Looking back at my log, it just dawned on me that I have not written one of these since last December, this is largely because of my new adventures: running, learning to build AI products, and starting an applied AI company (as a solo founder) has been thoroughly time consuming. I would probably take the time to write more about my experiences in the future.

In the meantime, Around the Web Issue 39.

For books, go the end of post.

[Talking Points]

Simulating the birth lottery.

NIH Chief X Ross Douthat.

The three African Guineas.

What happened to reading?

The death of the Stack Overflow forum.

Record number of Americans are living alone.

Milestone 70 Percent 5-Year Survival Rate for Cancer.

Most common religious affiliation by country.

Walmart hits $1 trillion market value.

Best Goals of AFCON 2025.

Hausa Insurgency in North Nigeria.

Violent Crime Rates in the United States.

Indian Students sing “Aimasiko“

Mutsuhiro Watanabe Interview

Yoruba tribe in Ketou, Benin Republic.

Youngest Catholic Bishop in America.

Nassim Taleb on hard copy books.

[Longer Reads/Watch]

😜Fun Criterion

Here, David Deutsch explains the “fun criterion”. My initial thoughts about what he seems to be saying at the beginning was that “fun” is the feeling that arises when you’re making progress on a problem: when things are moving forward, you naturally want to keep going, and when you’re stuck or frustrated, the absence of fun signals that progress has stalled. In this sense, fun functions like a motivational feedback loop and can be used as a heuristic for deciding whether to persist or change direction.

But his claim is deeper than this “progress-as-motivation”. My reading is Fun is not merely a reaction to visible success, nor a reward for effort; it is an epistemic signal that internal error correction is occurring. Human thinking consists of interacting layers of explicit, inexplicit, and unconscious ideas that are constantly in tension. Fun arises when those competing ideas are being productively resolved, when confusion is decreasing, explanations are improving, or even when a better problem is discovered. Conversely, frustration is not just “no progress” but evidence of unresolved internal conflict: rival explanations are clashing without resolution. I think I like this idea very much, the fun criterion is not “continue if it feels good,” or even “continue it, even if you don’t feel good” (no pain no gain); but “continue if you are reducing error and resolving contradictions,” even when the work is hard, slow, or externally unsuccessful. This explanation maps quiet well to my experiences in my startup, even though I am don’t even have a product yet.

🎲Casino is Not the Only Option

Late last year, an essay landed in my siblings group chat. I read it and started flooding the chat with messages, and then realized quickly that this probably wants to be an essay. So I wrote one.

The original essay, titled “The Prison of Financial Mediocrity” by Systematic Long Short, argues that younger generations, locked out of traditional wealth paths and facing accelerating disruption from AI, are rationally turning to high-variance “casino” systems (crypto, prediction markets, sports betting) as the only remaining sources of perceived agency.

As white-collar career ladders erode and social media amplifies status anxiety, gambling becomes less about ignorance and more about desperation under constraint. The author frames this as a structural, not moral, phenomenon and concludes that the winning move is to “long the platforms”, the argument that because the house always wins, the only rational move is to invest in, or align with, the “casinos” extracting fees from the desperation. I argued against the proposition.

💼 AI: Agency Strategy For next 5 years

David Shapiro argues that the AI transition will dismantle wage labor faster than most people expect, and that high-agency individuals should respond across four fronts: where they live, how they invest, what kind of work they pursue, and, most importantly, what mission gives their life meaning. Remote work and automation decouple income from geography, making lower-cost, values-aligned communities more attractive. Capital ownership and disciplined saving matter more than ever as jobs erode. Traditional knowledge and single-task labor will shrink, while attention, experience, authenticity, and meaning-driven roles persist, often augmented by AI rather than replaced by it. Shapiro contends that a post-labor future demands shedding hustle-based identities and orienting life around purpose, creativity, and human flourishing, because no institution will “empower” you unless you first empower yourself.

Our Mathematical Reality

In this episode of Uncommon Knowledge, Peter Robinson moderates an enjoyable discussion featuring mathematician David Berlinsky, Princeton professor Sergiu Klainerman, and philosopher of science Stephen Meyer on the mysterious “unreasonable effectiveness” of mathematics in describing the physical world. They explored how mathematical truths function as an immaterial reality that exists independently of human invention, directly challenging the foundations of strict materialism. Klainerman emphasizes the “mathematical test of reality” through his work on black hole stability, while Meyer argues that the conceptual nature of these objective truths suggests they originate in a transcendent mind.

I have a related video essay on this subject matter:

Share

Giza Plateau’s Hidden Structures

Filippo Biondi discusses his groundbreaking satellite-based radar tomography that he says revealed immense, undiscovered structures deep beneath the Giza Plateau. Utilizing technology originally developed for military and industrial applications, his team identified massive vertical columns, spiral coils, and horizontal corridors extending up to 1.2 kilometers below the surface. These findings challenge the traditional archaeological narrative that the pyramids served merely as tombs, suggesting instead that the structures functioned as sophisticated mechanical devices designed for harmonic resonance and energy generation. Biondi highlights evidence of ancient flooding within these shafts and is currently proposing a collaborative, robotic exploration of the site to physically verify what he describes as a “paradigm-shattering” discovery that could rewrite the history of human civilization.

I listened to the episode on an 18km run earlier in the year, and I found it utterly fascinating. But the thing is, what I believe to be the underlying paper for this discovery was recently redacted. That being stated; I did not get the time to dig into the weeds: what still plausibly stands, what’s definitively wrong, what’s still up for debate, that kind of thing.

See the redacted paper.

Nigeria’s Mysterious Art Treasures

This is a fine documentary about Ile-Ife art; one of the excavation sites featured is very close to where I lived as child and teenager. One of the archaeologists in the film also wrote this book (which I’ve been planning to read); it looks like it will offer a more “solid” perspective compared to the takes on the subject matter by Samuel Johnson and Professor Akintoye.

Claustrophobic Metaphysics

This essay from Ed Feser provides a clear-eyed intellectual history of the shift from classical realism to modern subjectivity in Western philosophy. He critiques how thinkers like Descartes, Locke, and Hume moved away from prioritizing the objective natural world to focusing exclusively on the internal human mind, a transition that ultimately led to skepticism and idealism. Immanuel Kant was presented as a pivotal figure who attempted to synthesize these views by proposing that the mind imposes structural order on sensory data, yet simultaneously restricted human knowledge to mere appearances rather than reality itself. This “revolution in thinking” move is examined through its expansion into moral philosophy, where Kant argued that reason legislates its own ethical laws through the Categorical Imperative. The final argument goes something like this: Kant’s “internal turn” inadvertently fostered a culture of moral relativism and a sense of human arrogance by detaching truth from an independent, external foundation.

Subscribe now

How Earth’s Night Lights Became Our Most Honest Economist

Fascinating look at the unexpected intersections of technology and economics, Frank Jacobs over at Big Think explores how a Cold War-era satellite program inadvertently created our most reliable global economic indicator. Originally launched by the U.S. Air Force to track cloud cover for military operations, the Defense Meteorological Satellite Program’s sensors also captured the planet’s artificial light at night, illuminating everything from city sprawl to gas flares. Today, this nighttime luminosity serves as a fiercely independent fact-checker against official government narratives, capable of accurately mapping GDP, poverty levels, and even illegal fishing. Interesting data science story.

[Books/Papers 📚]

Book(s) I recently completed: [-]

• Hatching Twitter [Link][Review]

• Unbroken [Link][Review]

• WhatI Talk about When I Talk about Running [Link][Review]

• Endurance [Link][Review]

• Self Help from the Middle Ages [Link]

What I am currently reading:

• The Hard Things About Hard Things [Link]

• Aquinas: A Beginners Guide [Link]

• Communion: Finding My Way Back to Faith [Link]

I have spent some time listening to the two sides of the AI (and the future of jobs) forecasting debate. From where I sit, Tyler Cowen has the more optimistic take. On the other hand, you have the AI Futures folks. I read their AI 2027 last summer, and from my recollection, its forecast was directionally strong, but too fast, I think. They recently came out with this new report, AI 2040, which I have not yet read. But I did listen to what I believe is the gist of it in a two-hour podcast the other day.

As I listened, I wondered what Tyler Cowen might say in response to Daniel Kokotajlo’s arguments. So, I assembled a prompt for GPT-5.6 Sol, as one does these days, asking it to stage a debate between Kokotajlo and Cowen using their latest available materials and content, something light I could read after a hard day of work. The result, which you can read below is quite good. And I am deliberately presenting what GPT 5.6 sol gave me back, largely untouched, because 1) there is not much I could add given the time constraint of writing this essay 2) I believe that the “medium” itself is the message: The synthesis itself makes a case about the subject matter.

The Crux of it is, or at least part of the crux of whole debate, is that Kokotajlo assumes that AI capability growth may outrun institutional adaptation, while Cowen assumes that institutional friction, expanding demand and human preference will remain powerful even after capability becomes extraordinary. My reading of it is that Cowen offers the stronger model of the next several years; Kokotajlo identifies the regime change that becomes decisive if autonomous AI research and general-purpose robotics advance rapidly.

Here are the References synthesized into the debate:

So you can read (or watch) the source materials for yourself.

Cowen, T. (2026, July 8). Human life in a post-AGI world. Google DeepMind. https://tylercowen.com/human-life-in-a-post-agi-world-talk/

Cowen, T. (2026, June 28). My ARC talk on AI and jobs. Marginal Revolution. https://marginalrevolution.com/marginalrevolution/2026/06/my-arc-talk-on-ai-and-jobs.html

Cowen, T. (2026, May 13). Some non-obvious reasons why AI will create some transitional problems in employment. Marginal Revolution. https://marginalrevolution.com/marginalrevolution/2026/05/some-non-obvious-reasons-why-ai-will-create-some-transitional-problems-in-employment.html

Roberts, R. (Host). (2026, March 30). AI, employment, and education (with Tyler Cowen). In EconTalk. Econlib. https://www.econtalk.org/ai-employment-and-education-with-tyler-cowen/

Pangambam, S. (2026, July 13). Transcript of Daniel Kokotajlo interview: Diary of a CEO podcast. The Singju Post. https://singjupost.com/transcript-of-daniel-kokotajlo-interview-diary-of-a-ceo-podcast/

The Diary Of A CEO. (2026, July 13). He risked everything to warn you: No one is ready for what’s coming, and the AI companies know it! YouTube.

Veld, S. (2026, January 11). What happens when superhuman AIs compete for control? AI Futures Project.

*** GPT 5.6 Sol Synthesis** *

Moderator: By the 2030s, will most people still have jobs?

Daniel Kokotajlo: Not if companies successfully build superintelligence. The usual response—“technology destroys some jobs but creates new ones”—works only while humans retain capabilities machines lack. Once AI is better, faster, and cheaper than humans across essentially every cognitive task, it can perform the newly created jobs too. Add sufficiently capable robotics, and eventually almost all economically useful human labour becomes technically replaceable.

Tyler Cowen: That moves too quickly from technically possible to economically and institutionally implemented. Economies are not benchmark suites. Firms, governments, laws, buildings, energy systems, professional norms and consumer preferences change slowly. Electricity took decades to diffuse fully. Even very capable AI must pass through human organisations, and those organisations are full of bottlenecks. I expect substantial disruption and more leisure, but not the disappearance of employment.

Round 1: Does comparative advantage save human labour?

Cowen: Even when one party is better at everything, comparative advantage can preserve trade and specialisation. Humans will remain useful where our participation is legally required, socially preferred or tied to trust, legitimacy and personal meaning. People will pay to see a human musician, talk to a human political representative, receive human reassurance or experience something created by a particular human life.

There will also be enormous new sectors: gathering real-world data, conducting physical and biomedical experiments, integrating AI into institutions and helping different societies adopt it on their own terms. Humans may use AI throughout these jobs, but that does not mean the humans vanish.

Kokotajlo: Comparative advantage does not guarantee that wages remain high enough to sustain a mass labour market. Horses technically retained comparative advantages after automobiles arrived; their economic role nevertheless collapsed.

Your examples are plausible transitional jobs. But an AI that can design experiments, operate robots, persuade people, administer organisations and rapidly improve the relevant technology will progressively absorb them. Eventually the remaining human jobs exist because society politically reserves them for humans—not because humans are economically necessary.

Cowen: But “politically reserved” is not a trivial exception. Politics, liability, status and preference are part of economic demand. We do not consume only the cheapest technically adequate output. We consume stories, identities, authenticity and relationships.

Kokotajlo: Agreed—but that supports a boutique market for human authenticity, not necessarily employment for billions of people.

Round 2: Why has mass unemployment not appeared already?

Cowen: The current evidence should reduce confidence in imminent labour-market collapse. Even young-worker unemployment has not clearly broken from its pre-AI trend. AI-exposed work is changing, but aggregate employment is not falling off a cliff.

The more realistic near-term problem is transition. New jobs may be concentrated in slow, regulated sectors such as energy and healthcare. Employers may struggle to identify who is actually good at supervising AI. Governments may spend money poorly because they cannot distinguish competent AI-enabled workers from people producing polished-looking nonsense. That can create painful mismatch without creating permanent mass unemployment.

Kokotajlo: Current employment data are a lagging indicator of the scenario I am worried about. Frontier companies are not trying to automate the whole economy first. Their sequence is more like:

1. Automate parts of AI research.

2. Use that acceleration to develop much more capable AI.

3. Deploy the resulting systems across the economy.

Therefore, widespread unemployment may arrive only after systems have become strategically powerful. Waiting for unemployment statistics to prove the danger could mean waiting until the crucial decisions have already been made.

Cowen: That depends on a fast intelligence explosion. I assign more probability to bottlenecks, diminishing returns and imperfect deployment. Humans will slow the process even when the underlying models improve quickly.

Kokotajlo: Yes. The disagreement is largely about whether AI research automation produces an unusually fast takeoff. If progress stalls, your account becomes much more plausible.

Round 3: Will AI create more work—or make work optional?

Cowen: AI is already making some highly motivated people work more, not less. It raises the return to initiative: one ambitious individual with agents may accomplish what previously required an organisation. The immediate response for young people should therefore be to learn these systems intensely rather than assume there will be nothing left to do.

The winners may not be conventional rule-following professionals. They may be unusually curious, energetic people who repeatedly experiment with the technology. Intelligence remains valuable, but initiative, judgement, charisma and the ability to discover worthwhile problems become relatively more important.

Kokotajlo: That is good advice during the transition. In the late 2020s or early 2030s, many jobs may indeed consist of managing teams of agents. But “learn to manage AI” is not a permanent refuge if the AI eventually becomes better than you at managing AI.

I expect a period in which AI greatly increases individual productivity, followed—conditional on superintelligence—by a world where human employment is largely a political choice.

Cowen: You are describing the technological endpoint. I am describing the lived economic path. That path might last decades and contain enormous quantities of work, entrepreneurship, relocation and institutional rebuilding.

Kokotajlo: And I worry that the path could compress into several years.

Round 4: Who loses first?

Cowen: Not necessarily the poorest workers. A particularly vulnerable group is the highly paid, upper-middle-class knowledge worker whose advantage consists of processing information within established systems: consultants, lawyers, administrators, junior analysts and some academics.

They may remain employed but suffer a large status and income reduction. A partner earning $1.5 million might become a manager in an energy company earning $300,000. That is not destitution, but it could produce intense political backlash because these people currently possess disproportionate cultural and institutional power.

Kokotajlo: I agree that white-collar status disruption comes early. But I would emphasise entry-level positions. If AI performs the basic writing, coding and analysis through which juniors learn, firms may hire fewer beginners. Eventually, however, the distinction between junior and senior matters less: sufficiently capable agents threaten both.

Cowen: The apprenticeship problem is real. We will have to redesign education and employment so that people learn by supervising, testing and challenging AI rather than by spending five years producing first drafts. I have argued that a substantial share of education should involve learning to use AI properly.

Round 5: Is income the main problem?

Kokotajlo: No. People need both income and power.

A citizen’s dividend, AI tax or broad ownership mechanism can distribute some of the enormous output produced by AI. But employment currently gives ordinary people leverage. Workers can strike, withhold expertise and make themselves economically costly to ignore.

If almost all production comes from AI companies, data centres and robot fleets, governments may become less dependent on the population. People could receive comfortable payments while losing political influence. A post-work settlement therefore needs durable democratic control, not merely UBI.

Cowen: That is one of Daniel’s strongest arguments. A wealthy post-work society is not automatically a liberal society. Concentrated AI capability could become a power magnet for corporations or the state.

But I resist designing the entire economy around the assumption that labour will become worthless. Prematurely freezing a particular redistribution system could entrench incumbents and slow the very growth that makes abundance possible. Build energy, housing, healthcare, research capacity and broad access to AI first.

Kokotajlo: The danger is that “wait until displacement is obvious” becomes “wait until the owners of AI possess overwhelming bargaining power.”

Closing statements

Cowen: AI will radically alter work, destroy particular career ladders and reduce the returns to some forms of intelligence. Yet human wants are expansive, institutions are slow and economic activity is not a fixed quantity. Expect turbulent reallocation, more experimentation, new human-facing work and eventually more leisure—not a permanent society in which virtually nobody works.

Start learning AI now. Develop initiative, embodied skills, judgement, persuasion and the ability to work on problems the models cannot cleanly specify.

Kokotajlo: Historical reassurance is unreliable when the technology being developed is a general substitute for human cognition and may accelerate its own development. If superintelligence arrives, the question will not be which jobs AI can perform. It will be which roles society deliberately keeps human.

Prepare before unemployment becomes the headline: slow dangerous races, solve alignment and control, distribute ownership, and protect the political power of citizens in a world that may no longer economically need their labour.

Subscribe now

Book: Self-Help From the Middle Ages: A Journey Into the Medieval Mind by Peter Jones.

Medieval writers and thinkers, especially those thinking through the Seven Deadly Sins, had a serious understanding of the human psyche. They may not have had our vocabulary of dopamine, trauma, burnout, attention economy, clinical depression and the likes, but they understood something fundamental about the human person. They knew that we are unstable creatures, easily captured by desire, comparison, resentment, appetite, and despair.

Jones did a great job fleshing this out in his book “Self Help from the Middle Ages” which he organizes around the Seven Deadly Sins: Pride, Envy, Anger, Sloth, Avarice, Gluttony, and Lust. But he does not treat them as religious vocabulary. Instead, he treats them as some sort of diagnostic categories.

Pride is not simply pride, it is the refusal to see oneself truthfully. Envy is not merely wanting what another person has. It is the spiritual sickness of being pained by another person’s good. Anger is not only loss of control. It is also seduction, because rage gives us the intoxicating feeling that we are obviously, and unquestionably right. Sloth, especially in its older form as acedia, is not laziness. It is the utter collapse of care.

A hurry-look-around, would convince even the most dim witted persons amongst us that we have not escaped the Seven Deadly Sins. If anything, we have industrialized them - we blew it up. This is the part worth sitting with. “Our problems” are not new; they are old. In some respects, what is new is how we have used our technologies to intensify them.

Envy used to require proximity. You had to see your neighbor’s field, your colleague’s promotion, your friend’s marriage, your rival’s applause. Now envy arrives infinitely refreshed, algorithmically arranged (with precision, I must add); and backlit in your hand.

Pride used to need an audience. Now the audience is always potentially there, just boast away. Anger used to burn in the room, the court, the tavern. Now anger can become a personality, a business model, a media strategy. Ever heard of rage bait? sounds familiar? And please, don’t get me started with Gluttony.

You put all these together and all the talks about the “modern” mental crisis suddenly makes sense (What would be shocking is such crisis not happening)

And then, on the other end, we use technology to solve the very problems technology has ramped up. We create apps to manage the attention destroyed by apps (I use the Freedom app, for example). We create wellness industries to soothe the exhaustion produced by overwork. We create subscription products for loneliness, meditation platforms for people whose lives have been engineered into fragmentation. Vast section of our economies now grow out of misery, with some, offering half-solutions to the conditions they quietly depend on.

Inspite of our insane material success, the argument wages on, why are we not happy? - that is another separate, but related, kettle of fish.

Or are we doomed to fail? The book does not flatter us with the idea that we are uniquely broken. It places us inside a longer human story. The names have changed. The technologies have changed. The economies have changed. But the soul, that stubborn old thing, is still fighting many of the same battles.

And perhaps that is comforting. Not because our problems are small, but because they are shared. Others have been here before. Others have mapped these dark roads. Jones’s book asks us to pick up those maps again, not as antiquarian curiosities, but as instruments, maps for living.

Subscribe now

I am launching an experiment, the first version of Missing Pages, an AI-powered learning navigator for people who read seriously and want a better way to understand what their reading is becoming.

The initial version is simple. If you track books on Goodreads or similar apps, Missing Pages can import your reading history and turn it into a living knowledge map. Each book becomes a point in an atlas. Related books cluster together. Dense regions show where your attention has accumulated. Sparse regions hint at what may be missing.

From there, the Compass Agent helps you interrogate the map. You can select a cluster of books and ask: what does this collection understand well, and what does it leave out? Missing Pages then identifies knowledge gaps and can generate a short research essay to begin filling one of them.

The immediate use case is practical: help avid readers see their intellectual landscape and decide what to explore next. But the larger question is more ambitious.

AI is making research, summarization, and synthesis much cheaper. That does not make human judgment less important. It may make it more important. The harder problem becomes deciding what is worth learning, what question should be asked, and how scattered fragments of reading become a coherent picture of the world.

Missing Pages is my first public experiment in that direction. It starts with books because book trackers give the product a clean wedge: Goodreads users already have structured data and a real habit of cataloging what they read.

But the larger idea is (personal) knowledge as a structured, navigable, eventually collaborative asset: maps, essays, learning sprint people can share, fork, and build on.

If you are not the initial target user, that’s OK, I am mostly looking for reaction to the idea and demo, especially if you can imagine this being useful in other contexts.

Link for Feedback: https://tally.so/r/Pd28kd

Subscribe now

A few weekends ago, I started building this real-time simulator of bacterial evolution and antibiotic resistance (calling it evodish).

I just got around to finishing it up.

Antibiotic resistance is still a very big problem. In the United States alone, more than 2.8 million antimicrobial-resistant infections occur each year.

In this simulator, you drop antibiotic discs into a digital petri dish, natural selection unfolds.

I tried to capture as much biology as possible without overcomplicating it (random mutations, horizontal gene transfer, metabolic fitness costs, etc.).

> 💻 Try it here: evodish.vercel.app

> 📺 Video Guide:

Subscribe now

Last Fall, I ate my own cooking, put my money where my mouth is. After my Postdoctoral Fellowship at Georgia Tech, I transited to a research affiliate role at the Institute and I decided not to take a new employment.

About a year ago now, I stumbled on a LinkedIn post, where a pertinent question was asked, just before a brief articulation of the obvious incursion of AI tools into knowledge work. The fellow wrote, “What’s left in research when so many things can be done by AI? 🤔 …”

Some people who might have found that take a bit hyperbolic at the time might not anymore, particularly if they have tried tools like CodeScientist, Autodiscovery, and Kosmos, to mention just a few.

My response was as follows:

"A researcher could, and probably should, deliberately increase their breadth (and depth) of knowledge/competency. Many aspects of research will be aggressively automated over the next few years (This is true for all knowledge work) So, insights (reads: your knowledge bank) would be ultimately highly priced in a world where some research can be done quasi-on-demand. And the 'insights' are going to scale to the extent to which one leverages AI; you probably don't want to be the third servant, in the Parable of Talents who buries the one talent received into the ground, and ends up with nothing. And this parable maps to what we will be seeing going forward. 5 to 10, 2 to 4, and so on. The talent is your knowledge of what you know is possible. The (research) questions you ask. It's probably the best time to be a generalist."

Today, my response could really use some workshopping, but that’s not the reason for this essay. The only thing I would add is that, as a hypothesis, we are approaching even more pronounced barbell territory: hyperspecialist or (hyper)generalist. A plausible scenario is that everything in the middle will be gobbled up by AI. We can argue about timeframes, but again, that isn’t the purpose of this essay. This essay is, in part, an attempt to show what acting on that hypothesis looks like in practice.

After my fellowship, I passed on a couple of opportunities that would have been great in the pre-ChatGPT era (we really need a name that rolls off the tongue for this). But as I was saying, history is no longer “settled”; it’s moving fast — too fast in some domains. The future is profoundly uncertain, and precisely because of this, actions matter more than usual.

And it was principally the barbell and the “moving history” arguments that propelled me in this direction, in spite of the fact that the move ain’t cheap, to say the least.

And now, with the privilege of hindsight, I realize I actually could have used some rest. The path through graduate school and a postdoc is fiercely intellectually stimulating, but undeniably stressful. Add the years spent navigating the U.S. immigration system as an international student, alongside the recent relentless pace of tracking the AI disruption like a hawk, and stepping back makes some sense.

Subscribe now

I have had a fairly eclectic academic journey so far, perhaps this exposes my generalist bias. I hold a PhD in Biochemistry and Molecular Biology. I did some work in yeast genomics and studied Caenorhabditis elegans early in grad school. As a senior graduate student and postdoc, I published several works on applied AI/ML in biomedicine.

One of the main ‘skill issues’ I’ve had over the past couple of years (and for the engineers reading, that pun is fully intended) arose simply by virtue of being in a research environment. It was the ability to take an idea and turn it into production-level software with all the bells and whistles. I could do bits and pieces, but not the full gamut.

To solve this, the core challenge of my break became building intelligent systems end-to-end: models, agents, backend infrastructure, and user experience. What people now call full-stack AI/agent engineering. It felt like a reasonable way to expand my wingspan. I was already adjacent to the field through a number of side projects, so this seemed like a natural escalation. It was a sufficiently hard problem, and increasingly, an important one, since the next iteration of knowledge work is clearly automation (for better or for worse). And how long the skillset itself will be “useful” simply depends on which Substack newsletter you read, or which AI lab CEO you listen to.

However, as I hinted earlier, my break soon turned into a much-needed respite. I worked less — by virtue of being more focused (and leveraging AI) — read more books, played video games, and exercised much more.

I’ll give an example. All my life, I had wrongly assumed I was not a good long-distance runner, despite being, at least by American standards, very active. These days, I run 10 kilometers most days of the week. (It’s crazy how we tell ourselves we are not good at things we simply haven’t tried enough.) Along the way, I built a holistic, bespoke personal operating system from the ground up, one that incorporates work, learning, exercise, rest, and more.

However, one of the main projects I am working, now close to an alpha launch, is an experiment to test a hypothesis on what knowledge navigation, consumption, and collaboration might look like in a world where research is commodified.

Over the next weeks and months, I will share more artifacts (essays, apps, videos) publicly, and I would be glad to hear your feedback. I will share my “in-the-trenches” micro-updates mostly on Substack Notes, and I will also collect them here. General essays will be on Episteme Engine, while technical AI and engineering deep dives will be on my AI blog — The Epsilon. Video updates will be shared here as well.

Calling this a career break is somewhat of a misnomer, more like building through an AI transition (admittedly, that doesn’t roll off the tongue either). The work didn’t stop; it changed shape. Whether this turns out to have been the right call is still an open question. But even if the barbell hypothesis is wrong (which I strongly doubt), a wider wingspan is not the worst thing to have built.

P.S.

1. If you are an experienced engineer, especially in the United States, looking for a new AI adventure around (scientific) research, biomedicine, or science education, feel free to reach out to discuss the possibility of teaming up. I have far too many projects and feature ideas, and too little time.

2. I am considering publishing a full length memoir essay at some point in the future, structured as a sequel to “Plans Are Overrated“, express your interest here.

I listened to the first part of this podcast on my 10k run today.

There are many threads running through the podcast, but I will focus on just one thread.

Dr. Alok Kanojia, whom I encountered for the first time today, eloquently explains that the ego is constructed from the labels and definitions we assign to ourselves, making it an inherently comparative entity. When we fall into the trap of comparing ourselves to others or copying what we see them pursuing (often driven by external conditioning from social media where we adopt others’ desires as our own) we end up chasing goals that do not align with our true selves. While allowing the ego’s comparative nature to fuel our ambition might lead to worldly success and gratify our desire to prove ourselves, it ultimately comes at the cost of genuine happiness because the ego is never satisfied and continually moves the goalposts. To find true fulfillment, he argues that we must step away from these externally programmed desires and peel back the comparative layers of the ego to tap into our authentic, intrinsic drive.

(As a Girardian of some sort myself, it wasn’t particularly hard for me to see that his analysis is drenching in the language of “Mimesis” and “Metaphysical Desire”)

This quote from the classic epistolary novel The Sorrows of Young Werther (published in 1774) by the German polymath Johann Wolfgang von Goethe, has to be the shortest best thing I have read on this matter at hand.

I urge the reader to read it carefully.

“Since we are so made that we compare everything with ourselves and ourselves with everything, happiness or misery lies in the objects we associate ourselves with. . . . The power of our imagination, driven by nature to elevate itself, nourished by the fantastic images of literature, raises up a series of beings of whom we are the lowest and everything outside ourselves seems more glorious, every other person more perfect. And that happens quite naturally. We feel so often that we are lacking so much, and just what we lack another person often seems to possess, to which we also add everything that we possess, and project a certain ideal contentment on top of it. And thus the other person is made happy, complete, and perfect, a creature of our own making.”

given the times we live in, this extract is even more telling:

“...nourished by the fantastic images of literature, raises up a series of beings of whom we are the lowest and everything outside ourselves seems more glorious, every other person more perfect.”

Literature?!

Talk about Social Media.

If Johann Wolfgang von Goethe could visit us for a day, I wager that he would lose his mind, and so would his compatriots.

Share

I have also published two essays on the subject matter in the past:

We Are All Copycats

Prestige as an Illusion

Subscribe now

An essay landed in a group chat the other day. I read it and started flooding the chat with messages, and then realized quickly that this probably wants to be an essay. So here we are.

The original essay, titled “The Prison of Financial Mediocrity” by , argues that younger generations, locked out of traditional wealth paths and facing accelerating disruption from AI, are rationally turning to high-variance “casino” systems (crypto, prediction markets, sports betting) as the only remaining sources of perceived agency.

As white-collar career ladders erode and social media amplifies status anxiety, gambling becomes less about ignorance and more about desperation under constraint. The author frames this as a structural, not moral, phenomenon and concludes that the winning move is to be “long the platforms”, the argument that because the house always wins, the only rational move is to invest in, or align with, the “casinos” extracting fees from the desperation.

(Another, more general side note on the original essay: I read what is going on as a symptom of late-stage capitalism, one in which pressure is no longer absorbed quietly but expelled through multiple “exhaust pipes.” The same blocked-agency pressure that vents economically through casino-styled systems is also venting politically (say, Groypers-styled movements). The casino is just one outlet. It is not the only one—and it may not even be the most dangerous.)

A False Dilemma

The essay quietly advances a binary:

1. White-collar work is cooked

2. Therefore, gamble

This framing collapses too many alternatives and overfits to a narrow slice of the economy. Even granting that many white-collar knowledge jobs are under threat from AI, the author underweights blue-collar and asset-backed paths (plumbing, construction, electrical work, welding, farming) fields that can scale into real businesses, compound locally, and remain stubbornly resistant to automation. These are not consolation prizes nor are they sexy but they are parallel wealth tracks that don’t depend on winner-take-all dynamics.

That critique stands. But there is another omission that matters just as much.

The Missing Third Option

What the essay leaves out is that AI does not only destroy ladders, it reshuffles where leverage lives.

The choice is not simply between grinding obsolete roles and betting your future on high-variance speculation. There is a third option: positioning oneself at the system level rather than the task level.

AI collapses the value of many individual tasks, but in doing so it raises the value of roles that define goals, set constraints, integrate outputs, arbitrate tradeoffs, and absorb responsibility when systems fail. These are not “AI-proof” jobs in the traditional sense. They are AI-adjacent leverage points, places where human judgment, accountability, and synthesis still matter precisely because the system has become more automated.

(Also, a friendly reminder that there is such a thing as the Jevons Paradox: while AI automates tasks, humans still orchestrate workflows, pushing expectations higher. There are legible, logical arguments that we should expect more work, not less, much of it entirely new)

The core of the argument here is not necessarily about learning the next hot skill or chasing “prompt engineering” as a credential. It is about understanding how AI reorganizes workflows and placing yourself where coordination, interpretation, and consequence still concentrate. That path does not necessarily promise the dopamine hit of a moonshot, but it preserves agency without requiring existential bets.

Gambling, But With a Seatbelt

On “gambling” itself, the sane position lies between moral panic and evangelism.

There is a categorical difference between capped, asymmetric optionality and full-blown, identity-level risk. Holding casino-styled financial assets as a speculative hedge is not the same thing as staking one’s future on memecoins or perpetuals. The essay claims to be descriptive, but its tone risks normalizing behavior that is, at scale, financially and psychologically ruinous.

The deeper problem is not ignorance of odds. Most participants understand the house edge. What they lack is a credible alternative path that feels agentic. When the baseline is “treading water indefinitely”, perpetual movement without motion, even a slim chance of escape starts to look rational.

The Inflation of What Feels Necessary

Another blind spot is excessive upside-seeking masquerading as inevitability. Not everyone is merely seeking dignity or stability; many are seeking extraordinary outcomes in a world where most outcomes are, by definition, ordinary.

Even venture capital is structurally a form of gambling. Most startups fail. A tiny fraction generate outsized returns. The difference is that good VC-style bets often leave participants with something even when they fail: skills, networks, judgment, and optionality (conditioned on who the participant is, of course)

Pure gambling leaves none of this behind.

That distinction points to the strongest corrective insight.

The Only Bets Worth Making

The best bets are those whose failure modes still resemble success.

Learning can compound even when companies die. Reputation can survive even when products don’t. Skills can transfer. Networks persist. Optionality expands. These are antifragile payoffs. Lose the bet, and you are still better positioned than before.

Pure gambling offers no such cushion. Lose, and you are strictly worse off. Win, and you may still be worse off psychologically.

All of these begs the question: what is even gambling in the context of these various assets that I am, admittedly, not terribly familiar with?

One way to answer this via negativa is, it is not pure gambling if the failure modes still resemble some form of success (and I know the devil is in the details, but I’m not ready to spill more ink here; reasonable people can fill in the gaps as they see fit.)

But of course, there are more straightforward(?) way of thinking about it - does it involve creation of real value? and is it merely a game of chance?

A Better Framework

All of these invites better questions:

• Where does agency still exist in this system?

• Where does effort still map to outcomes?

• What paths offer convex upside without existential downside?

• What roles become more valuable because AI moves through the economy?

To conclude, we should address the cynicism embedded in the “Long the Platforms” thesis. The proposal is not to escape the prison of financial mediocrity by breaking the bars, but by repositioning oneself as the warden. To “own the casino” is to place a levered bet on the continued desperation of one’s peers; it presumes a steady population that remains desperate enough to gamble, yet solvent enough to pay the rake. Structurally, this is a short on broad-based human flourishing. True agency shouldn’t be found in collecting tolls on the road to ruin, but in building the off-ramps.

Subscribe now

My Other Publications

My latest WILLs, “What I Listened to Last week” Issue 6.

You can also find my latest AI newsletter here: “Gemini 3.0 💻, Kosmos 🌌, and Why AI Changes Everything ⚪️”

(For books I have read, reviewed, or currently reading, go to the end of the newsletter)

[Talking Points]

👝Dust in the Baggie - Billy Strings

💃🏾Kpanlogo Dance, Ghana.

👨Man walking across the world.

🗝️35 year old lives in 1946.

🇺🇸Where can Americans afford to live solo?

⚽History of world cup match balls.

🎺Asake - Red Bull Symphonic.

🐍Backend Devs are switching to Python.

🫂Why we are getting more stupid.

🩸A New Cancer Blood Test.

➕The Strange Maths that predicts almost anything.

🇦🇫Public Execution in Afghanistan.

🤼‍♂️John Cena final entrance.

🎵AI generated song that hit #1 on Billboard.

🧗🏽‍♂️Sisyphus unpacked

[Longer Reads/Watch]

🏚️ End of Progress Against Extreme Poverty

Max Roser argues here that the historic era of rapid global poverty reduction is effectively ending, with progress projected to stall and potentially reverse after 2030, quoting World Bank projections. While past successes were driven by booming economic growth in Asian nations like China and India, the world’s remaining extreme poverty is now concentrated in stagnant Sub-Saharan African economies.

Source.

What is different today is that the majority of the world’s poorest people are stuck in economies that have been stagnating for a long time. Consider the case of Madagascar. In the long run, the country has not seen any growth at all: GDP per capita in Madagascar is about the same today as it was in 1950. As a consequence, the number of people in extreme poverty increased in line with the country’s population growth. In richer countries, it is possible to reduce poverty by reducing inequality through redistribution, but a country like Madagascar cannot reduce its share of people in extreme poverty through redistribution. This is because the mean income is lower than the poverty line; if everyone had the same income, everyone would be living in extreme poverty.

The situation is similar in other countries: in the Democratic Republic of Congo, Mozambique, Malawi, Burundi, and the Central African Republic, more than half of the population lives in extreme poverty.

How does this end of reduction of extreme poverty narrative square up with the so-called “AI Abundance”?

I think the conflict, if any exists, partly lies in the gap between digital and physical reality. While AI creates an abundance of intelligence and demonetizes some aspects of services like healthcare, it cannot be eaten, nor does it inherently build the roads, power grids, or sanitation required to lift the “bottom billion” out of destitution. Worse, AI-driven automation threatens to kick away the ladder of low-cost manufacturing that nations like China used to rise, potentially leaving African economies stranded without a growth model. Unless AI specifically solves physical costs (energy, logistics) or drives a shift from local growth to massive global redistribution (perhaps a more cynical read would be something along the lines of ‘recolonization?’), we face a bifurcated future: post-scarcity/abundance for the connected, and continued stagnation for the rest.

🔥 Girard’s Apocalypse.

I have written my fair bit on Girard’s mimetic theory, which you can find here, here, and here. But my analysis of the subject never took any serious theological-political dimension. In this pair of excellent essays from the Substack, João Pinheiro da Silva and Mauricio G. Righi explore the existential stakes of René Girard’s “mimetic theory” and how venture capitalist Peter Thiel has adapted it’s insights.

In Part 1, “Girard’s Apocalypse,” the authors outline Girard’s central thesis: that human societies were originally built on the “scapegoat mechanism,” a collective act of violence against an innocent victim that creates temporary peace. However, the authors argue that the Christian revelation of the victim’s innocence has permanently “broken” this mechanism, leaving modern society in an apocalyptic state. Without the ability to vent societal tensions through a shared sacrifice, mimetic rivalry (the imitation of others’ desires) escalates without a natural limit. This creates a global crisis where traditional “restraining” forces (katechon) have failed, leading to a world of total competition and the constant threat of escalating violence that no longer has a religious “safety valve.”

In Part 2, “Thiel’s Diabolical Deviation,” the focus shifts to Peter Thiel, a former student of Girard who, they argued, has “weaponized” these insights. Thiel accepts Girard’s diagnosis of a world in mimetic crisis, but rejects Girard’s theological conclusion of total non-violence and renunciation. Instead, Thiel performs a “diabolical” inversion by using Girardian tools to build a new, technological katechon, a sovereign power capable of restraining the apocalypse through surveillance, monopoly, and political domination. By attempting to re-engineer a form of “pagan” sovereignty in a post-Christian world, Thiel is seen as a “Girardian against Girard,” utilizing the mentor’s revelation of human violence not to transcend it, but to master and manage it for the sake of worldly survival and strategic advantage.

🚗 The Human Driver is a Failed Experiment?

I took my first Waymo ride this fall, and one of my first thoughts was something along the lines of “Jeez, the stuff is too good, how the hell are humans supposed to compete.” The drive, right in the middle of Atlanta, was too smooth and very cozy. I came to the conclusion that humans should probably be driving less, then I looked up the data. Boy, was it good.

You can find more of the safety data in the Waymo Safety Report.

In this New York Times opinion essay, a neurosurgeon effectively argued that the adoption of autonomous vehicles is a public health imperative:

If Waymo’s results are indicative of the broader future of autonomous vehicles, we may be on the path to eliminating traffic deaths as a leading cause of mortality in the United States. While many see this as a tech story, I view it as a public health breakthrough.

There’s a public health imperative to quickly expand the adoption of autonomous vehicles. More than 39,000 Americans died in motor vehicle crashes last year, more than homicide, plane crashes and natural disasters combined. Crashes are the No. 2 cause of death for children and young adults. But death is only part of the story. These crashes are also the leading cause of spinal cord injury. We surgeons see the aftermath of the 10,000 crash victims who come to emergency rooms every day. The combined economic and quality-of-life toll exceeds $1 trillion annually, more than the entire U.S. military or Medicare budget.

ℹ️Connection Among Information, Intelligence and Decision

I have always enjoyed reading Dembski’s essay on his many critiques of materialism in biology/life.

Here, he argues that information, intelligence, and decision are intrinsically linked by the fundamental act of “ruling out possibilities.” He argues that information is a top-down process where specific outcomes are realized to the exclusion of others, which mirrors the etymology of both intelligence (inter-legere, to choose between) and decision (decidere, to cut off). By defining the creation of information as a discriminatory act of choosing one path over countless others, Dembski contends that the “selection” observed in the biological world is not a byproduct of blind forces.

As a side note, this brief etymological tour reveals that Darwin’s great coup was to coopt the term selection, previously associated with the conscious choice of purposive agents, and combine it with the term natural as a way of negating the role of such intelligent agency. In the term natural selection, Darwin therefore attempted to recover all the benefits of choice as traditionally conceived, and yet without requiring the services of an actual intelligence. Thus to this day we read such claims, as by Francisco Ayala, that Darwin’s greatest discovery was to give us “design without designer.” Likewise, Richard Dawkins describes Darwin’s great coup as explaining the appearance of design in the absence of actual design.

And what about microevolution? If I am not wrong, I believe, elsewhere, Dembski reconciles the aforementioned argument with natural selection & microevolution by framing them as “subtractive” rather than creative mechanisms. Natural selection clearly generates information by weeding out the unfit, he argues that this process merely filters pre-existing genetic data without generating new Complex Specified Information. Blind evolutionary forces can fine-tune existing biological structures (microevolution), but they lack the decision-making capacity required to conceptualize and build them from scratch, say.

Share

🔔 Double Pendulum Aren’t Chaotic

This excellent video essay challenges the popular notion that double pendulums are inherently chaotic by visualizing their behavior through “phase space” fractals (geometric maps). While admitting that many starting positions lead to unpredictable, divergent motion (chaos), the creator demonstrates that specific initial conditions (particular angles or momentum levels) result in stable, repeating patterns like “pretzels,” “hearts,” or Lissajous curves.

By mapping these behaviors in such beautiful motions, the video reveals “islands of stability” hidden within chaotic regions and observes that while low-energy starts are typically stable and medium-energy ones chaotic, ultra-high momentum starts can surprisingly reintroduce order, proving the system contains pockets of beauty and predictability amidst the disorder.

I was reminded of Nassim Taleb’s “Barbell Strategy” take in one of his books, after watching the video, which advocates for operating at the extremes while avoiding the “muddled middle.” Taleb suggests allocating resources to either hyper-safe assets (stability) or/and hyper-aggressive bets (high variance but high potential) to avoid the hidden fragility of “moderate” risk. I think some “law of everything,” might be hiding here in plain sight.

👴🏻The Common Sense of John Searle

I have been interested in some contentious (philosophy) subject matter that had eventually led me to read about some of John Searle’s work, chief among them would be the nature of the intelligence in artificial intelligence, and free will.

Ed Feser wrote this short synopsis of some of Searle’s work, after he passed away in September.

On AI:

Imagine that Searle, who knows no Chinese, sits in a room with a set of Chinese symbols and a rulebook in English telling him which combinations of symbols to give out in response to written questions slipped to him through a slot in the door. The rulebook does not tell him what the symbols mean; it simply allows him to mimic the behavior of a person who does. What Searle would be doing in this scenario, he argued, is essentially what a computer does: manipulating symbols according to the rules of an algorithm. The resulting mimicry, no matter how convincing, would not yield genuine understanding of Chinese. Likewise, what computers do can never amount to the operations of true intelligence, but only a simulation of it.

(also see this nice TED animation on the Chinese room thought experiment)

On free will

…He argued against the idea that our actions are the necessitated effects of mental events of which we are merely the passive observers—as if everything we think and do simply happened to us. Choice, of its very nature, involves an irreducible and persisting self, which actively brings things about as a result of deliberation. There is a causal gap between our beliefs and desires on one hand and our behavior on the other, and only the self can fill that gap, by way of its agency. Searle’s view is that whether or not we can strictly prove the reality of freedom, we have no intelligible model of human action without it.

On his metaphysics:

The main weakness of Searle’s work is that he never developed a metaphysics that was as carefully worked out as his anthropology. Searle’s materialist critics often accused him of being a dualist in the Cartesian mold, a charge he always denied. He was as committed as the materialists were to the thesis that the natural world is all there is. He simply thought their way of fitting human beings into that world was simplistic. The trouble is that the conception of nature he shared with them, which he never seriously questioned, makes his position unstable. When he emphasized the continuity of human beings with the larger natural world, he sometimes sounded like he was offering just another riff on materialism. But when (as was more often the case) he emphasized how radically different the human mind is from everything else in nature, the charge of dualism was hard to rebut.

Subscribe now

[Books/Papers 📚]

Books I recently completed:

• Ultralearning by Scott Young [Review]

• Stillness is the Key [Review]

• Secrets of Sand Hill Road [Review]

• Reshuffle: Who wins when AI re stacks the knowledge economy [Review] (I plan to publish an essay about the ideas in the book early next year.)

What I am currently reading:

• Hatching Twitter: A True Story of Money, Power, Friendship, and Betrayal by Nick Bilton

• Shoe Dog: A Memoir by the Creator of Nike by Phil Knight

• Google’s Introduction to Agents.

Book Title: Secrets of Sand Hill Road.

I wanted to learn more about venture capital, so I picked up the book Secrets of Sand Hill Road written by Scott Kupor of Andreessen Horowitz. The following are some of my notes grouped (loosely) by themes.

I. The Tyranny of the Power Law

One cannot understand the behavior of a venture capitalist without first grasping the mathematical reality they live in: the Power Law. In most walks of life, outcomes follow a bell curve. In venture capital, returns are heavily skewed. As Kupor illustrates using a baseball analogy, VCs aren’t looking for a high batting average (getting a “hit” every time); they are looking for “at bats per home run”.

The math is stark: VCs expect >50%? of their investments to return nothing. To offset those losses and generate the massive returns their own investors (Limited Partners) demand, they need the winners to return 10 to 100 times the invested capital. This is why a VC might pass on a perfectly good business that has a high likelihood of becoming, say a $50 million company; it simply doesn’t move the needle for a fund that needs to return hundreds of millions.

Understanding this helps to realize that when a VC pushes for hyper-growth, they are simply trying to survive the math of their own industry.

II. The Structural Foundation: C Corps and Stock Classes

Everyone knows corporate structure isn’t just paperwork; it’s strategy. Kupor emphasizes that while the LLC or partnership model might seem tax-efficient for a small business, the C Corporation is the non-negotiable standard for venture-backed startups. This is largely because the Limited Partners (LPs), the endowments and foundations funding the VCs, are tax-exempt entities.

And if a tax-exempt entity receives income from an active trade or business (like a startup) through a “pass-through” entity like an LLC or partnership, that income is classified as Unrelated Business Income (UBI) and is subject to tax. This creates a “tax penalty” for the LP, effectively eroding their tax-free status for that investment. By investing in a C Corporation, the tax liability remains contained within the corporation itself, blocking it from passing through to the LPs and preserving their tax-exempt standing.

Furthermore, the C Corp structure facilitates the necessary hierarchy of ownership: Preferred Stock vs. Common Stock. VCs require Preferred Stock to secure specific economic and governance rights (like liquidation preferences), while founders and employees hold Common Stock.

Take liquidation preferences, it ensures that in a “downside” scenario (e.g., the company sells for less than hoped), the VCs get their money back before the founders see a dime. Conversely, if the company is a massive success, VCs can convert their Preferred shares to Common and participate in the upside. This structure protects the investor’s capital in failure while allowing them to share in the “home run,” aligning their risk profile with the high-stakes nature of the bet.

III. Misaligned Incentives, Fund Cycle, & “Ruling from the Grave”

One critical factor Kupor highlights is the Fund Lifecycle. Most VC funds have a 10-year lifespan. If a VC invests in your company during year 8 of their fund, they may be under immense pressure to force a liquidity event (like a sale) sooner than you would like, simply to return cash to their LPs before the fund expires .

Startups must also watch out for the “Zombie Fund” scenario. If a VC firm has performed poorly and is unlikely to raise a subsequent fund, they may lose the ability to “reserve” capital for future rounds. This can leave startup stranded without the follow-on capital they expected. Knowing where VC stands in their own fundraising cycle is just as important as them knowing startup’s financials.

This ‘alignment’ discussion also extends to governance. Kupor introduces the somewhat macabre but critical concept of “ruling from the grave”. This occurs when a founder retains a board seat purely by virtue of being a founder, even after they have left the company. To prevent a former operator from holding the company hostage, board seats should be tied to current service (say, the CEO role) rather than the individual person.

IV. The Art of the Pitch and Early-Stage Decision Making

When early-stage VCs decide where to invest, they are often operating with very little hard data. Consequently, their decision-making heuristic relies heavily on three pillars: People, Product, and Market. Of these, market size is paramount because of the aforementioned power law, a great team in a small market will still fail to return a venture fund.

A successful Pitch must therefore articulate a narrative around these elements:

• Market Size: convince the VC that the opportunity can generate billions in value.

• Team: Specifically, “founder-market fit”, why this team is uniquely qualified to solve this problem.

• Product: Is it a vitamin (nice to have) or an aspirin (need to have)?

• Go-to-Market: A plausible strategy for acquiring customers.

V. Valuation: The “Dark Art” and Convertible Debt

Valuation is often more art than science. How does a VC conduct valuation of startups? Kupor outlines a few methods, ranging from Comparable Company Analysis (looking at public comps) to the Option Pricing Model (OPM). However, for early-stage startups, it often boils down to a “what do I need to believe” analysis, calculating what the company must look like in ten years to generate a 10x return.

On a related subject: In earlier rounds, founders often use Convertible Debt. This allows them to raise capital quickly without setting an explicit valuation immediately, “kicking the can” down the road to a priced equity round. While efficient, founders must be wary of valuation caps. A cap sets the maximum valuation at which the debt will convert into equity. If your company grows wildly and raises a Series A at a valuation far above the cap, the debt holders get to convert at the lower cap price. Effectively, you have priced the round, you just set a ceiling on it, giving early investors a guaranteed lower price relative to your future success.

VI. Protecting the Cap Table: Vesting and Transfers

The “prenup” elements of a startup are vital for long-term health. Founder Stock Vesting is essential not because partners don’t trust each other, but to align incentives over the long haul which is typically a four-year period with a one-year cliff.

Equally important are Transfer Restrictions and Rights of First Refusal (ROFR).

ROFR: If you want to sell your shares to a third party, you must first offer them to the company (and often the investors) at the same price. This creates a “chilling effect” on buyers, who don’t want to negotiate a deal only to have it snatched away by the company.

Transfer Restrictions: These are even stricter. They dictate that you cannot sell your shares at all without the company’s consent. This ensures that the founders remain “locked in” and aligned with the long-term vision, preventing them from cashing out early while the VCs are still holding the bag.

Finally, regarding exits, Kupor champions Double Trigger acceleration as the industry standard for fairness. Single Trigger means your unvested stock immediately vests solely upon the sale of the company (Change of Control). Acquirers hate this because they are often buying the company to get the talent. If you fully vest the day they buy you, say, you have no financial incentive to stay. Double Trigger, on the other hand, solves this tension. It requires two events: 1) The sale of the company, AND 2) Your termination without cause (or leaving for “good reason”) . This protects the acquirer by keeping you incentivized to stay, but protects you from being fired right after the acquisition and losing your unearned equity.

VI.

There is so much more discussed in the book than highlighted here, including a detailed examination of Term Sheets. For those interested in venture capital, it seems this is one of the oft-recommended books, and I enjoyed reading it.

Subscribe now

I saw someone online the other day who claimed to be a “polymath.” And of course, the last person who would ever use that word to describe themselves is a genuine one—at least to the extent that such a thing is even possible these days. It struck me as a kind of Dunning–Kruger variant, or maybe just the plain old vanilla version.

It reminded me of when I first began reading fine books. When I came to the United States, and suddenly buying books wasn’t a luxury, I realized I couldn’t even come up with a list of ten really interesting titles to read.

Today, if you locked me in a room for five years and threw away the key, I wonder if I will be able to finish the books on my reading list, including those in my anti-library (see below).

There’s a certain irony to all of this. The more one learns, about the nature of things, about how the world works in its elastic, interwoven way, the clearer it becomes that the well is too deep. Knowledge doesn’t expand into mastery; there is a sense in which it dissolves into mystery. The horizon keeps receding.

Related: On Antilibrary

Nassim Taleb via Black Swan:

“The library should contain as much of what you do not know as your financial means … allow you to put there. You will accumulate more knowledge and more books as you grow older, and the growing number of unread books on the shelves will look at you menacingly. Indeed, the more you know, the larger the rows of unread books. Let us call this collection of unread books an Antilibrary.”

Subscribe now

There was once talk about AI as a leveler. I’m not sure how common that claim is anymore, but I’ve always thought it was off the mark. (I recently heard the idea parroted on some podcasts lately prompting me to write this.)

Think of intelligence in two forms: “raw” and “cooked.” Raw intelligence is what an AI model embodies, sheer pattern recognition at scale. Cooked intelligence, however, is different. It’s the lived context, the subtle navigation of a messy world.

AI struggles with that second kind. It doesn’t arrive with context baked in. Over time, context can be layered on, through careful prompt design, by building agents, or by extending memory systems (the kind I’m certain Meta is betting on, and OpenAI is aggressively implementing). Even then, the problem isn’t solved.

Because context is only half the story. What about the quality of your questions? What about the blind spots (what you don’t know that you don’t know?) Those edges define the limits of any intelligence, human or machine. And they’re exactly where the myth of AI as a leveler collapses.

There’s another dimension to this. It’s true that in many domains of knowledge work — particularly in task-centric contexts — the gap between expert and beginner has narrowed, and will continue to be. Yet one could also speak of a “greener field” hypothesis, where precisely because everyone has gotten a leg up, the advanced knowledge worker now holds the magic wand, the ability to do what was previously unthinkable (my go to analogy here is often the parable of the talents in the gospels).

But the real fracture appears on the economic front (which flows directly from the prior argument). This is where the “leveler” myth truly shatters.

AI is less a leveler and more a lever, dramatically amplifying the force of those who already know where to push (capital, know-how, etc.). The technology creates a dynamic where markets disproportionately reward those with sophisticated direction (and those who invest in them.)

A small class will thus guide AI toward novel, high-value creation, effectively setting the agenda for entire industries. The majority will find themselves executing more commoditized tasks, using powerful tools they don’t control, to fulfill objectives set by others.

This hierarchy will become self-reinforcing, as leading players reinvest their outsized gains into better models and proprietary data, solidifying their advantage and making the upper rungs of the economic ladder increasingly hard to reach.

There is precedent for this, of course, as such, I am unsure if many people, or anyone for that matter, would disagree with what we all think is coming.

Subscribe now

My Other Publications:

I have recently started a new series on my AI blog called, “What I Listened to Last week” (WILLs), see my latest edition; my latest AI newsletter.

My latest academic work, using neural topic modeling and LLMs to map the landscape of a biomedical research field, was published in Analytical Chemistry in July.

I haven’t published a video essay in a while, but you can check my latest one here, it’s largely a philosophical commentary on Alan Lightman book, the Miraculous from the Material.

(For books I have read, reviewed, or currently reading, go to the end of the newsletter)

[Talking Points]

🎮GTA VI Trailer

👶4-year-old sings Hallelujah

💵Bribe collection by country

🦄There are now 498 AI unicorns.

🇳🇬Nigeria vs Ghana 2002 AFCON.

🚖Driverless taxi increase in California.

🗞Americans are no longer paying for news.

💻Key differences between Javascript and Python

💦Two billion people don’t have safe drinking water.

💵The U.S. added 1,000 new millionaires a day in 2024.

🤲Migrants’ remittances almost triple global foreign aid.

🚘More than 1 in 5 new cars sold worldwide last year was electric.

🇪🇺100 years ago, ~½ of today’s independent countries were European colonies.

[Longer Reads/Watch]

[I] 👦🏿Can You Still Be Human?

In his thought-provoking piece “Can you still be human?”, sounds the alarm on the unprecedented threat AI poses to human dignity, freedom, and agency. He argues that technological advances—particularly in AI, surveillance, and monitoring—are being rapidly adopted under the guise of progress and convenience, yet risk enabling totalitarian control by governments and corporations, eroding civil liberties and reducing humans to mere cogs in a machine. McGilchrist warns against complacency and fatalism, urging readers to maintain skepticism, question the relentless march of technology, and safeguard what makes us truly human. In a world increasingly shaped by algorithms and bureaucracy, he calls for pushback: “The opposite of life is not death, but the machine. You are not a machine.”

My recent worry isn’t unlike some of Iain’s. I worry that how we treat humans in the world that appears to be looming will hinge squarely on a metaphysics of human nature. And it is no news that there are terrible ideas out there that undermine human dignity because of poor or plain wrong metaphysics, precisely one that could be even more vicious in a capitalist system where humans are essentially defined by their jobs, how much they earn, et cetera.

[II]: 🦯 How Hellen Keller Learned to Write.

In the essay I am writing at the moment, I am investigating people who lived in the most extreme of the human conditions, in terms of their senses, memory, or just plain hardship, and using their lives to stress test a philosophical conception of what it means to be a person. Long story short, I find myself reading this essay about Helen Keller.

In her essay, Cynthia Ozick (now 97 years old) examines the life of Helen Keller not simply as an inspirational figure, but as an artist who constantly battled accusations of inauthenticity. Ozick recounts how the brilliant and resilient teacher Annie Sullivan unlocked language for the young, feral Keller. This breakthrough, however, led to immense public scrutiny. At just eleven years old, Keller was accused of plagiarism over a short story, an event that Ozick presents as the first of many challenges to the legitimacy of her experience. Throughout her life, critics would dismiss Keller as a “living lie,” arguing that her use of visual language for a world she could not see or hear was a form of “verbalism”, parroting concepts rather than expressing genuine perception.

Ozick compellingly argues that these critiques misunderstand the very nature of imagination and art. Keller’s defense, which Ozick champions, was that her rich interior world was built from language and that “the bulk of the world’s knowledge is an imaginary construction.” Keller contended that her ability to describe a “flash of a swordblade” or “rainbow waters” was an act of the mind’s eye, the same faculty used by poets and writers throughout history. Ozick reframes Keller’s story from one of disability to one of profound artistic creation, asserting that Keller’s life serves as the ultimate proof that imagination is not a substitute for sight, but a powerful form of vision in itself.

[III]: 🔥The Hell House of Martinez

Occasionally I watch Jimmy Akin’s mysterious world, and I typically really enjoy listening in on his pod. One of the recent episodes I watched is the so-called “Hell House of Martinez”.

SPOILER ALERT!

He explores a case investigated by paranormal expert Lloyd Auerbach. In the 1980s, a family in Martinez, California, reported a series of strange phenomena in their home, including dizzy spells, noxious odors, flashes of light, and fireballs. The family believed the house was haunted, but Auerbach’s thorough investigation uncovered natural explanations for all the events. He discovered that the dizziness was caused by structural issues and low-frequency vibrations from nearby power lines, while the flashes and fireballs were the result of high static electricity interacting with methane gas from a nearby landfill. Auerbach’s investigation not only provided rational explanations but also helped the family get out of their lease and receive their money back due to the unsafe living conditions discovered during the investigation.

The episode also highlights the ‘paradoxical’ importance of paranormal experts in investigating issues like this because the family had been deflected away by psychologists who largely assumed that they were nuts, when in fact there was something there. A flesh and blood, paranormal expert in this case walked inside the building to essentially investigate what’s going on.

[IV]:🔫Lead and Serial Killers

An amazing podcast here on “lead‑crime hypothesis.”

The lead–crime hypothesis is the idea that widespread exposure to lead, especially from sources like leaded gasoline, paint, and industrial pollution, impairs brain development in children, particularly in areas related to impulse control, decision-making, and aggression. This neurological damage, which can be permanent, is thought to increase the likelihood of violent and antisocial behavior later in life. Studies have shown that trends in violent crime often lag about 20 years behind peaks in lead exposure, suggesting a generational effect: as lead levels in the environment declined after regulations in the 1970s–1990s, crime rates also fell. Book discussed: Murderland.

[V]: 🤳🏿How Social Media Shortens Your Life

unpacks how social media is quietly robbing us of our most precious asset, time, not only by swallowing hours in endless scrolls but by reshaping our very perception of time itself. Drawing parallels with the maze-like designs of casinos, Gurwinder wrote about how platforms engineer “curvilinear mazes” to keep us distracted, passive, and unable to form strong memories—shortening both our experienced and biological lives.

Social media’s endless, story-less feeds make days blur, disrupt sleep, accelerate aging, and heighten anxiety, but, I suppose there is hope: by intentionally seeking novelty, storytelling, and mindful habits, we can slow our rush through life and reclaim meaningful moments before they slip away. He reminds us that living longer isn’t just about more days, but fuller ones.

I haven’t read many essays on social media that are this well written.

Subscribe now

[VI] 🗽David Brooks on Audacity, AI, and the American Psyche

Interesting conversation between David Brooks and Tyler Cowen.

[VII] 🏃🏿 Easy Habits to Make You Run Faster

I have been running a lot more lately. As I am writing this sentence in late August, I have run over 100 kilometers. So, I have been all crazed about how to run faster, or better.

In this video, a former elite triathlete Nicklas Røssner breaks down 12 actionable habits that can transform your running game and help you run faster. He covers everything from setting achievable goals and building mental toughness, to smart shoe rotation, self-efficacy, and the power of consistent training. He highlights often-overlooked secrets like training your gut for nutrition, including strength workouts, investing in high-quality gear, and embracing rest as a tool for longevity.

[VIII] 🧗🏿Free Solo

Nice podcast, and super inspiring documentary

https://films.nationalgeographic.com/free-solo

[Books/Papers 📚]

Book(s) recently completed:

Deep Work [Review]

What I am currently reading:

Ultralearning: Master Hard Skills [Link]

Reshuffle: Who wins when AI restacks the knowledge economy [Link]

There is a sense in which by saying the things we are supposed to show, we are professing that we have lost the plot, either as an individual, or more importantly, as a culture. This applies specially if you can detect a tautological motif, or a kind of forced appendage in language use. It is worth noting that they can be hard to see as many are already normalized; and the normalization is really the point here.

For example, the obligatory use of the word “prompt” after a time is clearly stated, in some communities and culture goes a long way to show a broken culture of timeliness, or take deep work (I use this often myself), but what is wrong with just work? The modifier clearly betrays a landscape saturated with too much noise.

The pattern is everywhere once you tune your ear. Some can speak of “authentic cuisine.” Job ads promise “true ownership.” Even the truth sometimes can be insufficient, when one speaks of “honest truth.” I think social media birthed “real connection.” There is passion, and then there is “true passion.” And then there are the “active listening” workshops all over the place. And here is my ‘favorite’: “lived experience.”

Each redundant label is a linguistic tell, a nervous tick revealing the anxiety of its age. When substance fades, language rushes in to erect elaborate signposts, frantically assuring us that the thing still stands. But the very insistence is the confession.

And the normalization, like I intimated earlier, is really the key point, as it signals the silent triumph of the change.

The more syllables we stack, the clearer it becomes that we are circling an empty center, or that something is simply not right, trying to name back into existence what we can no longer, or struggling to, embody.

Subscribe now

I.

I first read Deep Work in 2016, at the start of grad school, and it remained one of the most impactful “productivity” books I read during that period. I decided to revisit it now as I prepare to embark on another major project, mainly to set the right ambience, the kind such projects demand. I’m well aware of the common critiques of books like Deep Work, that they “could as well be just a long blog post.” But such takes often miss the point. Sometimes you read not for novelty, but to create a particular atmosphere, a cognitive aftertaste if you will, that hopefully sets you straight. And that effect isn’t necessarily summoned by reading 600 bullet points of cold facts that you already know, but by stories woven carefully around those salient ideas.

II.

Cal Newport argues that integrating deep work into a professional life requires adopting a specific scheduling philosophy tailored to one's circumstances. He presents a spectrum of four distinct approaches. The most extreme is the Monastic philosophy, which involves minimizing shallow obligations to the barest minimum to maximize deep focus, as practiced by figures like Donald Knuth. It's suited for individuals whose work value is tied to a single, well-defined goal, and are privileged enough to do without the social media ‘networking’, and other things of sorts, that mere mortals like us have to endure to get our ideas out. A more balanced option is the Bimodal philosophy, where time is divided between long, intense, monastic-style deep work sessions and periods left open for everything. For those needing consistency, the Rhythmic philosophy turns deep work into a regular habit, removing the friction of deciding when to focus. Lastly, the Journalistic philosophy, he says, involves fitting deep work into any available free moment, a method best suited for those adept at quickly shifting their mental state and not for deep work novices.

For most serious knowledge workers, a rhythmic strategy might be the most reasonable, for example my recent deep work scheduling is between 5:30am and 11:30am, and getting rid of your smart phone during this period can only do you good.

III.

He also adapted ideas from the book titled “four disciplines of execution” as a framework for making deep work stick. The first is focusing on the wildly important, choosing one or two ambitious goals rather than vaguely promising yourself you’ll “work more deeply.” The second is acting on lead measures: instead of tracking only the final outcome, like a published paper or book, you track the hours you actually spend in deep concentration working towards that goal. The third is keeping a compelling scoreboard, something visible and concrete, even a simple tally card, so that progress feels tangible rather than abstract. And the fourth is creating a cadence of accountability, a regular rhythm of looking back at what worked, what didn’t, and what you’ll commit to next.

This stuff is really powerful stuff if you get to actually act them out. I have implemented a version of it, even long before reading the book.

IV.

In the chapter Embrace Boredom, Newport reminds us that the ability to concentrate is not an inborn gift but a skill, and like any skill it must be trained. The training, however, runs against the grain of our distracted age. We don’t build focus by taking occasional “internet sabbaths” while remaining tethered to distraction the rest of the week. Instead, we train it by scheduling our distraction deliberately, and refusing it outside those pre-allotted windows. Over time, the mind learns it can be content without constant novelty.

On this matter of training the mind, one of the most helpful (yet least used) subscriptions I have is Freedom, and the irony isn’t lost on me. Whenever I fall into a rut, say, binging politics new on YouTube or (doom) scrolling tech news on X, I only need to be strict with myself for about a week, using Freedom. After that, I rarely need it again.

He also offers a few practical drills. One is the “Roosevelt Dash”: set an artificially tight deadline for a meaningful task, then attack it with full intensity. The compression forces an almost athletic level of focus. Another is “productive meditation”: using walks, commutes, or other moments of physical routine to circle a single well-defined problem, returning to it again and again when the mind drifts. Such exercises don’t just improve concentration; they often yield insights that surface only under sustained attention.

The larger point is clear. Focus is a muscle, and muscles grow through deliberate strain. If you don’t practice being bored, it’s hard to develop the depth of attention that serious work requires.

In addition to books, I’ve been thinking about bringing radios back into my late-night routine. I’ve mostly cut off social media after 6 p.m., the damage those late scrolls do at night is hard to ignore, even when you start with good intentions. And don’t get me started on YouTube. I had to abandon my feeds altogether. These days, for the most part, I use an AI browser and LLM workflows to surface the specific content I want, where I want them such as to my email inbox.

V.

Now, talking about social media.

Newport’s advice is bracing: quit social media. However, to grasp just how successful social media platforms have been at capturing attention, you need only note some writers he cites (e.g., Malcolm Gladwell, J.K. Rowling) who once proudly avoided them. I suppose things has changed since then. There is a sense in which once enough of us jumped on the train, it’s compels others to join, even if the train is net bad business for many involved; and there is plenty of argument to be made on that front.

Anyways, there’s value in his call to quit (or at least sharply minimize) our use of social media. He critiques the common “any-benefit” mindset, where even the slightest upside is enough to justify adopting a new tool. The result is a cluttered digital life that fragments attention and weakens our ability to do meaningful work. His alternative, the Craftsman Approach to Tool Selection, is simple but demanding: identify the core factors that truly drive success and happiness in your life, and only use a tool if its benefits clearly outweigh its costs. Most platforms fail that test, he argues, and I agree.

To put this into practice, he suggests a 30-day experiment: quietly quit all non-essential social media, then afterward ask yourself two questions—would the month have been better if I had used it, and did anyone notice I was gone? If the answer to both is no, the verdict is clear.

Subscribe now

VI.

One of the most striking ideas in the book is fixed-schedule productivity. The practice is simple: set a hard, non-negotiable end time for your workday and refuse to work beyond it. At first glance, it feels like an arbitrary constraint, but that’s precisely the point. By creating scarcity, you are forced to strip away shallow tasks, organize your priorities, and channel energy toward what actually matters.

Abundance is good, but it often leads to wastage, and when it does, it’s probably not a good thing.

Immortal Souls by Ed Feser

I recently read AI 2027, an impressively detailed, heavily foot-noted, month-to-month AI scenario projection (You can read my review here), and I left with more questions than when I started. Long story short, while reading the scenario, I couldn't help myself but to do all these pondering in the background (e.g., how does the projected AI agency differ from human agency; what is really agency; what does it mean for a substance or a thing to reason? etc.)

Then I became smart enough to quickly realize my knowledge of philosophy of mind & metaphysics of human nature is indeed poor. You see, I have had Immortal Souls (by the brilliant Aristotelian-Thomistic philosopher, Ed Feser) on my to-read spreadsheet since it was released last year; and this was just a perfect excuse to read it. All it took was reading the aforementioned scenario, $35 and a couple days wait period for delivery.

It turned out to be a very detailed book, far more detailed than I have envisaged, but unsurprising given the subject matter. I don’t have a philosophy background, but my guess is that the book could be a legit textbook for higher level undergrad (or graduate level) class on the metaphysics of human nature – it’s that detailed and long, standing over 500 pages.

So, I have decided to focus solely on the chapters on the philosophy of mind (What is Mind?), and relevant chapters on machine learning/computers. I will most likely be writing about the subject matter either on this blog, or my AI blog.

The Master and His Emissary by Iain McGilchrist

One book that has been taking me forever to read, partly because it’s a door stopper, partly because I can’t help myself from being distracted with other less dense books, is the masterly work by Iain McGilchrist, The Mater and His Emissary. In brief, the book explores how the brain’s two hemispheres (left and right) differ in how they perceive and relate to the world, and argues that Western culture has become dangerously dominated by the narrow, analytic worldview of the left hemisphere, neglecting the more holistic, contextual understanding of the right.

I have added it to my list of books I am reading this summer hoping to knock out more than a few pages and at least complete the first part of the book.

So far, one of my favorite bits in the book: “The kind of attention we bring to bear on the world changes the nature of the world we attend to, the very nature of the world in which those 'functions' would be carried out, and in which those things' would exist.”

Subscribe now

All Things Wise and Wonderful by James Herriot

Admittedly, the two books above are quite dense, and sometimes after a protracted run you get bored from reading them, so I thought I would add something light that will allow me to continue reading, hence All Things Wise and Wonderful, the third installment in the All Creatures Great and Small Series.

My introduction to the book series came in the form of this nice PBS series about the book, which I highly recommend. After watching a few episodes, I immediately ordered the entire book series. I have read Book 1: All Creatures Great and Small, and Book 2: All Things Bright and Beautiful. Leaving me with Book 3.

And what is it about the book? It’s a memoir-style book that recounts the humorous, heartwarming, and often poignant experiences of a country veterinarian in rural England, capturing the richness of life through stories of animals, farmers, and the countryside during and after World War II.

Papers

some papers I plan to read in-depth.

• The AI Scientist-v2: Workshop-Level Automated Scientific Discovery via Agentic Tree Search [Link] [Code]

• Code Scientist: End-to-End Semi-automated Scientific Discovery with Code-based Experimentation [Link] [Blog]

• Towards an AI co-scientist [Link] [Hugging Face]

• And at least a few papers on LLM interpretability.

Video Essay:

Text:

I.

When I was planning to visit Harvard to give a talk last fall, I looked up events I could attend while in Boston. I saw that Alan Lightman was going to give a talk at the Harvard Science Book Talk about his new book, and I quickly reserved a spot to attend. I hadn't read any of Lightman’s book at the time, but of course, I had heard a lot about him. I believe I first heard about him about a decade ago when, spurred by my intellectual interests, I began researching professors who had modeled their careers to encompass a wide variety of pursuits. He was the first professor at MIT to receive a joint appointment in both the sciences and the humanities. I recalled reading about him and reassured myself that I wasn’t that crazy after all (or may be I was).

When I got to the venue of the talk, he sat beside me outside the lecture hall for a while, I didn’t realize he was the one until someone yelled “Prof Lightman” or something to that effect. After the talk, I bought a signed copy of the book.

Alan Lightman's "The Miraculous from the Material" is a captivating exploration of the wonders of the natural world, viewed through the lens of a self-proclaimed "spiritual materialist" (I will come to this materialist bit in a minute). In this book, Lightman examines various natural phenomena, from the mundane to the spectacular (mostly spectacular than mudane), and elucidates how the material world, with its rules and laws, is capable of extraordinary displays of beauty and wonder.

Lightman's journey as a materialist, he says, began in his youth, where he meticulously constructed a laboratory in his bedroom closet. It was there that he embarked on his investigations of the material world, building pendulums and even a light-borne communication device. Through these experiments, he discovered that the physical world, or at least his little corner of it, obeyed reliable, logical, quantitative laws.

Lightman's fascination with the material world did not diminish his awe and amazement of natural phenomena. From the microscopic world of Paramecia to the rings of Saturn, Lightman found beauty and wonder in the material world and its laws.

One of the parts of the book I enjoyed the most is the way Lightman weaves together science and personal experiences. He recounts his first encounter with bioluminescence while returning to his island house in Maine, where he noticed the wake of his boat glowing. He writes about the science behind this phenomenon, explaining how it is caused by a chemical reaction in organisms like fireflies and algae.

Lightman's fascination with the natural world extends to the intricate structures created by living organisms. He writes about the spider web, considering it the most beautiful structure created by any living being other than Homo sapiens. Spiders utilize different types of silk for various purposes, including trapping prey and wrapping victims in a deathly cocoon-like structure. Lightman also elucidates the remarkable ability of spider webs to spring out and capture flying insects, a feat achieved through electrical attraction.

His curiosity extends to the skies, where he explores the physics of bird flight. He explains how birds generate lift through the shape of their wings and their forward motion, enabling them to seemingly defy gravity. There is also this intriguing phenomenon of birds flying in V formations, revealing how each bird, except the leader, conserves energy by taking advantage of the uplifting pockets of air produced by the bird in front.

I wish I had read books like this when I was much younger, it would have saved me the horror of memorizing some of my text books. Lightman writes so well about things you might otherwise think are impossible to explain. I probably wouldn’t have read the book if I hadn’t attended the talk, so I’m glad I did.

Subscribe now

II.

In addition to a general review of the book, I'd like to add a few commentaries. While this book is not about philosophy (I should state this very clearly), a few important, opinionated statements were made that I think warrant a full-fledged philosophical commentary. And those statements essentially can be summarized in what Lightman called spiritual materialism, which seeks to harmonize transcendent experiences with materialism.

I believe this framework faces a foundational challenge from Alvin Plantinga’s Evolutionary Argument Against Naturalism. Plantinga argues that if naturalism and evolution are true, our cognitive faculties—including those generating spiritual feelings—evolved for survival, not truth, rendering even materialist explanations of spirituality epistemically unreliable. For instance, sensations of cosmic connection or awe might be adaptive illusions fostering social cohesion or resilience, not genuine insights into reality. This creates a paradox I think: Lightman’s framework depends on neuroscience and evolution to explain spirituality, yet naturalism undermines trust in all cognitive outputs, including the scientific reasoning used to validate such spiritual materialism. And one could respond and say that spiritual experiences require no truth-claims beyond their material origins, sidestepping the argument’s focus on belief reliability.

But there is more to be said on this topic: the mysterious alignment between abstract mathematical truths and physical reality (as seen in Eugene Wigner’s popular paper: “unreasonable effectiveness of mathematics”) poses a dual challenge to materialism. Materialism struggles to explain why accidental cosmic processes would conform to elegant, discoverable mathematical laws or why survival-driven cognitive faculties could reliably uncover these laws. The latter intersects with Plantinga’s Evolutionary Argument Against Naturalism: if naturalism undermines trust in cognition’s truth-aimedness, even science’s reliance on mathematics becomes suspect, as evolved brains might prioritize adaptive fictions over truth. Thus, materialism cannot coherently account for mathematics’ predictive power or the reliability of the minds that wield it.

And there is this scientific antirealism angle too but it’s doesn’t look like a crisp argument to me, but I will still flesh it out anyways. When I first read about scientific anti-realism, I didn't quite get it. But now I'm beginning to think that I might be able to make sense of scientific anti-realism on the grounds of materialism. In brief, (some) scientific antirealist challenges the notion that there are mind-independent truths about reality. They suggest that scientific theories are primarily instrumental tools that help us predict, explain and control stuff, they do not necessarily provide a true picture of how the world actually is. Granting the Evolutionary Argument Against Naturalism - and I am now back to my initial point - it appears to me that materialism paradoxically supports antirealism: if our minds are products of blind evolutionary processes, we lack justification for believing they can grasp fundamental truths about the universe.

We can also go through the route of accounting for information in the evolutionary landscape: Darwinian accounts of how complex biological information arises rely on a "displacement fallacy." In brief, instead of explaining where the information originally comes from, the Darwinian account simply shift the burden onto natural selection or specially designed fitness landscapes, effectively assuming the very information they're supposed to explain.

I suppose this will do for now. I should also add that my philosophical commentary should not be taken as a response to Professor Lightman's complete philosophical framework, as I have not read him beyond this book. It's just a commentary to a few comments that actually popped at me while reading this particular book.

In summary, The Miraculous from the Material is an enjoyable read that invites us to marvel at the wonders of the natural world regardless of our philosophical leanings. That being said, if I were to give the book a title, I will keep things as it is, except to swap the positions of “Material” and “Miraculous.”

My Other Publications:

Check out my latest AI newsletter: ε Pulse: Issue #24

The Epsilon

ε Pulse: Issue #24

Read more

a year ago · Olatomiwa Bifarin

Watch my latest video essays on my Youtube Channel:

The Story in Your Head

Pillars of Modern Civilization

What Are You Hiding?

Working in The Storm

(For books I have read, reviewed, or currently read, go to the end of the newsletter)

[Talking Points]

🏃🏿David Goggins X Israel Adesanya Training.

🥩Ketogenic Diet in Pancreatic Cancer Treatment.

⌨️Cursor Writes 1 Billion Lines of Accepted Code/Day

🎶Nigerian Students Sing the UEFA Champions league Anthem

🏠US: Homebuyers Need $114K Salary for Median-Priced Homes

🌍Sub Saharan Africa Has 67% of People Living in Extreme Poverty

🧑In 2025, If You Are Over 30, You Are Older than Most People in the World.

[Longer Reads/Watch]

[I]: 🔰 Missing Pages: An AI Assistant for Avid Readers.

Back story to an app I am currently working on with a friend called Missing Pages.

Here is the waitlist.

[II]: 💻 A.I. Is Changing How Silicon Valley Builds Start-Ups

The evolution is underway: more with less with AI. Here is a New York Times profile of Gamma, an AI design partner for presentations, websites, etc.

The Cofounder Grant Lee wrote about it on X:

The New York Times profiled a start-up with 28 employees serving nearly 50 million users. That company is us. The traditional startup playbook: raise massive funding, hire hundreds of employees, and worry about profitability "later." But there's another way. Everyone at Gamma could fit in a small restaurant. We're not just surviving—we've been profitable for 15+ consecutive months, with revenue growing month over month, and lifetime negative net burn (we have more money in the bank than we've raised). This isn't an accident. We've deliberately designed our organization to maximize impact per person. Instead of creating specialist silos, we hire versatile generalists who can solve problems across domains. Rather than building management hierarchies, we find player-coaches who both lead and execute. Our team leverages AI tools throughout our workflow - Claude for data analysis, Cursor for coding efficiency, NotebookLM for customer research synthesis. These aren't just productivity hacks; they're force multipliers. Examples: — When our growth PM needed better analytics, he didn't file a ticket with a data team—he built a self-serve system that anyone can use without SQL knowledge. — When our marketing lead needed to understand our customers better, she fed thousands of interactions into an LLM and created actionable personas that now guide our entire strategy. — When our design team needs to test a hypothesis, we create a rapid prototype and show it to our power users. What we're seeing isn't just about "doing more with less." It's about fundamentally changing what's possible per person. The most valuable employees aren't specialists who excel in narrow domains - they're resourceful problem-solvers who continuously expand their capabilities. This approach creates remarkable resilience. Since everyone understands multiple functions, we don't have single points of failure when someone leaves or moves to another project. If you're building today, the question isn't how quickly you can scale headcount … it's how much impact you can create with the smallest possible team. The future belongs to tiny teams of extraordinary people.

Subscribe now

[III]: 🇻🇦The Next Pope

The first pope I knew was Pope John Paul II, but I wasn't particularly aware of him because I was quite young so “knew” is an exaggeration. I do recall his death being major news, and I vaguely remember the talk of an African Pope over the radio in my dad’s car, maybe Cardinal Arinze? While I wasn’t particularly interested in Catholicism (and hence the papacy) then, my appreciation for Catholicism grew after reading some of Pope Benedict's books such as this, this, & this; supplemented with plenty podcast. Now, because of this sequence of admittedly abbreviated events, I've become too interested in who becomes pope, and I have read perhaps an unhealthy quantity of material about who is who, and the goings-on. So here are a few. The best resource is this fantastic web page on the College of Cardinals. It provides in-depth profiles of Catholic cardinals-especially those considered potential future popes (“papabili”) - to help folks better understand their backgrounds, beliefs, and positions on key issues. The site features comprehensive biographical data, summaries of each cardinal’s qualities, detailed analysis of their work as bishops. It also includes a fairly detailed piece on the origin of cardinals. Another one on the role of present-day cardinals (which clarified my confusion with regards emeritus, curial, & diocesian status within the cardinalate). The one I enjoyed the most was on electing the pope. Interesting bit from the article:

In the 1970s, Pope Paul VI introduced the age limit of 80 for electors, and 120 as the recommended maximum number of voting cardinals. For a man to be validly elected pope, he must receive two-thirds of the votes. Pope John Paul II changed the rules in 1996 so that after 33 or 34 ballots without a two-thirds majority, a candidate could be elected by a simple majority.

But Benedict XVI reinstated the requirement for a two-thirds majority to elect a pope, reversing John Paul II’s change which had been viewed as a “radical” innovation compared to the two-thirds rule that had existed since 1179. Benedict’s amendment stated that if a deadlock persisted after 13 days of voting, the two top candidates would enter a runoff. He did this to prevent a situation where a majority bloc could push through a candidate by simply holding out until the 34th ballot when there would have been a simple majority. It was also felt that the change would guarantee the greatest possible consensus among the cardinals rather than a candidate who could win with just a narrow majority. Benedict’s reversal was a return to the historical norm.

I have read a few profiles about some of the papabili, written by John Allen of Crux: Parolin, Zuppi, Erdo, and Besungu. At the time of writing it appears profiles of the following priests have been added: Pizzaballa, Aveline, Tagle, Prevost, and Arborelius. They are running this “Papabile of the day" till the conclave begins. On the world's largest prediction market, Polymarket (at the time of writing) Cardinal Parolin is leading the pack, closely followed by the charismatic Tagle of the Philippines. Also see Ed Feser on pope’s first duty, profiles of African Cardinal electors, and the youngest priest in the college of cardinals (45 year old Cardinal?!).

[IV]: 🌎Ten Charts That Show How the World is Changing

In his Digital Native essay, Rex Woodbury uses a curated set of charts to visually narrate the significant transformations shaping our world. Prominent economic and technological shifts are clearly visible. The tech market, for example, is shifting focus, now prioritizing profitability and efficiency over unrestrained growth, marking a significant change from previous years. Echoing this recalibration, venture capital distributions to investors have slowed, indicating a tighter funding and exit environment for startups and their backers. At the same time, a powerful trend is the rapid acceleration of AI adoption, which is driving significant increases in web traffic and engagement for AI tools, signaling a major technological disruption.

Beyond core tech and finance, the charts illuminate evolving patterns in commerce, culture, and society. While e-commerce continues its expansion, startups like DoorDash are shown dominance in their sector. Within the digital media landscape, success and engagement are increasingly concentrated among the largest platforms and most popular creators, following a distinct power law, which will get accelerated with continued AI adoption. (See my essay: Winners Take All). This digital environment highlights how viral cultural phenomena can create substantial and measurable boosts for brand visibility (this was discussed in the context of Guinness recent sales boost). Finally, the data underscores distinct generational economic realities, as Gen Z confronts unique hurdles compared to previous generations, starkly illustrated by trends like lower relative homeownership rates.

Share

[V]: 🦠Recipe for a Cell

What makes up a cell? I learnt many of these in college, and even more complicated stuff in grad school (most of my formal education is in biology), but this is refreshing stuff to read nevertheless. For example see this bit:

In general, however, the mass of an E. coli is about 1 picogram, or one one-trillionth of a gram. That’s astonishingly small, roughly equal to the weight of the DNA inside a single hummingbird cell.

Which molecules make up this mass?

About 70 percent is just water. The other 30 percent — the so-called “dry mass” — is composed of everything else: DNA, RNA, proteins, lipids, and ions. Surprisingly, DNA accounts for just 3 percent of this dry mass, yet encodes all the information needed for the cell to grow, divide, adapt, and evolve. Most of a cell’s dry mass (55 percent) is proteins, which dominate because they are the “executors” of cellular functions; they catalyze reactions, form structural scaffolds, and regulate gene expression. About 20 percent of the dry mass is RNA, 10 percent is lipids, and the rest is ions and signaling molecules.

…

building a functioning cell from scratch still remains one of biology’s most elusive goals. The challenge is not merely mechanical, either. Even if one gathers all the right molecules in the right quantities, arranging those molecules with sufficient precision to “create life” remains far beyond current capabilities.

This is, in part, because cells are dynamic. They move and change quickly. Sugar molecules fly through a cell at 250 miles per hour, and each protein is bombarded by 10¹³ water molecules every second. This dynamism is often overlooked in the methods used to study living cells. A list of ingredients is inherently static. The next step, then, is to capture how each piece moves and interacts. Only then might we turn our catalogs of parts into recipes for a cell.

Full essay.

[Books/Papers 📚]

Book(s) recently completed: [-]

What I am currently reading:

The Master and His Emissary [Link]

AI 2027 [Link]

Hands-on Guide to Fine-Tuning Large Language Models [Link] [Code]

The Mom Test [Link]

Administer Container in Azure [Link]

The Alchemist [Link]

This essay was previously published a few years ago. If you want to read the text, see the first version here.

Share

Link to Waitlist: https://tally.so/r/wgOJvM

Missing Pages is my latest app project— an app prototype to solve a personal habituated pain point.

To stop reading in a circle, find knowledge gaps, read on-demand, and dig in.

If you are interested in trying this out, here is the Waitlist: https://tally.so/r/wgOJvM

(h/t to Michael Judge for the apt name)

Subscribe now