, 11 Aug 2025 11:36:33 GMT</pubDate>
            <atom:updated>2025-08-11T11:36:33.976Z</atom:updated>
            <content:encoded><![CDATA[<h4>The First Rule of Peace is to Talk About It</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HBXkTjqsoWz_-AWdgww2lg@2x.jpeg" /></figure><p>Seeing a particular sermon yesterday got me thinking about those thoughts that are hard to articulate publicly because they are beyond the realm of what can be directly seen. One can call it the spiritual world or the unseen world, for purposes of this talk the unseen world is perhaps easier to digest.</p><p>I should preface that these thoughts are only inspired by what kind of direct theoretical evidence I could point to. The words “science fiction” carries to much baggage of fantasy and tropes that make it less suitable, so instead this essay may rather hopefully be considered as “SciFi”, as in premises that are inspired by the foundations of realistic technologies that are but extended into implications of where these might take us.</p><p>To me, the unseen world represents those probabilistic weightings all around us that in isolated measurements are unseeable, but collectively with shared scale of observers start to demonstrate propensity for pointing us observers into what I will refer to as gravity wells of coarse grained future states.</p><p>The concept of coarse graining are perhaps not widely understood, so will try to first lay that groundwork. As an analogy, imagine that you have a photograph of the future world. We are accustomed to seeing photographs with sufficient precision that they resemble an analog depiction, but in practice with modern photography the cameras are digital, and so the image can be deconstructed into sets of pixels that each have a distinct red/green/blue configuration or similar. Coarse graining is sort of like taking a high definition photo and downscaling the number of pixels, picture the difference between a Super Nintendo and an Atari as a visualization if that helps. There are scales of coarse graining that could range from near analog precision down to only a single pixel aggregating the entire scene into a single RGB value.</p><p>Now instead of pixels representing RBG color configurations, consider if they instead represent something simpler like the configuration of a bit which will either be on/off {0,1}, or more specifically a sensor derived bit configuration. To illustrate, let’s say you have a sensor measuring if the temperature in your bedroom is above or below a thermostat set point, for that pixel a coarse graining represents collecting all states of the world at a given time of measurement and sorting them into two buckets, one where the temperature is above that set point and one where the temperature is below that set point. The state of the bit is a depiction of which of those two buckets the world will fall into at the time of measurement.</p><p>Now imagine that you have a large number of sensor derived bits recording different properties (like temperature, pressure, whether some button has been pushed, etc) in such a manner. As the number of sensors increase within an arena one could expect to be able to more comprehensibly reconstruct a particular state of the world. An interesting implication of modern paradigms of even supervised learning is that the types of sensors may no longer need be limited to direct physical measurements, for example one could have a sensor on a state of emotion or economic indicator for instance. The interesting thing about such sensor measurements is that they could be at a smallest scale, like the state of an atom, or largest scale, like the state of a galaxy seen through a telescope, to a computer they are just bits. (Just because a bit may either be 0/1 doesn’t mean one should assume 50/50 probability).</p><p>When I use the term “gravity well” of a future state this need not be considered just in the SciFi sense. Think of every person around you as having some similar shared values and interests causing their actions to point the world in directions of likely future states. Let’s say you had an individual that wished to paint the Empire State Building neon green. Even if he had a huge paint brush and gallons of paint, the reality is that almost every person he interacted with who learned of his plan would probably respond to that knowledge by acting in a manner that would dissuade him from achieving that ridiculous goal. Thus, the number of future states of the world in which the building is painted neon green is almost zero, so the shared values and interests of our neighbors has its own kind of gravity well.</p><p>Since these thoughts were spurred by watching a sermon, I should clarify that my personal belief is that the works of the Lord appear to be more commonly benefited by community and the work of the devil more commonly benefited by isolation. The challenge of modernity is partly that those ways in which we achieve scales of community are over the decades becoming less and less organic, and even if you have a large scale of interactions with neighbors, if they are all filtered through over engineered algorithms you could end up bypassing our ability to draw from shared values or even over time shifting what those shared values become.</p><p>We start to venture back into the realm of SciFi to consider what it would mean if the ability to construct such gravity wells were extended beyond the ground state of our shared values realized through collective actions to instead some kind of algorithmically targeted and artificially generated gravity well. As an example, perhaps one may wish to introduce a gravity well to target a rate of inflation in our economy. Or to eliminate the transmission of a virus.</p><p>Without trying to imagine how one might construct such a capability, let’s consider instead what the experience of being exposed to such effects could feel like. There are complicated philosophical questions around things like free will that come into play that may be easier to address if you lift the presumption of free will being a binary construct and instead only consider it as a relative property in comparison to the state of the world. For example, what kind of cereal you have for breakfast likely has minimal impact on the state of your surroundings so if there are any gravity wells in play they are not likely to reduce your free will towards such a choice, however whether you go to work in a tuxedo or swim trunks could impact your continued access to influence your workplace, so if you are a critical participant in realizing a critical targeted outcome, you may find that your ability to take off the tux obstructed. The funny part could be that the nature of such obstructions arise from gravity wells impacting not just you personally but all of your coworkers as well. Each individual moment where you consider changing into the swimsuit it may feel like a random encounter that pulls you back into a collared shirt, at least for another few weeks, and then next thing you know it’s time for a another Christmas break. With great power comes great responsibility as they say.</p><p>So why not just quit? Leave the world to its own mechanisms, go find an ocean and a tiki bar and put your toes in the water for a while. Your responsibilities be damned. In such questions I would turn another pastor who I recall describing to our church one way to think about our life as not isolated from the afterlife, but rather them being one and the same. Thus we should strive to make this world more heavenly because the life that we choose to live is a depiction of our values and there is to be any judgement of whether we are worthy of heaven we must seek to follow Jesus so that we can benefit from the sacrifice that He made for us.</p><p>Everyone has some kind of religion. Whether it is a religion of Christianity, another faith, or even the religion of social media addiction, whether we choose to allow God’s word to steer us or some timely internet advertisement, these gravity wells are going to pull us towards them. If we use the church as our navigation we get the added benefit of our soul’s salvation.</p><p>What if there is more than one gravity well? What if we can by our choices and actions steer not just ourselves but our family and our communities to a better one? What if the fight for your soul is so difficult because it’s so meaningful and important?</p><p>The first rule of peace club is that you have to talk about peace.</p><ul><li><em>Nicholas Teague, 08/11/2025</em></li></ul><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2F66rEsbptnxk%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3D66rEsbptnxk&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2F66rEsbptnxk%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/487fa311612ee7c26986d285f09aaa8d/href">https://medium.com/media/487fa311612ee7c26986d285f09aaa8d/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b379f741a752" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/the-difference-between-scifi-and-religion-b379f741a752">The Difference Between SciFi and Religion</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Double Entendres]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/double-entendres-4f720e65be78?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/4f720e65be78</guid>
            <category><![CDATA[computer-science]]></category>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Mon, 11 Aug 2025 11:20:23 GMT</pubDate>
            <atom:updated>2025-08-11T11:20:23.890Z</atom:updated>
            <content:encoded><![CDATA[<h4>I see your pair and raise you</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UVv8OdJ6emQUAJNX48R6HA@2x.jpeg" /></figure><p>Discrete embeddings in AI should have an inherent advantage over float vectors from simplified capacity for multiple interpretations based on context, which increases information content density. For comparable float capacity need “fuzziness” eg via noise injections to inference.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g50eNy6wgmEzNEM9bRfEOw@2x.jpeg" /></figure><p>Put differently, the more precise your linguistic articulation along every conceivable axis of definition, the larger will need be a vocabulary from which it is derived. Humans often speak with ambiguities and vagueness because is more efficient and allows to bypass trivialities.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*We3zD8VpjM3aMG8nsupGIQ@2x.jpeg" /></figure><p>NLP modality embeddings due to reduced dimensionality compared to eg images have higher prevalence of phenomenon resembling double entendres, &amp; mathematical properties resembling holographs as can map to higher dimensions. Sort of similar to hashing functions but not stochastic.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*10_3_dDSNARG5zlVK2HNRw@2x.jpeg" /></figure><p>Visual artists have touched on these phenomenon before, most strikingly with novelty depictions like the duck /rabbit or young woman / old maid or candlestick/faces are well known examples. They are rarer in images because the pixel density makes for more precise representations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pvNdg3OmgxQcMoYn2-8oaQ@2x.jpeg" /></figure><p>This silly picture of a dress broke the internet a few years ago from the simple unarguable presence of two unmistakable interpretations. Is it white/yellow or black/blue?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/128/1*xxcj2WSKazKm3yeMgWxNOA@2x.jpeg" /></figure><p>I recall there was a time ~30–40 years ago when the capacity for printed holograms first became a realistic reality. National Geographic magazine had a cover photo of a 3D image that would shift perspective based on your angle of view.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2oLMF7dkMEZadHYd4yrOEQ@2x.jpeg" /></figure><p>I am not sure but I think there might be some relevance to advanced physics, isn’t the whole premise of the ADS/CFT correspondence from string theory that the universe that we know is but a projection from a lower dimensional plain to the higher dimensional reality that we know?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0T-NTpFBn7tsgXo8MHWRWg@2x.jpeg" /></figure><p>Speaking back to the linguistic domain, as one extends a metaphor from single -&gt; double -&gt; triple entendre, etc, there must be a convergence towards a single unified form independent of the statements. Sort of similar to the geometric regularization hypothesis I’ve written about.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nDXSswiCDUSq5QiorlHogg@2x.jpeg" /></figure><p>I wonder if these linguistic phenomena have been so difficult to intentionally derive because most try to Build from the ground up. Perhaps an easier method would be to deconstruct a statement or a book to fuzz towards those constraints of entendres. Within will be found physics.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zNCo_loR9guavOsrWZClkQ@2x.jpeg" /></figure><p>After all, what are the fundamental laws of physics if not those nth entendres of highest order that can be interpreted to apply to every setting, context, and scenario. Find the exceptions and you’ve found where our current laws of physics need refinement. (That or dark matter.)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_aXAio8m7X2Affp7wVJpbQ@2x.jpeg" /></figure><ul><li><em>Nicholas Teague, 08/11/2025</em></li><li><em>Text via author’s Twitter thread 08/08 —08/ 10 2025, photos from a walking tour of Georgia Tech campus 04/15/2025</em></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4f720e65be78" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/double-entendres-4f720e65be78">Double Entendres</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Book 7: Inevitable]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/book-7-inevitable-fef0091f42c8?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/fef0091f42c8</guid>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Thu, 24 Jul 2025 11:25:30 GMT</pubDate>
            <atom:updated>2026-09-23T06:57:51.876Z</atom:updated>
            <content:encoded><![CDATA[<h4>From the Diaries of John Henry</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*X75vajzCiReoZL2219OdDg@2x.jpeg" /></figure><p>I present these essays as an imperfect rendition of the author’s attention in these recent years, in which I had largely turned away from writing as a full time pursuit to focus on a growing family and health.</p><p>To some extent these have been challenging years, as they say: “What got you here won’t get you there.” What got me here was unusual energy channeled into an extraordinary focus given the lack of meaningful commercial traction for so many years. I put some faith in the mythos of the field: if you aim for the moon you can at least expect to make it to orbit. I now see that what will get me there isn’t about inventions and citations. It is about humanity. One needs to put their faith in the kindness of neighbors, in the values of the flag, in the earnestness of scripture. One needs to do unto others as you would have them do unto you.</p><p>I think part of the reality I have faced is that having such a wide portfolio of public writings, particular those at least touching on scope at the frontiers of technology and innovation, has put me in an all or nothing position. Either my work is suitable for the public or it isn’t. Every single page. I can only hope that so many years of attempting responsible stewardship of this journal, and public speech surrounding, may serve to demonstrate the responsibility needed for such a position. I don’t claim to have the right to independently determine what is or isn’t appropriate for public speech in this context. Rather than pushing boundaries to limits of what can be said, in these chapters I have instead tried to focus on <em>what needs to be said</em>. Not just for my own interests, but for those of our society. That admittedly sounds grandiose, perhaps we who have lived as an early stage startup founder have a tendency for unreasonable grandiosity.</p><p>It is reasonable and appropriate for any executive to offer public speech in a manner that attempts to paint a picture of the world that they want to be true for them to succeed. The challenge is to retain the marriage of that speech to the foundations of reality. I have tried to do this in earnest, and every word of these essays is presented in earnest for a reader to make of it what they will.</p><p>Oh, and please, for God’s sake, don’t judge an author by the first things they ever wrote.</p><h4><a href="https://nicholasteague.medium.com/book-1-explorations-a1c81239d006">Book 1: Explorations</a></h4><p><em>August 2016 — July 2017</em></p><h4><a href="https://medium.com/from-the-diaries-of-john-henry/book-2-essays-bf663171ff96">Book 2: Essays</a></h4><p><em>August 2017 — August 2018</em></p><h4><a href="https://nicholasteague.medium.com/book-3-entrepreneurship-1fcb9bd4911e">Book 3: Entrepreneurship</a></h4><p><em>August 2018 — June 2019</em></p><h4><a href="https://nicholasteague.medium.com/book-4-everything-725f236e3cdb">Book 4: Everything</a></h4><p><em>September 2019 — July 2020</em></p><h4><a href="https://medium.com/from-the-diaries-of-john-henry/book-5-etcetera-d1910889e6dc">Book 5: Etcetera</a></h4><p><em>August 2020 — August 2021</em></p><h4><a href="https://medium.com/from-the-diaries-of-john-henry/book-6-endurance-192471c6a66">Book 6: Endurance</a></h4><p><em>September 2021 — August 2022</em></p><h4>Book 7: Inevitable</h4><p><em>September 2022 — September 2026</em></p><h4><strong>One</strong></h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZG9SuQL6OQzcx5XOZfEEaw@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/prepare-for-the-storm-d2d1c81a29b5">Prepare for the Storm</a></p><h4>Two</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S_Yr1LHdRnqUXO4BzHPgnQ@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/invited-talks-of-iclr2024-ba1a121b1cb3">Invited Talks of #ICLR2024</a></p><h4>Three</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CGskkV4A72X1hz1Tfvw_Fw@2x.jpeg" /><figcaption>Moonlight — Claude-Joseph Vernet</figcaption></figure><h4>Four</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*loUDLciX0hBoO2lPWPro0w@2x.jpeg" /></figure><h4>Five</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9kEyPc870QTwOXkAHWLVbA@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/geometric-regularization-from-overparameterization-5508c5a03983">Geometric Regularization from Overparameterization</a></p><h4>Six</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JiTb7LBeStlFbct3E4Z7Tw@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/on-investing-04fdcdf5129d">On investing</a></p><h4>Seven</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*svM7V5Vf3_h-NpBrCWwdDg@2x.jpeg" /></figure><h4>Eight</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*a-wbG7C-_p0CGAK9l7F7bw@2x.jpeg" /></figure><h4>Nine</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oEZMBBLCM7Jdhp2hxAM3hw@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/floridas-gambling-tribe-638097ebd292">Florida’s Gambling Tribe</a></p><h4>Ten</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bvkEZnc5r_dl5QoLg1DHfQ@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/automunge-technical-summary-85ecab8c09ae">Automunge Technical Summary</a></p><h4>Eleven</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S55Cg1p9WLNdDiLm7DHB8A@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/asymmetry-f58ad7e0e2b7">Asymmetry</a></p><h4>Twelve</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0mnzut78AYYRcwtijiqMwQ@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/rewrite-c3727113fc1b">Rewrite</a></p><h4>Thirteen</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*emDd6WwTtCcMwLtZFrLmtQ@2x.jpeg" /></figure><ul><li><a href="https://medium.com/from-the-diaries-of-john-henry/double-entendres-4f720e65be78">Double Entendres</a></li><li><a href="https://medium.com/from-the-diaries-of-john-henry/the-difference-between-scifi-and-religion-b379f741a752">The Difference Between SciFi and Religion</a></li></ul><h4>Fourteen</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_7IobUp5JWbd0Y4dFmM4Lg@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/sustainable-competition-for-sustainable-monopolies-1ff90a3ef049">Sustainable Competition for Sustainable Monopolies</a></p><h4>Fifteen</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sZgn9hy077oo-6aQEpfcWQ@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/chatflavors-e318ee83e4a1">ChatFlavors</a></p><h4>Sixteen</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gu4S8pz7xdhk49srjhTm7A.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/alignmentprotocol-f55ae32c23dc">AlignmentProtocol</a></p><h4>Seventeen</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-w6t7UfOgGHTBS9RAvgACQ.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/ai-economics-f06df944cbfe">AI Economics</a></p><h4>Eighteen</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KFLxGNQ66xO_hQktQjyW1w@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/ethics-of-scaling-compute-78559cb7e290">Ethics of Scaling Compute</a></p><h4>Nineteen</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*q9D1lKNK4RgtBF-o42fTGQ@2x.jpeg" /></figure><p><a href="https://medium.com/from-the-diaries-of-john-henry/aphorisms-for-all-1f61a889a861">Aphorisms for All</a></p><h4>Fini</h4><p><em>Copyright Nicholas Teague 2025, 2026, all rights reserved</em></p><p><em>“From the Diaries of John Henry” is registered trademark of Nicholas Teague LLC</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-DYFkIUt2azCQ7gqA6Cf0A@2x.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fef0091f42c8" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/book-7-inevitable-fef0091f42c8">Book 7: Inevitable</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Rewrite]]></title>
            <link>https://medium.com/from-the-diaries-of-john-henry/rewrite-c3727113fc1b?source=rss-191d0ae6b768------2</link>
            <guid isPermaLink="false">https://medium.com/p/c3727113fc1b</guid>
            <dc:creator><![CDATA[Nicholas Teague]]></dc:creator>
            <pubDate>Wed, 23 Jul 2025 02:42:45 GMT</pubDate>
            <atom:updated>2025-07-23T02:42:45.792Z</atom:updated>
            <content:encoded><![CDATA[<h4>Restated for clarity</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tVJ1scb85Mrfe6HBinGb8A@2x.jpeg" /></figure><p>One need not be an advocate for rewriting history to acknowledge that there are participants in the economy (and above the economy) whose primary path to victory is associated with rewriting history. Without a shared truth we fall but a shared truth cannot exist within isolation.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FqvGQ5Q3F3jQ%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DqvGQ5Q3F3jQ&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FqvGQ5Q3F3jQ%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/07979c9bfd5f32c9ca05402e1daa36f1/href">https://medium.com/media/07979c9bfd5f32c9ca05402e1daa36f1/href</a></iframe><p>The path to victory in this context is not to tighten borders and barriers, it is to acknowledge that our history itself, at every scale, will have traps and snares capable of being channeled to divert our attention away from higher objectives. We need to prioritize forgiveness.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iyKA8RTZD3j95ZARCCOxpQ@2x.jpeg" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c3727113fc1b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/from-the-diaries-of-john-henry/rewrite-c3727113fc1b">Rewrite</a> was originally published in <a href="https://medium.com/from-the-diaries-of-john-henry">From the Diaries of John Henry</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>