There is a discipline called philosophy of physics. It sits at the intersection of science, mathematics, and philosophy proper, and it is uncomfortable precisely because the boundaries between those three things dissolve at that point. The physicist thinks they are doing science. The philosopher thinks they are doing conceptual analysis. The mathematician thinks they are doing mathematics. In practice, at the frontier, all three are doing something that none of their home disciplines can cleanly describe.

This interests me, not as a specialist (I am most certainly not one, as many PhDs have pointed out to me on LinkedIn), but as someone who thinks seriously about how we produce and evaluate knowledge. The question that keeps surfacing when you spend time in this territory is the testability question. And followed honestly, it leads somewhere uncomfortable.

The line between theory and faith

String theory is the canonical example. It began as an attempt to unify quantum mechanics and general relativity: two frameworks that each describe reality with extraordinary precision within their own domains, but which contradict each other at the scale where both should apply. String theory proposed that the contradiction resolves if the fundamental constituents of the universe are not point particles but one-dimensional vibrating strings - and that the mathematics only works in more dimensions than we can observe.

The initial formulation required 26 dimensions (as a reminder, we only know of 3. 4 if you include time). Later refinements brought that to ten or eleven. The theory has been revised repeatedly, not because new experimental evidence demanded revision, but because internal mathematical consistency required it.

And here is the uncomfortable part: no prediction from string theory has ever been tested. Not because physicists lack the will, but because the energy scales required to probe string-length physics are so far beyond anything achievable that testing a prediction would require something in the vicinity of a Big Bang. The theory is internally consistent, mathematically sophisticated, and empirically inert.

At what point does that stop being physics and start being something else?

My reading of the philosophy of physics literature is that it becomes an article of faith. Not in a dismissive sense. Faith in the sense that the theory is held primarily because of its internal coherence and aesthetic appeal, rather than because of evidence that the thing it describes actually exists. The people who work on string theory are not deluded. They are often genuinely brilliant. But the relationship between the theory and the world has become, at minimum, attenuated.

What is interesting, and what the philosophy of physics keeps pressing on, is the question of why someone develops strong commitment to a theory in the absence of evidence for it. The obvious answers are real: the mathematics is beautiful, the framework is generative, it is the dominant paradigm and careers have been built inside it. But underneath all of those, I suspect there is something more personal. An initial insight that felt true, and everything after that becomes an attempt to prove the initial intuition rather than to test it honestly. The ego of the founding idea persists through all subsequent revision.

This is not unique to physics. But physics is where it becomes most visible, because physics is the domain most committed to the idea that this kind of thing does not happen.

What mathematics can and cannot do

The reason this matters is what it implies about mathematics.

Mathematics has a particular authority in our epistemology. We treat mathematical proof as the highest form of certainty. And within mathematics itself, that is justified: if the axioms hold and the derivation is valid, the conclusion is necessary.

But mathematical consistency is not the same as truth about the physical world. A theory can be internally coherent, formally beautiful, and empirically disconnected. When that happens, mathematics is not serving as a tool for understanding the world. It is serving as an internal validation mechanism, confirming that the framework hangs together without engaging the question of whether the framework corresponds to anything real.

That’s what I mean by mathematics becoming a tool of fantasy. Not that it’s wrong or useless. But that it’s been decoupled from the thing it is supposed to illuminate. The map has been drawn with extraordinary precision and care. Whether it corresponds to the territory is a question the map cannot answer about itself.

The philosophy of physics discipline exists precisely because this situation keeps arising. Science, mathematics, and philosophy intermingling means the methods of each discipline alone are insufficient to evaluate what is happening. You need philosophical analysis to distinguish between “this theory is mathematically consistent” and “this theory describes the physical world.” You need scientific methodology to insist that the latter requires evidence. And you need mathematics to work out what the theories actually say.

When these disciplines stop talking to each other, things that function as articles of faith get treated as empirical science, with all the authority science implies.

Share

Testability as the honest constraint

The current most critical question for any unified theory of physics is whether it can either quantise gravity or gravitise quantum mechanics: that is, whether it can bring the two frameworks into a single, consistent, testable description of reality. Currently, neither general relativity nor quantum mechanics can do this. They each work in their own domain; they contradict each other at the boundary.

A theory that resolves this contradiction without producing any testable predictions is not a scientific achievement. It is a demonstration of mathematical ingenuity. That is still valuable. But it should be categorised honestly.

Testability isn’t an arbitrary philosophical demand. It’s what distinguishes a claim about the world from a claim about a formal system. If there is no possible observation that could, even in principle, disconfirm a theory, the theory is not in competition with reality. It’s coexisting with it.

String theory’s defenders sometimes argue that the energy scales required for testing are simply beyond current technology, and that this does not make the theory unscientific. This is partially right: falsifiability in principle is the standard, not falsifiability in practice. But, “in principle, we could test this if we had energy equivalent to a Big Bang”, is doing a lot of work in that argument.

The more honest position is this: what we have is frontier mathematics with physical motivation. It might eventually produce a testable prediction. Until it does, the appropriate posture is not belief or disbelief but sustained scepticism, held alongside genuine interest.

Why this matters beyond physics

Every domain has versions of this problem. Frameworks that become self-confirming, that generate internal complexity without external traction, that attract commitment because of their elegance rather than their accuracy. Philosophy of physics is one explicit attempt to maintain the distinction between what the mathematics says and what the world is like. But the underlying failure mode is not specific to physics.

What strikes me is that the resolution is not to abandon ambitious theory. It is to be honest about what category of thing you are doing at any given point. Exploring the mathematical implications of an idea is a different activity from testing a hypothesis about reality. Both are valuable. Confusing them is where the problems start.

The most defensible position on string theory is not dismissal and not commitment. It is curiosity held alongside a clear recognition of what kind of evidence would actually change anything.

Until then: interesting mathematics. Not yet physics.

If this piece added something to your week, please consider subscribing :)

Subscribe now

There’s a thing that happens when you’re standing somewhere genuinely stunning, some landscape that stops you mid-step, and you reach for your phone to take a picture. You frame it, you tap the shutter, you look at the result, and it’s just... not that. The picture might even be good. But it doesn’t look anything like what you’re seeing. There’s something missing, something the camera can’t hold.

I’ve been turning that feeling over for years now, because I think it’s pointing at something much larger than photography. It’s pointing at a fundamental limitation of digital mediation, one that runs through every platform we’ve built to connect people at scale, and one we keep pretending doesn’t exist.

The compression problem

Digital, by definition, computes human expression. It takes something continuous, something alive with infinite variables, and it reduces it to something a machine can process. It always takes something away. It’s always going to compress something. It’s always going to remove something.

This isn’t a bug. It’s the mechanism. It’s how the thing works.

Consider live music versus a record. A live performance has something a recording can never have, because the record is going to play the same way every single time. There’s no room for expression. No room for any finite variable, any time-bound variable to enter the equation. The performer’s mood, the acoustics of the room, the cough from the third row that shifts the timing of a phrase by half a second: all of that’s gone. The record is perfect in the way that only dead things are perfect. It’s fixed. And in being fixed, it’s lost something essential.

Now scale that same principle to human relationships.

What gets lost in the flattening

The thing is, human relationships have always been mediated. That’s not new. They’ve been mediated by geography, by religious institutions, by family, workplaces, by mutual friends. But none of those mediators required the same sacrifice that digital ones do.

What makes digital mediation different is the mechanism. To process relationships at scale, algorithms have to flatten people into signals: engagement patterns, keyword matches, behavioural data. Multi-dimensional human beings get compressed into something software can handle. And what’s lost in that compression is far more than we tend to acknowledge.

Take the mutual friend as an example. If someone introduces you to a friend of theirs, that friend can account for nuance. They can say, “Look, he seems quiet at first, but give him twenty minutes and he’s the funniest person in the room.” Or, “She comes across as intense, but that’s because she actually cares about what you’re saying.” A mutual friend can account for the gap between how someone presents in a snapshot and who they actually are. They can account for consciousness.

An algorithm can’t do that. It can’t account for consciousness. It can process signals: how long someone looked at a profile, which photos got the most engagement, what keywords appear in a bio. But the distance between those signals and the actual texture of a human being is enormous. And that distance is where everything that matters about connection actually lives.

The neurological cost

If your brain acclimatises to swiping through TikTok or Instagram reels, to consuming human beings as ‘content’ at the rate of one every three seconds, how is it supposed to cope with the slowness of real relationships? The patience required. The ambiguity. The discomfort of sitting with someone and not knowing where a conversation is going.

There’s a parallel I keep coming back to, between social media’s effect on our neurology and the documented effects of pornography. Both involve the over-stimulation of reward systems through compressed, mediated versions of something that’s supposed to be experienced in full fidelity. Both create a gap between expectation and reality that makes the real thing feel somehow insufficient. And both operate through the same basic mechanism: take something complex, strip it down to its most stimulating signals, and deliver those signals at a frequency the brain wasn’t designed to cope with.

The infinite scroll is part of this. TikTok, Instagram, all of them: you never have the experience of closing the loop. There’s no moment where you’ve finished. No moment where the thing is done and you can set it down. It’s an unclosed loop, running constantly, and I think that hidden absence of completion is doing more damage to mental health than the content itself.

Creation doesn’t fix it either

Here’s what surprised me. I’ve gone from something like 100% content consumption to now roughly 90% content creation. And I don’t feel any better for it.

I thought the problem was passivity. I thought if I switched from consuming to creating, from scrolling to making, the flatness would lift. But it hasn’t, not fundamentally. Because the medium itself is the issue. Whether you’re consuming compressed human expression or producing it, you’re still operating within a system that can’t hold the full thing. Every time I log into Substack or LinkedIn, it’s just the same stuff. An ocean of content that’s been processed into the same shape, the same cadence, the same optimised structure.

The creation side is arguably worse in some ways, because it adds a layer of performance. You’re not just experiencing the compression; you’re actively compressing yourself. Flattening your own thinking into formats the algorithm can distribute.

The Instagram spot problem

There’s a telling behaviour around those famous Instagram spots, the locations that become popular specifically because they photograph well. People travel to them, but I’m not convinced they’re going to experience the place. They’re going to acquire the asset. The photo. The proof of having been there. It’s a kind of status collection, trophies on a digital shelf.

The experience itself, the actual standing-in-a-place-and-being-stunned-by-it part, is almost incidental. The platform has inverted the relationship between experience and documentation. The documentation is the point. The experience is just the means of production.

And this connects back to the compression problem, because what you end up with is a photograph that doesn’t capture what was there, shared on a platform that compresses it further, consumed by people scrolling past it in under a second. At every stage, something is stripped away. The landscape becomes a rectangle. The rectangle becomes a thumbnail. The thumbnail becomes a data point in an engagement metric.

What previous mediators preserved

Geography mediated relationships by limiting who you could meet, but it didn’t flatten the people you did meet. A church mediated relationships through shared ritual and moral framework, but it didn’t reduce its members to swipeable profiles. A workplace mediated relationships through proximity and shared purpose, but it preserved the full complexity of daily human interaction.

What’s distinctive about algorithmic mediation isn’t that it mediates. It’s what it has to destroy in order to do so. Previous mediators were constraints on access. Digital mediators are constraints on fidelity. They don’t limit who you can reach; they limit how much of a person can travel through the connection.

That’s a fundamentally different trade-off, and I don’t think we’ve properly understood what we’ve given up by making it.

The trust question

There’s a deeper question underneath all of this about whether trust itself can be computed. Blockchain tried to remove trust as a factor entirely, replacing it with verification - ‘Don’t trust, verify’. But I think that gets it backwards. What we actually want, fundamentally, is to encourage trust, (not least of all because trust and economic prosperity are positively correlated - more on this in a forthcoming article). The question is on what basis.

A mutual friend gives you a basis for trust that’s grounded in shared experience and accumulated knowledge of a person. A verified profile gives you a basis for trust that’s grounded in... what, exactly? That someone’s photos match their face? That their employment history checks out? These are signals, but they’re thin ones. They’re the data that survived the compression.

The thing I keep circling back to is that there’s no way to mediate human relationships at scale without flattening them. Without thinning them out. Without removing a lot of the things that are important for human connection. The furniture maker carving wood in a workshop, shaping something purely through consciousness and physical contact with the material, that’s the opposite end of the spectrum. That’s human expression utterly unmediated. And everything we build digitally moves further from it.

I don’t have a clean answer for what to do with this. I’m not sure there is one. But I think the first step is being honest about the trade-off: that every platform promising to connect us is, by the mechanics of how it works, also compressing us. And that compression has a disconnection cost we’re only beginning to measure.

If this piece added something to your week, please consider subscribing :)

Subscribe now

Most productivity software gets marketed on a single promise: more output, less effort. More posts. Faster drafts. Content in half the time. This framing treats output as the goal and the human as the chokepoint. The better the tool, the less the human has to do.

There’s an extent to which this makes sense. Removing genuinely unnecessary friction is good design. But there’s another viewpoint, and it’s where most productivity tooling can risk ending up: optimising output velocity by progressively removing the human from the cognitive process. Not just the administrative parts. The thinking parts.

The result is what we might call a human donkey. The user fills in fields, selects from menus, drags blocks into templates. The system handles the rest. Output is consistent, reliable, and increasingly divorced from anything the user actually thought.

The human is no longer creating. They are executing. Efficiently, reliably, and with diminishing understanding of what they are producing or why.

The drag-and-drop trap

Over-structured content tools are the most visible version of this. The logic is intuitive: give users a clear structure, reduce the open-ended ambiguity of a blank page, and they will produce more. Which is true. They will produce more.

What tends not to survive this process is creative ownership. When the structure is provided rather than built, when the user’s job is to fill rather than to form, the cognitive load shifts from the person to the system. That sounds like a good trade. But what gets relocated is not just the complexity of deciding how to structure a piece. It’s the process of working out what you actually think.

Anyone who has written something properly knows that the thinking and the writing are not separate activities. You do not think first and then write. You think by writing. Remove that process, and you do not have a faster version of the same outcome. You have a different outcome entirely, one where the author has produced content without developing understanding.

I’ve found it useful to distinguish between two things a tool can do: it can increase what you produce, or it can increase what you can think. Over-structured, drag-and-drop systems optimise aggressively for the former. The latter barely registers as a design goal.

Why this is hard to see

The disquieting part of this dynamic is that it tends to feel productive. The human donkey is not idle. Posts go out. Metrics accumulate. The dashboard looks healthy.

What is harder to measure is whether the person operating the system is developing anything. Whether they have a clearer sense, six months in, of what they think and why it is worth reading. Whether they could hold a conversation on their stated area of expertise that goes deeper than the surface their content has skimmed. Whether they are building fluency or merely generating volume.

These questions do not surface naturally when the output numbers are good. They require a deliberate decision to ask them.

And this is exactly the problem. The default success metric for a productivity tool is user-reported satisfaction, almost always correlated with output volume. “I doubled my content in two weeks” is a testimonial that moves product. “I have a clearer sense of what I think about my domain” is almost impossible to attribute to any specific tool and does not feature prominently in any growth playbook.

The incentives, in other words, are structurally aligned against the right outcome.

The commercial pressure

This is where the design ethics become genuinely complicated. A tool that converts users into efficient, low-agency executors is, in many cases, commercially successful. Retention is strong. The value proposition is legible. The comparison to a world without the tool is flattering.

The pressure to build toward this outcome is real, and it is not cynical. It is structural. Optimise for user-reported satisfaction, optimise for output metrics, and you eventually arrive at exactly the kind of tool that severs the user from the creative process. Not through bad intentions, but through the accumulated weight of incentive.

The guiding principle I keep returning to is this: the right test for any tool is not whether it increases what a person produces, but whether it increases what a person can think. These are not always the same thing. They can, in fact, run directly against each other.

A tool that holds your hand through every step of the content process may genuinely produce more output. But it is also training you, session by session, to rely on the scaffold rather than develop the capability. The productivity gain is real. The dependency that accumulates alongside it is also real.

Subscribe now

What extending thinking actually looks like

There’s a different category of tool that is harder to describe precisely because it does not optimise for one clean metric.

A tool that genuinely extends thinking creates friction in the right places. It prompts before it generates. It asks what you mean rather than assuming. It holds enough structure to be navigable, but enough open space to allow genuine discovery. When you are done with it, you have produced something, but you have also clarified something. The output and the thinking moved forward together.

Two pieces of content can look identical on the surface: one produced by someone who was thinking throughout the process and one produced by someone who was filling templates. What differs is what the author knows afterward, and what they are capable of the next time they sit down.

That second kind of difference does not feature in any dashboard. But it is the only one that compounds.

The line is fine, and it moves

I do not think this is a problem that can be solved once at the design stage and then set aside. The commercial pressure to drift toward high-efficiency output tooling is continuous. It does not announce itself.

It arrives as small decisions. Adding one more template option. Making one more step automatic. Removing one more moment of friction that users reported as a pain point. Each of these, individually, looks like an improvement. Cumulatively, they can shift a tool from one that extends human thinking to one that replaces it, without anyone having made that decision explicitly.

The only reliable response is to treat this as a guiding principle that needs to be surfaced actively and often. The purpose of the tool is to help people think and express themselves better. Not just more. Better. And each design decision should be tested against that honestly, with awareness that the commercial pressure will consistently pull in the opposite direction.

The line between a tool that extends human capability and one that silently replaces it may be fine. That is exactly why it requires sustained attention rather than a one-time commitment.

Where this leaves us

The human donkey problem is not a critique of any particular tool or platform. It is a description of a failure mode that most productivity software is structurally incentivised to drift toward over time.

Understanding that drift is the first condition for designing against it. And recognising it in tools you already use is the first condition for making deliberate choices about which kind of productivity you are actually after.

The goal worth pursuing is not a tool that makes humans more efficient. It is a tool that makes humans more capable. These are different ambitions. They lead to different products. And the distance between them is not always as obvious as it should be.

I’ve been working on just such a tool and I’m looking for early testers. If you’re interested, click below, I’d love to hear from you!

The dominant voice in the personal brand building world says: niche down, find your audience, build an offer.

I’ve been resistant to this for a long time, and I think I’ve finally worked out why it bothers me so much. It isn’t that the advice is wrong in some narrow, tactical sense. If you want to grow a LinkedIn following about B2B SaaS marketing, then yes, talk about B2B SaaS marketing every day. The algorithm will reward you. But the advice contains a hidden assumption that I think, for some people at least, is genuinely destructive: that you should know where you’re going before you start walking. I don’t know exactly where I’m going, and I’m starting to think that’s not a problem to solve. It’s the whole point.

Direction as a Product, Not an Input

The standard model treats direction as an input. You decide what you’re about, you build content around that identity, and then you attract an audience that validates the decision. It’s clean. It’s logical. It’s also completely backwards for some people.

What I’ve found is that direction isn’t something you choose and then execute on. It’s something that emerges from the act of thinking out loud over time. I want to have thoughts and talk about things, and those things then create the direction. The direction is a product of where my mind is already going, not some strategic decision made in advance.

When you niche down prematurely, you’re forcing yourself into a corridor before you’ve even explored the building. You’re committing to a path based on what you think you should be interested in, or what seems commercially viable, or what some growth strategist told you would work. And then you’re stuck creating content about something that might not actually matter to you, which is precisely how you end up as one of those people on LinkedIn whose posts read like they were written by someone who doesn’t really care about what they’re saying - because they don’t.

The Optionality Argument

Here’s what I think content actually does when you let it breathe: it creates optionality. Not obligation. Not commitment. Options.

There’s an unknown quantity of opportunities that you can’t even possibly imagine right now. Not in a vague, motivational-poster sense, but in a very practical one. You’re limited in your experience of the world, we all are. Particularly in a professional capacity. And so there are possibilities available to you that you literally can’t conceive of from where you’re currently standing, because you haven’t been exposed to the people, ideas, or contexts that would make those possibilities visible.

Content is how you expand that exposure. But only if you’re creating content that actually reflects where your thinking is, rather than content optimised around some predetermined niche that you chose six months ago and now feel trapped by.

I recently spoke to a friend who wants to create content and build their profile but the fear stopping them is that putting their thoughts and ideas out into the world will create situations they don’t want to be in. That if something resonates, they’ll be forced into conversations or opportunities that feel misaligned.

And whilst I understand the concern, the mistake is imagining that the totality of opportunities that would get created are ones you don’t want to participate in. In other words, they’re not opportunities for you. They’re threats. And that framing just isn’t true. You have agency at every stage of the process. Clicking “post” doesn’t set off a chain of dominoes that ends with you trapped in a meeting you hate. You can say no at any point. The content creates the options. You choose which ones to take.

Share

The Three-Year Thought Experiment

I keep coming back to this thought experiment, and it’s the one that makes the anti-niche case most clearly for me.

Even if today you’re commenting on people’s posts about AI agent swarms, in three years you might be commenting on someone’s posts about how to design a traditional pattern on top of sourdough bread as you bake it. There’s no rules. We can do whatever we want.

That sounds flippant, but I mean it seriously. The assumption behind niching down is that your interests today will be your interests in three years. That’s almost certainly not true. And if you’ve spent those three years building an audience and an identity around a specific niche, you’ve also built a cage. You’ve optimised yourself into a corner where the algorithm, the audience, and your own professional identity all resist the very evolution that makes a creative life worth living.

But if you’ve spent those three years talking about whatever genuinely interests you, following the threads of your own curiosity, the body of work you’ve built is something much more interesting. It’s a record of how your thinking evolved. It’s a demonstration of intellectual honesty. And it’s created a surface area for connection with people who share your actual values and interests, not just the ones you strategically ‘target’.

You might be up for an opportunity that presents itself in three years that you’re not up for today - that you couldn’t even imagine today. And in three years, you’ll have wished you’d been creating content and building audience and creating that domain authority and track record, so that when you do feel ready, you’re well positioned for whatever comes next.

Refinement Happens Naturally

The other thing I’ve noticed is that the refinement everyone’s so anxious to impose from the start happens on its own if you give it time.

Over time, the stuff you’re putting out gets refined. Not just in the structure and the quality and the mechanics of it, but in what you’re talking about. You start to zero in on different things. You notice which topics you keep returning to, which conversations energise you, which responses from people make you think, “yes, that’s the thing.” The niche reveals itself. You don’t have to manufacture it.

This is fundamentally different from the “niche down and build an offer” approach. That approach says: decide, then execute. What I’m describing is: explore, then notice. The exploration is the strategy. The noticing is what turns it into something coherent over time.

The Real Risk Isn’t Doing It Wrong

I think there’s a part of all of this that I’m speaking to myself about as much as anyone else. There are days, sometimes whole weeks, where I struggle to believe that writing without a clear destination leads anywhere positive. And in the short term, it doesn’t. There’s no immediate payoff for talking about whatever happens to be on your mind this week. The algorithm doesn’t reward intellectual honesty. The growth strategists would tear their hair out.

But in the long term, it creates something that a niche strategy never can: a life you actually want to live. Because you’ve been following your own curiosity the entire time, the opportunities that present themselves are, by definition, aligned with who you actually are, not with some strategic persona you constructed years ago.

The real risk isn’t that you fail to build an audience by refusing to niche down. The real risk is that you succeed at niching down and then discover you’ve built an audience for someone you don’t want to be. You’ve optimised yourself into a professional identity that has nothing to do with what actually matters to you. And then you’re right back where you started, except now you’ve got 10,000 followers watching you figure out that you hate your own content.

Talk about what you want to talk about. Push content out that you’re happy to stand behind - let the rest take care of itself. And if in three years you’re writing about sourdough bread, excellent. At least it’ll be something that actually fulfils you.

If this piece added something to your week, please consider subscribing :)

Subscribe now

There’s a level of understanding you can’t get from user interviews, market research, or competitive analysis. It doesn’t show up in survey data. You won’t find it in a customer discovery playbook. It’s the knowledge that comes from being personally, viscerally frustrated by a problem you haven’t solved yet.

These days I keep catching myself in moments where the friction I’m experiencing isn’t a distraction from the work, realising instead, it is the work.

A few days ago, I spent two and a half hours in a waiting room while my car’s smashed windscreen was replaced. I didn’t have my laptop, I was working from my phone, trying to be productive in conditions that fought me at every turn. And somewhere in the middle of that, it hit me: this is exactly the constraint users of a product I’ve been working on actually face. Not the windscreen, obviously, but the feeling of trying to do meaningful work when your bandwidth is crushed, your tools aren’t quite right, and the gap between what you want to accomplish and what your environment allows feels enormous.

That’s not an inconvenience - that’s a gift: product research you couldn’t outsource, and you couldn’t buy.

The knowledge gap no amount of research closes

Canonical founder advice says: talk to your users, understand their pain points, map their workflows. And that’s sound advice as far as it goes, but it only produces a secondhand understanding - you’re collecting descriptions of friction, not experiencing it.

The difference matters more than most people think. When you’ve personally felt the sharp edge of a problem, you don’t need someone to explain why it’s urgent. You don’t need to be convinced the pain is real. You carry that conviction in your bones, and it shows up in every product and marketing decision you make.

When you’re living inside the problem, you notice things that never surface in interviews. The micro-frustrations. The workarounds people have normalised so completely they don’t even mention them. The moment where energy plummets not because the task is hard, but because the tools make it harder than it needs to be.

A founder who’s studied the problem knows the what. A founder who’s lived it knows the why, and more importantly, the feeling. That feeling is what separates products that technically solve a problem from products that people genuinely love.

Secondhand empathy has a ceiling.

I’ve watched this pattern play out enough times to trust it: the teams that build from lived frustration make different decisions than the teams that build from observed frustration.

When you’ve only observed the problem, you optimise for the metrics you can see. You build features that look right on paper. You solve the problem as described. And sometimes that’s enough.

But when you’ve lived the problem, you catch the things that never make it into a brief. You know that the issue isn’t the five-minute task itself but the twenty minutes of context-switching around it. You know that what users say frustrates them and what actually frustrates them are often different things entirely.

This is why experiencing low bandwidth, experiencing real constraint, is so valuable. Not as a philosophical exercise, but as a direct input into what you’re creating. Every time I find myself frustrated by the gap between what I want to do and what my tools or circumstances allow, I’m learning something I couldn’t learn any other way.

The friction isn’t blocking the insight, the friction is the insight.

The weight of small tasks

One thing I keep coming back to is how badly we misjudge the weight of small tasks when we haven’t done them ourselves. It’s easy to design a workflow and assume it’s simple. It’s another thing entirely to execute it when you’re tired, distracted, low on bandwidth, and juggling six other priorities.

The tasks that look trivial on a whiteboard become genuine obstacles under real constraint. I’ve sat in that waiting room, phone in hand, trying to get something meaningful done in the cracks between interruptions. The gap between what a task should take and what it actually takes under pressure is where most product assumptions fall apart.

You can’t credibly simplify what you haven’t personally found complicated. If you’re designing workflows for people under pressure, you need to have done those workflows yourself under the worst conditions they’ll face. Otherwise you’re not simplifying. You’re guessing.

Why dismissing your own frustration is a mistake

There’s a temptation, especially when things get busy, to treat personal friction as noise. To push through it. To think “I don’t have time to be frustrated right now, I just need to get this done.”

I’ve seen this pattern in teams. Someone flags their lack of bandwidth, their difficulty keeping up with a particular workflow, and the instinct is to solve around it. Find an efficiency. Remove the blocker. Move on.

But sometimes the better move is to sit with the frustration. To ask: what exactly is making this hard? Where does the energy drop? What’s the real chokepoint, and is it the task itself or everything surrounding it?

When the people closest to the problem are struggling with the workflow, that tells you something about the design, not about the people. The only way to see it clearly is to be close enough to feel it yourself.

The danger of distance

The further you get from the problem, the more abstract your solutions become. This is how products end up technically correct but emotionally wrong. The features work. The flows make sense on a whiteboard. But the experience feels off because nobody who built it has recently felt what it’s like to use it under real pressure.

Distance from the problem doesn’t just reduce empathy. It reduces pattern recognition. When you’re living in the friction daily, you start to notice recurring failure points. You spot the moments where people give up, not because the task is impossible, but because the accumulated weight of small irritations becomes too much.

That pattern recognition is a competitive advantage you can’t hire for and can’t shortcut. Someone closer to the problem, someone who feels it more acutely, will always have a sharper intuition for what needs to change and how to communicate the value proposition of what they have built. And if that person isn’t you, or at least someone on your team, they’ll eventually outperform you.

Frustration as a compass

None of this is comfortable. Living inside the problem means being perpetually aware of what doesn’t work yet. It means experiencing your own product’s shortcomings as personal annoyances. It means your car windscreen gets smashed, and instead of just being frustrated, part of your brain is cataloguing the experience as input.

But that discomfort is precisely what makes it valuable. Comfort and insight rarely coexist. The moments where everything runs smoothly are the moments where you learn the least about what needs to change.

So the question isn’t whether you understand your users’ problems. It’s whether you feel them. Whether the friction that defines their experience is something you encounter in your own day, unprompted, without having to simulate it.

If the answer is no, that’s worth paying attention to. Not because you need to manufacture struggle, but because the gap between your experience and your users’ experience is where blind spots live. And blind spots, in the long run, are what get you outbuilt.

Final thought - this is perhaps the single greatest reason (of many) to treat consultants, and agencies, with a healthy double-dose of skepticism.

If this piece added something to your week, please consider subscribing :)

Subscribe now

There’s a definition of high performance that most people carry around without ever questioning it. It sounds something like this: wake up at 5am, hit the gym, execute the morning routine, clear the inbox, crush the to-do list. Performance as a set of objective standards applied uniformly across every day, regardless of context.

The trouble with this version is that it works brilliantly right up until it doesn’t. And the moments where it stops working, illness, exhaustion, life falling apart in small or large ways, are precisely the moments where a useful definition of performance would matter most.

I’ve been very much enjoying a different definition for a while now. High performance, defined subjectively: doing the best you can, with what you have, at any given time. That’s it. No absolute standard. No comparison to yesterday or to someone else’s output. Just an honest assessment of what’s available right now, and whether it’s being used well.

The objective trap

The standard model of performance is seductive because it’s measurable. Five gym sessions a week. Eight hours of deep work. A streak of 30 consecutive days. Numbers give the illusion of control, and control feels like progress.

But measurement creates a binary: you either hit the target or you didn’t. And the moment you miss, the entire framework turns against you. A missed gym session becomes evidence of failure. A low-energy day becomes a broken streak. The system that was supposed to drive performance now generates guilt, and guilt is one of the least productive emotions available.

The deeper problem is what happens during inevitable low periods. Everyone gets ill. Everyone has periods where sleep is broken, energy is depleted, circumstances shift. An objective framework has nothing useful to say about these periods except “you’re falling behind.” Which is completely unhelpful.

Subjective performance as a practice

The subjective reframe changes the question entirely. Instead of “did I hit my target?” the question becomes “given what I actually have available today, am I using it?”

On a good day, that might mean a full training session and four hours of focused work. On a day where man-flu has stripped capacity down to 30%, it might mean a 15-minute walk and one meaningful task. Both can qualify as high performance, because the benchmark adjusts to reality rather than ignoring it.

The standards remain high. They just become responsive to actual conditions. A pilot doesn’t fly the same approach in clear skies and heavy fog. The skill is in adjusting to what’s real, not in pretending conditions don’t exist.

People who operate from an objective framework tend to have two modes: performing and failing. There’s no middle ground, and it’s brutal on the self. The subjective frame introduces a third option: adapting. And adaptation, over long enough timescales, compounds into something far more durable than any streak.

Remove the conditions for failure first

This is where it gets practical, and where most advice gets the sequence backwards.

The default approach to improving performance is additive: add a meditation practice, add a workout routine, add a journaling habit, add a morning routine. Stack enough positive behaviours and eventually the compound effect kicks in.

The problem is that adding behaviours to an already strained system increases the number of things that can go wrong. Every new habit is a new opportunity to fail. And when someone is already in a low period, adding more is almost guaranteed to produce more failure and more experience of defeat, not more progress.

The better sequence is subtractive. Before adding anything, identify and remove the contexts where failure is most likely. If social obligations reliably drain energy and create stress, reduce them. If an ambitious gym schedule generates more missed sessions than completed ones, strip it back. If the current daily structure contains three or four things that regularly don’t get done, stop pretending they will.

This feels like giving up. In practice, it builds a foundation that can hold weight.

Achievable daily wins

What replaces the removed items matters. The goal is a small set of daily activities so achievable they’re almost impossible to fail at. A walk. Five minutes of stretching. One focused task. These aren’t impressive. That’s the point.

The function of achievable daily wins isn’t the activity itself. It’s the feeling of consistency, control, and agency that comes from completing them. Day after day, the experience shifts from “I keep failing at my ambitious plan” to “I keep showing up for my realistic one.” Confidence builds from that foundation, not from the ambition of the plan.

Consistency over intensity. I keep returning to this because the evidence for it is everywhere. The people who sustain performance over years aren’t the ones who had the most intense periods. They’re the ones who found a baseline they could maintain through the low points and built upward from there.

The intense periods come and go. The baseline remains. And the baseline is what determines the long-term trajectory.

The long view

None of this produces dramatic short-term results. That’s partially why it’s undervalued. The culture around performance favours visible intensity: the before-and-after transformation, the 30-day challenge, the dramatic overhaul. Quiet consistency doesn’t photograph well.

But over three months, six months, a year, the person who maintained a sustainable baseline through every dip and disruption is further along than the person who sprinted and crashed repeatedly. The maths is simple, even if the daily experience of it feels underwhelming.

The sequence matters: define performance subjectively, remove the conditions where failure is most likely, build from wins so small they’re almost impossible to miss, and let consistency do the compounding.

This applies to everything: health, entrepreneurship, relationships, investing - you name it. The ambitious plan can come later. First, build something that doesn’t break, and neither will you.

If this piece added something to your week, please consider subscribing :)

Subscribe now

A question I’ve only recently started asking myself, long before any pitch deck, application form or term sheet, is whether an investor’s world actually matches the experience I really want to have. This is fundamentally a journey Vs destination type consideration, and I think it’s one that the vast majority of entrepreneurs ignore.

The default advice is simple. Raise from the best. Get into Y Combinator. Move to San Francisco. Surround yourself with the most financially ambitious people on earth. The logic is clean: better network, better signal, better outcomes.

I’ve come to strongly disagree. Not with the logic, but with the premise.

The premise assumes that the only variable worth optimising for is the probability of a billion-dollar outcome. And if that’s genuinely your single priority, then yes, YC is probably the right call. The data supports it, the alumni network is extraordinary, the brand opens doors that stay closed for everyone else, so on and so forth.

But walking through those doors has a cost, and it’s not financial. It’s philosophical.

What you actually absorb

Y Combinator is not just a programme. It is a culture, a geography, and a specific set of assumptions about what building a company should look like. It assumes you want to live in San Francisco, or at least that you are willing to. It assumes you want to spend your time with YC people, absorb YC values, operate on YC timelines. None of this is hidden, it is the product. And for a lot of founders, that product is exactly what they want. They thrive in it. The density of ambition, the pace, the constant proximity to people who think in terms of scale and speed. That environment is genuinely transformative for certain personalities and certain kinds of companies.

As it turns out though, I’m not one of those founders.

I'd rather not optimise for trying to build a billion-dollar company than spend my time in San Francisco surrounded by people whose definition of success I don’t share. That’s not defeatism. It’s a form of clarity, which it took me a lot of work and a long time to arrive at, about what my motivations are and what I actually value.

Clarity equals trust. I use that phrase often in the context of branding, but it applies just as well to funding decisions. If you aren’t clear on what you’re optimising for, you end up optimising for someone else’s version of success - and by the time you notice, you’re deep into a life you didn’t choose. Good luck ‘trusting the process’ at that point.

Subscribe now

What prestige funding actually buys

When you take money from a prestigious accelerator or fund, you’re not just receiving capital. You’re entering a relationship with a set of expectations. Those expectations shape decisions in ways that are very, very easy to underestimate from the outside.

The geography matters. San Francisco is expensive, intense, and culturally specific. If your life is in Cambridge or Lisbon, relocating is not a logistical inconvenience. It’s a fundamental change to how you spend your days, who you see, what you prioritise.

The peer group matters. YC batches are designed to be competitive and high-velocity. That’s a feature, not a bug, for the right person. But for someone who might build best with space, patience, and a slower rhythm of iteration, that environment can be corrosive rather than catalytic.

The incentive structure also matters: YC is optimised for manufacturing supercharged, venture-scale outcomes. The entire model depends on a small number of companies returning the fund many times over. That means the advice you receive, the metrics you’re encouraged to track, and the definition of progress you absorb will all be calibrated toward rapid scale. If your ambition is simply to build something profitable, sustainable, and deeply good at what it does, you may find yourself constantly swimming against the current of the programme’s own logic.

None of this is a criticism of YC. It’s an observation about fit.

The alternative is not settling

There’s an assumption embedded in the “raise from the best” advice: that choosing a different path is choosing a lesser one. That if you do not apply to YC, you’re either not ambitious enough or not good enough.

This framing is wrong, and it costs people years.

Alignment is the concept that matters most here. The best investor for your company isn’t the one with the biggest brand. It’s the one whose thesis and model of success most closely matches yours, and whose value proposition best matches to your needs and wants. If you want to build a venture-scale business in the Bay Area, YC is probably the best fit on the planet. If you want to build something excellent on your own terms, in a place you actually want to live, with people whose values and character traits you genuinely share, then the “best” investor is likely to be someone else entirely.

The conviction test

I think about this as a conviction test. Not “can I get in?” but “do I actually want what comes with getting in?”

Most founder discourse skips this (mission critical) question. The assumption is that any rational person would want YC if they could get it, and the only interesting question is how to maximise your chances. But that assumption treats founders as interchangeable units of ambition, differing only in ability. It ignores the fact that fundamentally different people start companies for fundamentally different reasons, and those reasons should inform every structural decision that follows.

Some people start companies because they want to build something enormous. Some because they want to solve a specific problem. Some because they want autonomy and craft. Some because they want to prove something to themselves, some because they want to prove something to others. These motivations are simply different. And they lead to different optimal funding structures, different geographies, different timescales, different ‘tribes’ and different definitions of success.

The conventional advice collapses all of this into a single ladder: raise from the best, grow as fast as possible, aim for the biggest outcome. That ladder works brilliantly for a specific kind of founder. For everyone else, it creates quiet pressure to optimise for outcomes they never actually wanted.

What I am actually optimising for

I want to build something that is genuinely good at what it does and connects meaningfully with those who interact with it. I want to do it from a place I choose to live, with people I respect and whose company I enjoy, at a pace that allows for depth rather than just speed. I want the freedom to make decisions based on what I think is right, not what a programme’s incentive structure requires.

That means forgoing certain advantages. The YC network is real, the signal it sends to future investors is real, what it does for your CV is real, the acceleration it provides is real. I’m not pretending those things don’t matter.

But the alignment cost is also real. And for me, it’s too high.

This isn’t a universal prescription. It’s a specific position, held with conviction. If YC, and other entities like it, are the ideal environment for a particular founder, they should pursue it with everything they have. But that quiet resistance some founders might feel, where something about the geography or the culture or the pace does not sit right, isn’t a lack of ambition.

It might be the clearest signal available about what they are actually building for.

If this piece added something to your week, please consider subscribing :)

Subscribe now

Imagine an apartment. You know the layout: where the kitchen is, which window faces north, why the radiator makes that noise in January. The frame is tight. The information set is bounded. A well-designed agent operating within that frame can do genuinely useful things.

Now step back. The apartment sits inside a building. The building sits in a block. The block is part of a neighbourhood, the neighbourhood part of a city, the city part of a country. The frame has expanded, but not linearly. Exponentially. Each zoom level introduces not just more data but different categories of data, different rules, different relevant history - and some of those rules contradict the ones that applied at the level below. The agent that managed your apartment perfectly now has no idea what to do.

This is a structural property of frames.

Subscribe now

The Frame Is Not Just a Technical Limit

The reason I choose this metaphor is that it describes something more fundamental than context window length. Yes, token limits are real. Yes, an agent’s working memory has a ceiling. But even if those technical constraints disappeared overnight, the frame problem would remain.

Because the frame problem is not about information volume. It is about the impossibility of evaluating a frame from inside it.

This is the same argument Gödel made about formal systems, which I presented in a recent essay on the limits of knowledge. A system cannot prove its own completeness using its own axioms. To evaluate whether a frame is the right frame, you have to step outside it. Agents cannot do that. They can accumulate knowledge and apply judgment within a predefined frame. But true judgment - the kind that decides which frame to use, when to abandon it, and whether the metric you are optimising for is even the right metric - requires a vantage point that is not inside the system being evaluated. That capacity is not algorithmic.

So when people suggest that AI will eventually handle every decision, every context, every strategic move, they are proposing that the Gödel problem will be solved. I find that doubtful, not as a pessimist, but as someone who takes the argument seriously.

Why Zooming Out Breaks Things

The conventional assumption is that more context is better. Feed the agent more background, more history, more nuance, and it performs better. There is a version of this that holds, within a frame. But it reverses the moment you cross a frame boundary.

When you zoom from apartment to building, you do not add one layer of context. You introduce a new class of constraints and a new set of actors whose priorities sometimes contradict the apartment-level rules. The agent has no mechanism for arbitrating between conflicting frame-level inputs. It tries to resolve the conflict using the rules it has. Those rules were written for a different scale.

The practical consequence: the more a fully automated system scales up in scope, the more it becomes an echo chamber of its own initial assumptions. It engages with what it was trained to expect. Inputs that fit the mould get processed. Inputs that don’t fit the mould get either forced into the nearest available category or excluded entirely.

This is the hidden cost of full automation. An agent operating in a narrow frame fails visibly, and you can correct it. An agent operating in a wide frame fails silently, by quietly excluding everything its frame cannot accommodate. The output looks coherent. The gaps are not gaps - they are absences. You would not know to look for them.

The Echo Chamber Is Not a Bug

The instinct is to treat this as something that better training will fix. I am not sure it will.

An agent can only engage with what is inside its wheelhouse. That wheelhouse was defined by whoever built the system and by whatever data shaped it. Authentic human engagement - the kind that produces unexpected surface area, where you end up in a conversation you did not plan for that genuinely changes your thinking - is structurally incompatible with a fully closed loop.

The randomness of genuine human impulse is not noise. It is the mechanism by which the frame gets updated. The experience of waking up and finding yourself unexpectedly in an argument with people you have never met about the philosophy of mathematics is not inefficiency (this happened to me yesterday). It is the system working. It creates new context that the agent, operating within its predefined scope, would never have surfaced.

The fully automated version of that person produces content that stays within expected themes, responds to expected inputs, and misses every oblique angle that would have stretched the frame. The output is higher in volume and lower in information.

What Stays Human

Taste and judgment get invoked constantly in discussions about AI, usually as a placeholder for the category of things we haven’t figured out how to automate yet. I think this underestimates what they actually are.

Taste and judgment are the skills capable of evaluating themselves. That is why they compound in a way that pure knowledge accumulation does not. An agent can accumulate facts about what content performs well. It cannot evaluate whether the frame in which performance is measured is the right frame. It cannot decide that the metric itself is wrong and that a different metric would produce better outcomes. Those are frame-level decisions.

An agent can also be wrong about something within a frame and correct itself with more data. It cannot be wrong about the frame and correct itself, because the tools it uses for correction are inside the frame it would need to evaluate. This is not a software version problem. It is a structural feature of what it means to operate within a frame at all.

The implication for anyone building with agents is fairly direct: the human in the loop is the mechanism that evaluates and updates the frame. Remove that function entirely and the system optimises intelligently within a box it can never question.

Design Follows from This

If the frame problem is structural rather than technical, it changes what good system design looks like.

The first question is what to zoom into. The tighter the frame, the more stable the context, the more effective the agent. This is not a limitation to work around - it is a feature to exploit deliberately. An agent that is brilliant within a very tight scope is more useful than an agent that is mediocre across a wide one.

The second question is where the frame boundary is. At what point does the agent’s confidence break down? At what point does the relevant context expand beyond what the system was designed for? That boundary is where human judgment enters. The design has to make that entry point explicit rather than hoping the agent handles it gracefully. It won’t.

The third question is what feeds the frame. If the human is the mechanism for updating the frame, then the quality and variety of human inputs matters significantly. A system that filters all inputs through an automated layer before they reach the human is degrading the frame-update function, not improving it. The agent handles the work within the frame. The human updates the frame. Those are distinct jobs, and conflating them is how you end up with a system that runs efficiently toward the wrong destination.

The best version of building with agents is this: commit to the close range, respect the frame boundary, and preserve the function that evaluates the frame itself. Everything after that is implementation detail.

If this piece added something to your week, please consider subscribing :)

Subscribe now

There’s a sensation i’ve noticed showing up at times of disruption. Not chaos, more like a pressure that says something in life now must change. A trip starts to seem overdue. A career move takes on extra force. Plans that had been sitting quietly in the background suddenly feel urgent.

I’ve been noticing that pressure recently. A significant relationship has ended, and with it the familiar impulse to rearrange everything else around it. Maybe I should trade my car in for a motorhome and take a roadtrip. Maybe I should finally move on the conservation work I’ve been putting off. Maybe I should take the job offer that’s come in. Maybe I should start trying to learn an instrument again. Maybe I should, erm, move to Portugal? Each option arrives with the same feeling attached to it: forward motion. Empowerment.

Subscribe now

The trouble is that forward motion and strategic thinking are not the same thing. But during disruption, they can feel identical from the inside. The problem isn’t action itself. It’s action taken from a disrupted baseline, where the desire for something different starts to impersonate good judgment.

The Difference Between Novelty and Strategy

When I look back at past decisions I’d (wincingly) call non-strategic, the issue usually wasn’t the decision on paper. Some of them worked out well enough. The issue was the state I was in when I made them. The present felt difficult to tolerate, and the future needed to look different quickly.

Grief does this. So does uncertainty. So does any period where the structures that normally give days their shape have loosened or disappeared. When the usual anchors are gone, my mind starts generating alternatives at speed. Each one seems charged. Each one carries the suggestion that acting on it will settle something internally.

Usually it doesn’t. It just changes the setting. Strategy is about improving your position over time. The pull toward novelty is often about getting away from how things feel now. Sometimes those point in the same direction. Often they don’t.

Travel can be a good idea. So can a new career move. So can a move toward work that matters more deeply. But when the main appeal is that life would look unmistakably different, that’s worth treating with more than a fleeting sense of suspicion. A life change can be right and still be chosen for the wrong reason.

Baseline Resilience

At the age of 35 and a half, the phrase baseline resilience finally clicked into place for me. It means something fairly plain: the minimum level of functioning I want to protect even when everything else feels unsettled. Sleep. Food. Training. Work that still matters. Some basic structure to the day.

Not because these things are profound, but because they keep me from becoming unreliable to myself.

This matters because disruption has a way of making stability look suspect. Routine starts to feel like stagnation. Consistency feels passive. The ordinary disciplines can seem too small for the size of the feeling. When there’s a lot moving internally, keeping the same habits can almost look like denial.

But I’ve found the opposite. The boring things are often the only things keeping the floor in place. And once that floor goes, judgment tends to go with it.

I’ve learned to be wary of any plan that requires me to abandon the few things that reliably keep me clear-headed. That doesn’t mean the plan is wrong. It does mean it deserves more scrutiny, not less.

Share

Preserving Optionality Isn’t Stalling

There is, obviously, a way to use patience as camouflage. It’s possible to delay decisions indefinitely and call it discernment. Not every pause is wise. But preserving optionality isn’t the same as drifting. It means not forcing a commitment while your judgment is being bent by urgency. It means keeping paths open long enough to assess them from steadier ground.

For me, that often looks less impressive than the mind wants it to. Stay where I am a little longer. Keep the process going on opportunities that might matter. Don’t turn every new option into an answer to the whole period I’m in.

More often than I’d like, the right move is the least dramatic one: stay put, stay consistent, and keep doing the things that move the needle in ways that are hard to appreciate day to day but become abundantly obvious over time.

That kind of action rarely feels good on contact. It doesn’t produce relief. It doesn’t give the satisfying sense that a new chapter has begun. Which is exactly why it’s easy to miss.

Movement Without Abandoning the Baseline

None of this means the answer is always to do nothing. Sometimes a change in scene helps. Sometimes a move really is right. Sometimes the thing you’ve been putting off genuinely is the next step. The question is whether the impulse survives inspection.

I keep coming back to that with things I do genuinely want, like volunteer/conservation work, and with ideas that seem to offer movement without total instability. What interests me now is less the fantasy of escape than whether there’s a way to move without wrecking the baseline in the process.

That feels like a better test than asking whether something is exciting or overdue. Does this choice support the parts of life that keep me steady? Or is it mainly attractive because it lets me feel that something has changed? And that I have recovered some sense of being the agent of change in my own life?

A desire can be real and still get recruited into avoidance. That’s part of what makes disrupted periods hard to think clearly inside.

Recognising the Pattern Is the Work

I’ve come to think of this as a skill: noticing urgency before it hardens into a story about what must happen next.

I’ve made enough non-strategic choices to recognise the sequence by now. Something destabilises. Big options start to glow. A convincing narrative forms around why one of them is clearly the answer. Sometimes the move itself isn’t even bad. But the reasoning behind it is thinner than it first appears, and the timing is doing more work than I want to admit.

What’s different now isn’t that the pull has gone away. It hasn’t. It’s that I can see it earlier. And seeing it earlier usually points me back to the same place: hold the baseline, keep the options open, and let the more dramatic ideas survive a period of steadiness before reorganising life around them.

If a plan still makes sense after a stretch of ordinary, disciplined life, it’s probably real. If it only made sense when I felt desperate for things to look different, that tells me something too.

Now back to browsing the classifieds for a new motorhome. And a piano.

If this piece added something to your week, please consider subscribing :)

Subscribe now

There’s a pattern I’ve seen play out across multiple crypto projects, and it’s not about technology, funding, or even timing. It’s about professionalism - or more precisely, its absence.

I spent a formative stretch of my career working alongside elite forces operators in the middle east. Whatever your preconceptions about that world, the thing that left the deepest mark on me wasn’t the intensity or the risk. It was the standard. The absolute, non-negotiable standard of accountability, communication, and execution that governed every interaction. There wasn’t room for ambiguity about who owned what. There wasn’t tolerance for sloppy handoffs. Emotions didn’t dictate the conversation - clarity did. You said things exactly how they were, people absorbed it, and everyone moved forward.

That experience recalibrated my sense of what “professional” actually means. And it made everything that came after - building and running multiple companies simultaneously in the UAE—where, at the time, if you missed a payment or bounced a cheque (easy to do when nobody pays anyone, if they can possibly help it), you did not pass GO, you did not collect £200, you went straight to jail—working high-profile security and risk management jobs, managing multi-million dollar film and event projects, overseeing an $11.3M fleet of 74 luxury vehicles in the hands of spectacularly irresponsible clients - all manageable. Not because any of it was simple, but because the operating standard had been set so high early on that everything else was just execution against that baseline.

Then, after a couple of years as a partner at a CleanTech fund and a couple more consulting in that space, I entered the wild world of crypto.

The Horse That Didn’t Want to Drink

When I started consulting for crypto projects, I expected the usual startup chaos - under-resourced teams, unclear priorities, people wearing too many hats. That’s normal. That’s solvable. What I encountered was something fundamentally different.

The teams I worked with weren’t struggling because they lacked skills or resources. They were struggling because they actively resisted the structures that would have made them functional. Operational rigour, clear accountability, systematic decision-making - these weren’t things they hadn’t discovered yet. These were things they had consciously rejected.

The meat of my experience consulting in this space was trying to build operational structure around teams that genuinely didn’t want it. They’d got into crypto to escape those things in real life. The corporate hierarchies, the reporting lines, the performance reviews, the structured decision-making processes - for many of these people, crypto represented freedom from all of that. And they weren’t about to let some consultant reimpose it.

It was very much a case of trying to lead a horse to water and make it drink, and it was not remotely thirsty.

This is the part that I think gets misdiagnosed. The conventional narrative is that crypto teams are “immature” - that the industry is young and these organisations will professionalise over time as they grow. I don’t think that’s what’s happening. What I observed was a self-selection problem. The same anti-institutional impulse that draws brilliant people into decentralised technology actively repels the operational discipline needed to capitalise on what they build.

Brilliance Without Structure Is Just Expensive Chaos

The projects I worked with weren’t short on talent. There were excellent people. The technical foundations were often genuinely impressive. But hardly anybody - and I don’t say this lightly - was overly characterised by the idea of professionalism. There were traders, marketers, developers, people with a general sense of business. But there was no culture of rigour around how decisions got made, how information flowed, or how accountability was distributed.

One project I spent over a year with was navigating a strategic pivot. It should have been a complex but manageable transition. Instead, it became an object lesson in what happens when an organisation lacks the connective tissue of basic operational discipline.

The ecosystem they were inhabiting could itself have really done with guiding lights and real organisation and structure, because it was all a total mess. It made everything exponentially harder. Decisions that should have taken days took months. Debilitating and brutal politics. Dependencies went untracked. Communication happened in fragments. The technology was interesting. The execution environment was borderline non-functional.

And this wasn’t unique to one team. It was a pattern across almost every project I consulted for. Founders who could architect complex systems but couldn’t run a structured meeting. Teams where nobody knew who owned what decision. Organisations where the very concept of process was treated as an imposition rather than an enabler.

Subscribe now

The Self-Selection Trap

Here’s a hypothesis for what is actually happening, and why the “they’ll grow out of it” thesis might be wrong.

Crypto, by its philosophical nature, attracts people who are sceptical of centralised authority, formal hierarchies, and institutional control. That scepticism is precisely what makes them good at imagining and building decentralised systems. It’s also what makes them terrible at running organisations.

This isn’t a temporary phase. It’s a structural feature of who the industry attracts. The people most drawn to the vision of trustless, permissionless systems are often the people least inclined to submit to the very human, very trust-dependent dynamics that make teams work - clear ownership, honest feedback loops, structured accountability, the willingness to defer to someone else’s judgement when the situation demands it.

The result is an industry full of extraordinary technology and dysfunctional organisations. Teams that can write elegant smart contracts but can’t agree on a quarterly roadmap. Ecosystems with billions in notional value and no one willing to make a difficult decision and own the consequences.

What Actually Works

I don’t want to overstate my own credentials here, but I have seen what works. And it’s not complicated. It just feels a bit too much like adulting for people who entered this space specifically to avoid it.

It starts with someone being willing to say things exactly how they are. Not with cruelty, but with clarity. The environments I’ve worked in that functioned at the highest level shared one trait above all others: information moved without distortion. Problems were named plainly. Ownership was unambiguous. And nobody confused transparency with hostility.

That directness has always been a gift and a curse - in environments that value it, it accelerates everything. In environments that don’t - it creates friction. I’ve spent the last few years learning to filter that instinct and introduce structure in ways that don’t feel like an attack on the culture. But the underlying principle hasn’t changed: organisations that can’t handle honest, direct communication about what’s working and what isn’t are organisations that will underperform their potential indefinitely.

The Gap That Matters

The crypto industry doesn’t have a technology problem. It arguably doesn’t even have a product-market fit problem - at least not primarily. It has a professionalism problem. And it’s a professionalism problem that’s unusually resistant to solving because the people who need to solve it are the same people who chose this industry partly to avoid solving it.

The projects that break through - the ones that actually capture the opportunities their technology creates - will be the ones that figure out how to marry the anti-institutional creativity that makes crypto compelling with the operational discipline that makes organisations functional. That’s not about imposing corporate structure on decentralised teams. It’s about recognising that accountability, clarity, and rigour aren’t corporate artefacts. They’re human necessities. They’re what allow talented people to actually do the work they’re capable of, rather than losing it to noise, misalignment, and the slow bleed of decisions that never quite get made.

The technology in this space is extraordinary. The gap between what’s being built and what’s being executed against isn’t the most interesting problem in the industry, but it is one of the most fundamental. And it’s almost entirely an attitude problem.

Retail investors, the hype will return. And when it does, remember, you have been warned.

I publish these essays without a paywall so they remain open to anyone who finds them valuable. If this piece added something to your week, you are welcome to support the work by buying me some research fuel!

Buy Me A Coffee

AI isn’t making knowledge redundant. It’s making the public performance of knowledge cheaper, and that’s exposing a difference modern culture has spent a long time blurring: the difference between knowing more and understanding better. For years, we rewarded people for storing, retrieving, summarising, and displaying information. Now that machines can do much of that on demand, a harder question returns. What does understanding consist in, and why does it matter more when answers are easy to produce?

This isn’t a small shift. It changes how we think about education, work, intelligence, and even what sort of person we’re trying to become. For a long time, modern societies treated knowledge as both power and status. That wasn’t irrational. Information was unevenly distributed, difficult to find, and often slow to verify. If you knew more than other people, or if you could organise what you knew better than they could, you gained a real advantage. Schools rewarded recall. Professions rewarded expertise. Public life rewarded the appearance of being informed.

AI doesn’t remove the need for truth, and it doesn’t make genuine expertise unnecessary. But it does weaken the social value of certain visible performances of knowledge. It can produce a plausible summary, a tidy explanation, a competent comparison, or a polished draft in very little time. That changes the meaning of activities that once looked like strong evidence of intelligence. We may not be entering a post-knowledge society, but we may be entering an early stage of a society in which knowledge, by itself, no longer carries the status it once did.

That shift matters because knowledge and understanding aren’t the same thing. Knowledge can often be stated. It can be listed, tested, quoted, and transferred. Understanding is less portable. It involves seeing how something fits together, where an explanation holds, what it leaves out, and why it matters. A person can memorise the rules of a game and still not understand how the game is actually played. They can learn the language of economics and still not understand how fear moves through a market. They can read about grief and still not understand what grief does to the shape of a life.

This is why Hermann Hesse’s Siddhartha still feels relevant. The novel’s often remembered as a spiritual book, but one of its deepest concerns is the gap between borrowed wisdom and realised understanding. Siddhartha passes through teachers, disciplines, pleasures, disappointments, ambition, and loss. Again and again he encounters systems that promise meaning. Again and again he discovers that no teaching, however refined, can do the seeing for him. That isn’t because teaching is worthless. It’s because there are some truths that can’t simply be handed over in words and possessed like objects. They have to become real within the person who hears them.

At first glance, this may seem like a question in epistemology, the branch of philosophy concerned with knowledge: what it is, how we justify it, and where its limits are. But it quickly becomes an ontological question as well, because the way we know the world shapes the kind of people we become within it. A culture that trains people to collect facts, perform fluency, and move quickly from question to answer will produce one kind of self. A culture that trains people to stay with uncertainty, compare explanations, and return repeatedly to lived reality will produce another.

Hilary Lawson’s idea of closure helps explain why this matters. The term can sound abstract, but the core insight is simple. Reality is always richer, messier, and more open than the descriptions we use to manage it. In order to think and act at all, we draw boundaries around that openness. We simplify. We name. We classify. We frame. Lawson calls these acts of simplification closures.

A tree is a good place to start. To a builder, a tree may be timber. To a walker, it may be shade. To a bird, it’s a habitat. To a child, it might be a place to climb or hide. To a climate scientist, it may be part of a carbon cycle. None of these descriptions is false. Each captures something real. But none of them is the whole tree. Each is a way of closing a larger reality into a form that serves a purpose.

Subscribe now

The same thing happens with people. Think about how often we describe one another by professional titles. Someone’s a doctor, a teacher, a founder, a manager, a solicitor. These labels are useful. They tell us something real. But they also reduce a person to a narrow function. The title doesn’t tell you about their private fears, their loyalties, their grief, their humour, their moral blind spots, or the contradictions that shape their life. It gives you a workable frame, not the whole person. That, too, is a closure.

Lawson’s wider philosophy is sometimes described as post-reality, which can sound more dramatic than the point really is. He isn’t saying that reality is unreal, or that truth doesn’t matter. He’s saying that we never encounter reality in a pure and final way, untouched by interpretation. We meet the world through closures. We use maps, categories, models, and names because we must. They’re how we cope with the openness of things. The danger begins when we forget that these closures are partial and start treating them as the whole.

Once that idea is in view, AI looks slightly different. It isn’t just a machine for retrieving facts. It’s a machine for producing closures quickly and persuasively. You bring it an open question and it gives you a usable frame. It can turn confusion into structure, abundance into summary, and uncertainty into a list of options. That’s one reason it feels so helpful. But a useful closure isn’t the same thing as understanding. Understanding includes some awareness of what the closure leaves out, where it stops working, and how it changes what becomes visible to us.

This is also why the growing presence of AI can feel unsettling in a way that goes beyond economics. The anxiety isn’t only about jobs or competition. It’s also about identity. Many people have built their sense of value around being knowledgeable, articulate, or mentally quick. But, when a machine can imitate those outward signs of intelligence, the older question returns with more force. What, exactly, was valuable in the first place? Was it the possession of information, or was it something deeper that information only sometimes serves?

At this point, consciousness becomes hard to avoid. Part of what we ordinarily mean by understanding seems to involve more than correct output. It involves having a point of view within experience. Things matter to us. They wound us, attract us, burden us, embarrass us, and change us. We don’t merely process grief, beauty, shame, love, loyalty, responsibility, or mortality as topics. We encounter them as realities within a life. A system may be able to describe grief with great fluency. That doesn’t settle whether it understands grief in the way a bereaved person does. A system may generate language about love or fear, but those aren’t only concepts. They’re lived states of being.

This doesn’t prove that machines can never understand. That claim is much harder to establish than many people assume. But it does point to something important. Our ordinary sense of understanding includes an experiential dimension. It includes salience, consequence, and inwardness. To understand something, in the richest human sense, isn’t just to be able to describe it. It’s to have been changed by contact with it.

Gödel enters the discussion at precisely this pressure point, though he’s often used too carelessly. His incompleteness theorems don’t prove that human beings are magical, nor do they decisively prove that machines can never think. What they do show is that any formal system rich enough to express arithmetic contains truths that can’t be derived from within the system itself. Gödel himself took this to raise a serious question about whether human mathematical understanding could be fully captured by any finite procedure. One doesn’t have to accept the strongest anti-machine reading of Gödel to see what makes it important. It reminds us that any picture of intelligence as nothing more than rule-following inside a closed system is likely to be incomplete.

That matters beyond mathematics. Understanding often involves more than operating competently within a framework. It also involves recognising the limits of the framework, seeing what it excludes, and sometimes stepping outside it altogether. In practice, this is something human beings do all the time. We notice that a model is too simple, that a category no longer fits, that a theory explains one level of reality while obscuring another. We revise our frames, sometimes painfully. We don’t just calculate within them.

This is part of what makes the pursuit of knowledge so double-edged. Knowledge can steady us, guide us, and in many fields save lives. But when it’s pursued mainly as possession, performance, or defence against uncertainty, it can become a source of mental unrest rather than peace. The person who treats knowledge as status or control never quite has enough of it. There’s always another book to read, another framework to master, another opinion to track, another update to absorb. The mind becomes crowded, but not necessarily settled.

This isn’t only a modern problem, but modern culture intensifies it. We live amid endless information, endless commentary, and endless incentives to stay current. The result is a peculiar form of strain. We’re invited to know more and more while being given fewer conditions under which understanding can deepen. Speed is rewarded. Reaction is rewarded. Accumulation is rewarded. But understanding often requires the opposite conditions. It needs time, repetition, perspective, and contact with something more stubborn than discourse.

That’s why the distinction between knowledge and understanding matters so much in the age of AI. The technology didn’t invent the confusion, but it’s made it harder to ignore. When answers become cheap, the value of understanding becomes easier to see. Not because understanding is mystical or anti-intellectual, but because it includes the things that cheap answers don’t automatically give us: judgement, context, proportion, first-hand contact with reality, and some capacity to recognise when a neat frame is too neat.

If that’s right, then the task ahead isn’t simply to defend knowledge from machines, as if the important thing were to preserve our monopoly over information. The deeper task is to recover forms of learning and culture that value understanding more highly than performance. In education, that would mean rewarding explanation, application, and comparison more than recall alone. In work, it would mean valuing the person who can ask the better question, spot what a model has missed, or recognise which human reality is being flattened by an efficient abstraction. In ordinary life, it would mean reading less frantically, returning more often to direct experience, and using AI as a tool for first drafts rather than final judgements.

It would also mean becoming more alert to closures in our own thinking. When we reduce a person to a title, a problem to a slogan, or a life to a metric, we gain speed and lose depth. Sometimes that trade is worth making. Often it isn’t. Understanding grows in the ability to notice the trade, rather than living inside it unconsciously.

The point, then, isn’t to set knowledge and understanding against each other as though one were scientific and the other spiritual. We need knowledge. We depend on it. But knowledge is at its best when it serves understanding rather than pretending to replace it. It’s a tool for orientation, not a substitute for it. AI is making that harder to miss. The cheaper answers become, the more clearly we can see that a human life can’t be guided by answers alone. It also needs the slower, less transferable, and more demanding work of learning what those answers mean, where they fail, and how to live with what they reveal.

I publish these essays without a paywall so they remain open to anyone who finds them valuable. If this piece added something to your week, you are welcome to support the work by buying me some research fuel!

Buy me a coffee

The revealing part of a meeting often comes after it ends.

Everyone agrees, or says they do. People stand up. The energy breaks. Then one person stays behind for another ten minutes and makes the work real. They send the note nobody formally owned, fix the sentence that would otherwise be misread, attach the file that was mentioned three times and still not attached. The meeting presents itself as collective effort. Completion is usually more personal than that.

The Ringelmann effect is the old finding that as groups get larger, individual effort tends to fall. It’s usually introduced as a story about loafing. People hide in the crowd. Responsibility diffuses. Fair enough. But that’s only half the phenomenon. When effort diffuses in a group, the missing effort doesn’t simply vanish. Very often it concentrates elsewhere.

That’s the part worth thinking about.

In most teams, missing effort migrates toward the person least able to tolerate an avoidable failure. This is why conscientious people so often end up overextended. Not because they’re formally assigned more work, though sometimes they are, but because they absorb more ambiguity. And ambiguity is where a surprising amount of modern work actually lives.

An unclear brief is work waiting to happen. A vague promise made in a meeting is work waiting to happen. A delicate client relationship nobody has really taken charge of is work waiting to happen. The same goes for a decision that everyone assumes someone else will make, or a deadline that exists socially before it exists operationally. Much of what exhausts competent people isn’t the visible task list. It’s the unclaimed residue around the task list.

That residue has to go somewhere.

The conscientious are vulnerable because they experience that residue differently. A loose end doesn’t register as mild disorder. It registers as a claim on attention. A missing follow-up isn’t just something that would be nice to resolve; it becomes difficult to leave unresolved. This is what many people miss about conscientiousness. It’s not merely diligence. It’s a low tolerance for unfinishedness.

From the inside, stepping in feels like fidelity to a standard. From the outside, it becomes a free resource.

Those are the same event, viewed from two levels.

At the personal level, the conscientious person thinks, this shouldn’t be allowed to go wrong for such a stupid reason. At the group level, the system learns that certain kinds of vagueness will be metabolised by a particular person. Once that learning happens, standards start migrating out of process and into personality. The team doesn’t need to become clearer because someone inside it is reliably clearer than the team.

This is why praise can be misleading. When an organisation says someone is indispensable, that can mean one of two things. Either the person is genuinely exceptional in a healthy system, or the system has discovered a cheap way to avoid improving itself. In practice it’s often the second. A weak handoff, a fuzzy role boundary, a manager who avoids crisp decisions, an overcommitted team that keeps promising more than it can structure properly, all of these can appear workable for much longer than they really are if one careful person keeps converting disorder into output.

Subscribe now

The more reliable you are, the less feedback the system receives about its own design failures.

That’s the real competence tax. It isn’t just that capable people do more. It’s that their competence conceals the amount of dysfunction around them. They make the team look more coordinated than it is, the timeline more realistic than it is, the manager more competent than he is, the culture more responsible than it is. Their private standards subsidise the public appearance of order.

This creates a nasty psychological trap. If the conscientious person stops stepping in, standards drop. Things are missed. Clients notice. Colleagues get exposed. The person feels complicit in a decline they can already see coming. If they do step in, the immediate problem is solved, but the system learns again that it can outsource clarity to their conscience. Either way, they lose. They’re caught between an internal standard and an external structure that increasingly depends on it.

That’s why so many competent people look calm while living with a low-grade sense of pursuit. They aren’t simply busy. They’re surrounded by small pockets of unowned responsibility, each one emitting a faint signal that they find harder to ignore than other people do. Over time this changes how they move through the world. They start scanning by default. They anticipate failure before it happens. They read the room for what hasn’t been said. They listen for the promise inside the vague statement. They become good at compensation, which makes them even easier to depend on.

Then usefulness hardens into identity.

This is where the problem gets moral as much as organisational. Reliable people often derive self-respect from being the one who can be counted on. That isn’t vanity. It’s usually bound up with sincere values: care, precision, dislike of waste, dislike of embarrassment, a wish not to let others suffer for preventable mistakes. But once usefulness fuses with self-concept, boundaries start to feel dishonourable. Saying this isn’t mine feels petty. Letting someone else experience the consequence of their own vagueness feels cruel. So the person continues to intervene, and every intervention confirms the system’s quiet expectation that they will.

A lot of modern burnout isn’t overwork in the simple sense. It’s moral overextension. It comes from repeatedly treating every visible gap as if it were yours to close.

This isn’t confined to work. The same pattern shows up in families, friendships, and communities. One person remembers birthdays, notices tensions early, keeps track of what was promised, smooths the logistics, absorbs the social awkwardness, and notices what would otherwise go unsaid. That person is often described as generous, mature, or emotionally intelligent. They may be all three. But there’s still a structural question underneath the praise: why has one person’s conscience become everyone else’s backup system?

Once you see the pattern, a lot of common advice starts to look shallow. “Care less” isn’t serious advice for people whose value lies partly in caring. Nor is it even desirable. The world does in fact depend on those who notice what others miss. The better move is sharper discrimination. Not every gap is yours. Not every standard requires private enforcement. Not every failure should be quietly prevented.

The key distinction is between caring about an outcome and owning the machinery required to produce it. Competent people often collapse those two. They see a standard and assume custody. But care isn’t ownership. You can care deeply that a thing be done well without agreeing to become the place where all unresolved parts of it accumulate.

That distinction has practical consequences. It means naming ownership early, before social convenience has a chance to blur it. It means asking who is actually responsible, what “done” specifically means, and what happens if it isn’t done. It means noticing when you’re being handed not a task but an atmosphere. It also means allowing some failures to become visible. Not out of spite, and not to teach anyone a theatrical lesson, but because hidden failure teaches the system nothing. If every dropped ball is caught by the same person, the group never learns how often it’s dropping them.

Good leaders understand this instinctively. They don’t treat repeated heroics as a sign of health. They treat them as diagnostic data. If the same people are always rescuing the same classes of problem, there is almost certainly a design flaw somewhere upstream. The job isn’t to thank them more eloquently. The job is to stop using character as a substitute for structure.

The usual reading of the Ringelmann effect is that groups make individuals lazy. The more interesting reading is that groups also teach conscientious people to live as spare capacity. One story is about reduced effort. The other is about concentrated burden. They’re two sides of the same phenomenon.

The missing effort in a group rarely disappears. It settles on whoever finds it hardest to watch things fall apart.

I publish these essays without a paywall so they remain open to anyone who finds them valuable. If this piece added something to your week, you are welcome to support the work by buying me some research fuel!

Buy me a coffee

There’s a thought experiment I keep coming back to. It’s simple, and frankly it’s a bit depressing.

Say you’re building or using an AI content engine (as many are starting to). You feed it your transcripts, your notes, your half-formed ideas. It synthesises all of that into posts, articles, threads. Fine. That’s basically what anyone using AI for content is doing now, to varying degrees of sophistication.

But then you think: why stop at my own material? Why not slide Peter Thiel’s “Zero to One” into the content library? Or a few Naval Ravikant essays? Maybe some Rob Henderson, some Richard Branson, a bit of Nassim Taleb for texture. Before long, your content engine is drawing on the same canon of ideas that every other ambitious person on the internet has also fed into their content engine.

And then what? Everyone’s output starts converging. Not word for word, obviously. But the ideas, the frameworks, the references, the worldview. It all starts to sound like it was written by the same well-read ghost. That is what’s going to happen.

‘Show me the incentive and I’ll show you the outcome’ - Charlie Munger.

If the thought has popped into my head, it’ll pop into everyone else’s over time too.

Synthesis is not the problem

I want to be precise about what I’m describing, because there’s a version of this that’s completely fine, even admirable.

Chris Williamson is a good example. He reads widely, interviews deeply, and synthesises ideas from dozens of thinkers into something accessible and well-presented. He’s not claiming the thoughts are original. He’s a conduit, a discovery layer. You watch a Williamson clip and you encounter an idea from, say, evolutionary psychology or behavioural economics that you might not have found on your own. That’s valuable. That’s how knowledge moves through culture: it gets recycled, reapplied, repackaged for new audiences, and sparks new thinking in the process.

He doesn’t tend to produce deep original insights. He synthesises and re-hypothecates, which is absolutely nothing wrong with that. If I could do it with the fidelity and consumability and presentation quality that he does, I’d be pretty happy with that.

The key is that Williamson does this openly. The synthesis is transparent.

What I’m worried about is the non-transparent version. The version where AI handles the synthesis invisibly, and the person presenting the ideas has nothing other than a moral obligation to tell you where any of it came from.

The intellectual laundering problem

Here’s the disparity that matters: there’s synthesis, and then there’s laundering.

Synthesis is taking ideas from multiple sources, combining them in ways that produce new understanding, and being upfront about the ingredients. Every good essayist, every good podcaster, every good teacher does this. You stand on the shoulders of others and you say so.

Laundering is feeding other people’s thinking into a system that outputs it in your voice (AI content systems will get better and better at this over time), under your name, with no trail back to the source. The ideas pass through the AI like money through a shell company. They come out clean, they sound like yours, and nobody can easily prove otherwise.

The uncomfortable part: these two things look identical from the outside. An synthesised post drawing on Thiel and Henderson and Taleb reads exactly like a thoughtful person who’s read Thiel and Henderson and Taleb. The output is the same. The process is completely different.

And here’s the deeper issue: as the tools get better, even the process distinction starts to blur. Is it laundering if you genuinely understood the source material and just used AI to articulate it more clearly? Is it synthesis if you never actually read the books but your AI did? Where’s the line?

I don’t have a clean answer to this yet, but intuitively, we all know where the line is.

The convergence endpoint

The practical consequence is easier to see. When everyone’s content engine is drawing on the same pool of high-status thinking, the internet’s intellectual diversity collapses.

Right now we’re in an awkward transition period. Some people are still writing from genuine first-person experience and hard-won insight. Others are running the laundering playbook, consciously or not. The two coexist because you can still, if you’re paying attention, feel the difference. There’s a texture to writing that comes from actual experience, actual uncertainty, actual thinking-in-progress. It doesn’t always sound polished. It contradicts itself sometimes. It has rough edges.

AI-synthesised content, by contrast, tends toward a certain effortless competence. Every point is well-made. Every transition is smooth. Every conclusion is tidy. It’s the uncanny valley of thought leadership: technically impressive, emotionally hollow.

But the tools are getting better fast. The rough edges will be engineered in. The “human feel” will become a style parameter, not a signal of authenticity.

The endgame, as a colleague put it to me recently, may just be agents talking to agents. Your content engine publishes something. My content engine reads it, incorporates anything useful, and publishes its own response. Your engine reads that, updates its model of the discourse, and produces the next iteration. The humans involved are, at best, approving outputs. At worst, they’re not even in the loop.

At that point, the entire content ecosystem is just a closed system of AI-generated ideas circulating between AI consumers, with humans occasionally dipping in to skim the output.

Doesn’t that sound healthy and fulfilling? Not really.

And depressingly, this certainly seems to have started playing out to me. Human ‘agents’ all over LinkedIn, Substack and I assume elsewhere too are simply prompting LLMs to create content, then other human agents prompting LLMs to generate a response. And so it goes, back and forth in a depressingly transparent game of AI slop tennis of which I am assumed to be a naive and impressed spectator.

The citation gap

One thing that keeps nagging at me: we have citation infrastructure for academic work. If you write a paper and use someone else’s findings, you cite them. There are norms, there are systems, there are consequences for not doing it.

Content has nothing like this.

As a bit of a thought experiment, imagine a tool, something like a browser extension or an overlay, that could trace a claim or idea back to its likely origin. You’re watching a Williamson video, he surfaces some concept about status signalling, and the tool flags: “The original articulation of this idea was Rob Henderson, ‘Luxury Beliefs’, 2019. Here are the primary sources.” Not to discredit Williamson. Just to give you the lineage.

This would be genuinely useful. Not as a gotcha mechanism, but as a discovery tool. The value of synthesis is that it surfaces ideas you wouldn’t have found on your own. The limitation of synthesis is that it often strips away the context, the nuance, the original argument. A citation layer reconnects you to the source.

And if you built that kind of verification into a content pipeline from the start, if your own system tracked where every idea came from and included that provenance in the output, you’d have something that most content creators don’t: transparent intellectual supply chains. You could scale that to read anyone else’s content and do the same analysis.

That might actually be a meaningful differentiator in a world where everyone’s content engine is regurgitating the same inputs. Not “my ideas are more original than yours.” Just: “I can show you where everything came from, and I do.”

I’m planning to expand on this thought experiment in a forthcoming article.

What’s Actually Scarce

If synthesis is abundant and getting more abundant by the day, the scarce thing isn’t well-articulated thinking. It’s the raw material: genuine experience, genuine uncertainty, genuine first-person encounters with reality that haven’t been pre-digested by anyone else’s framework.

The person who actually built the thing, who actually failed at the thing, who actually sat with the confusion long enough to develop a non-obvious take. That’s the input that no content engine can fabricate, at least not yet. Everything downstream of that, the synthesis, the articulation, the distribution, AI handles increasingly well.

So the question isn’t really “how do we stop the homogeneity crisis.” It’s probably unstoppable. The question is: what are you feeding into the machine that nobody else has?

If the answer is “the same books and podcasts everyone else consumes,” then yeah, your output is going to converge with everyone else’s. That’s just the maths.

If the answer is “things I’ve actually lived through, actually struggled with, actually thought about at 3am when I couldn’t sleep,” then maybe you’ve got something the content singularity can’t easily replicate.

Maybe. Hopefully. We’ll see.

Subscribe now

There’s a version of self-help advice that tells you to trust your gut, follow your heart, listen to your feelings. It sounds right. It feels empowering. And for a long time, I believed it.

Then I spent most of my thirties unpicking the motivations behind the decisions I’d made in my twenties, and I realised something uncomfortable: I had been completely blind to the forces pulling me. Not partially blind. Not occasionally misled. Completely blind.

The emotions I’d trusted, the instincts I’d followed, the gut feelings I’d acted on, they weren’t broken exactly. They were doing their job. The problem was that their job was not what I thought it was.

Emotions are reactive, not proactive

Here’s the distinction that changed everything for me. I’d always treated my emotions as forward-looking signals. A feeling of anxiety meant something ahead was dangerous. A feeling of desire meant something ahead was good. A feeling of responsibility toward someone meant I was needed.

But emotions don’t work that way. They’re reactive. They’re responding to patterns your nervous system learned years or decades ago, often in circumstances that bear no resemblance to where you are now. The anxiety you feel before a conversation might have nothing to do with that conversation and everything to do with a dynamic you learned at seven years old. The desire you feel for a particular person might have nothing to do with compatibility and everything to do with a wound that person’s presence temporarily soothes.

This is not some abstract psychological theory. This is what I found when I started honestly examining why I had done what I’d done. The reasons I’d given myself at the time, the narratives I’d constructed, they were coherent and plausible and almost entirely wrong. The actual motivations were buried underneath, operating silently, and I’d been following their instructions without ever questioning who was giving the orders.

The underlying motivation is often untrustworthy

I want to be precise about what I mean here, because “don’t trust your emotions” can sound like “be a robot” or “suppress everything you feel.” That’s not it.

The feeling itself is real. If you feel pulled toward someone, that pull is genuine. If you feel responsible for another person’s wellbeing, that sense of responsibility is genuinely there. The problem is not the emotion. The problem is the story you tell yourself about why you feel it.

You feel responsible for someone, so you tell yourself: “They need me. I’m the only one who can help.” But the underlying motivation might be: “Rescuing people is how I learned to earn love as a child, and if I stop rescuing, I have to face the fact that I don’t know how to receive love that isn’t transactional.”

You feel drawn back to a situation that hurt you, so you tell yourself: “It’s different now. Things have changed.” But the underlying motivation might be: “Familiar pain feels safer than unfamiliar growth, and my nervous system would rather repeat a known pattern than face the uncertainty of something new.”

The emotion is the messenger. But the message it’s carrying was written by a version of you that no longer exists, for circumstances that no longer apply, in service of survival strategies you no longer need. And if you act on the message without reading it carefully, you end up serving those old strategies instead of your actual life.

The decade of discernment

I don’t say this from a position of having it figured out. I say it as someone who has spent roughly a decade doing the slow, tedious, often humbling work of asking: “Why did I actually do that?”

Not the comfortable answer. Not the one that preserves the story I tell about myself. The real one.

And the pattern I kept finding was the same: decisions I’d been proud of were often driven by something I hadn’t examined. Relationships I’d thought were about commitment were partly about rescue. Risks I’d thought were about ambition were partly about proving something to someone who wasn’t even watching any more. Values I’d thought were mine had been borrowed from contexts I’d outgrown.

This is not a process you complete. There’s no moment where you’ve fully mapped every hidden motivation. But there is a significant difference between someone who has spent years doing this work and someone who hasn’t started. The difference is not that the first person always makes better decisions. It’s that the first person knows to pause before acting on a strong feeling and ask: “What is this actually about?”

The cost of self-betrayal

Here’s where this becomes practical rather than philosophical.

When you act on an emotion you haven’t examined, and that action goes against what you actually want for your life, something specific happens. You betray yourself. And self-betrayal has a compounding cost that people wildly underestimate.

Every time you make a decision that contradicts your own standards, your own knowledge of what’s good for you, your own hard-won understanding of what works and what doesn’t, you lose a small piece of self-trust. And self-trust, once eroded, is one of the most difficult things to rebuild.

It works like this: you know you shouldn’t go back to the situation that hurt you. You feel pulled back anyway. You follow the feeling. It goes wrong, exactly as you predicted. Now you have two problems instead of one. The original problem is back. And you’ve also proven to yourself that your own judgement can’t be relied upon, that you’ll override your own conclusions when the feeling is strong enough.

That second problem is worse than the first. Because the next time you need to make a hard decision, there’s a voice that says: “Why bother thinking this through? You’ll just do whatever you feel like anyway.” The erosion of self-trust doesn’t just affect one decision. It undermines the foundation of every future decision.

What discernment actually looks like

I’m not suggesting that the answer is to become purely rational, to suppress emotion and operate on logic alone. That’s its own form of dissociation, and I’ve done enough of that to know where it leads.

What I am suggesting is something more like a delay. A gap between feeling and acting. Not suppression, but examination.

When a strong emotion arises, particularly one that’s pushing you toward a decision, the practice is to sit with it long enough to identify what’s underneath. Not the surface narrative. The actual driver.

Sometimes the examination confirms the emotion. You feel drawn to something, you examine why, and the answer is: “Because it genuinely aligns with what I want and who I’m becoming.” In that case, follow it.

But often, especially with the really intense feelings, the ones that feel urgent and non-negotiable, the examination reveals something else entirely. A pattern. A wound. A borrowed survival strategy. And once you see it, the urgency often dissolves. Not because the feeling was fake, but because the feeling was serving something you no longer need to serve.

The freedom in not trusting yourself

There’s a strange liberation in admitting that your emotions might be unreliable narrators. It sounds like it should be destabilising, to look at the thing you’ve always used as a compass and acknowledge that it might be pointing at magnetic north rather than true north. But in practice, it’s the opposite.

When you stop treating every feeling as a command that must be obeyed, you give yourself room to choose. You stop being a passenger in your own decision-making. The feelings are still there, still informative, still worth listening to. But they become data rather than directives. Input rather than instructions.

And the decisions you make from that position, examined decisions, decisions where you’ve identified the motivation and chosen deliberately, those are the decisions that build self-trust rather than eroding it. Even when they’re wrong, they were honestly made. And honest mistakes are infinitely easier to learn from than self-deceptions you refuse to look at.

I spent my twenties acting on feelings I hadn’t examined. I’ve spent my thirties examining them. If the pattern holds, I’ll spend my forties acting from a foundation that’s actually mine, rather than one I inherited from circumstances I didn’t choose.

That’s the work. It’s slow. It’s unglamorous. And it’s the most important thing I’ve done.

Subscribe now

There’s a guy I watch on YouTube called Liam Brown. All he does is get a one-way flight somewhere, throw on a backpack, and walk around with his tent for weeks at a time. When he’s done, he flies home. Last year he went from around 60,000 subscribers to 229,000. Monetised. Sponsors lining up. Free gear. The whole thing.

His content strategy? Be a human doing something interesting and film it.

I had something like that once. Not at that scale, but I had a version of it, back when I was an ambassador for Land Rover. It took a long time to build and a lot of work, but when it was working, it was great. Then I abandoned it to sit on a laptop arguing with bots all day. And I keep asking myself: what am I doing?

The attention ponzi recap

I hate to admit this, and I’m genuinely surprised it’s the case, but the state of content platforms is really getting to me these days. Every week it’s something new. My Substack timeline is the same recycled post pasted by hundreds of different accounts. LinkedIn is a graveyard of people writing posts about how to grow your LinkedIn following, which is itself the thing growing their LinkedIn following. The output is the input. There’s no underlying value.

And the advice is always the same: comment on other people’s posts, join engagement groups, be active in communities, like and reply your way to visibility. I refuse to spend any of my life doing that. Not just because it’s a time investment I’m not prepared to make, but because it’s such a cynical exercise. You’re not engaging because you care. You’re engaging because an algorithm rewards the appearance of caring.

The whole thing is pathetic.

The Corsica thought experiment

So here’s where my head went. What if the entire distribution side of content creation could run without me? Not the thinking. Not the conversations. Not the living. Just the mechanical bit: the formatting, the scheduling, the posting, the platform-specific sizing and copy.

Picture this. You’re hiking through Corsica with someone. You’re having a real conversation about something that actually matters to you. You hit record on your phone. That recording gets transcribed. Themes get extracted. Long-form pieces get drafted. Supporting posts get written. Platform-specific assets get generated. Everything gets scheduled and pushed out. You never log into a single platform.

You come back from a year of walking around the world, and the whole time your ideas have been reaching people. Not because you were gaming an algorithm, but because the ideas were worth hearing and the system handled the rest.

That’s the version of content creation I’m interested in.

The real bottleneck isn’t distribution

What struck me about Liam Brown’s growth isn’t the numbers. It’s what he didn’t do. No growth hacks. No engagement pods. No AI-optimised posting schedule. He just went and did something genuinely interesting, documented it honestly, and the audience came.

The lesson isn’t “go hiking and you’ll get famous.” The lesson is that the impediment for most people’s content isn’t distribution. It’s having something worth distributing. The entire growth industry has it backwards. They’re optimising the pipe when the problem is the water.

If you automate the distribution side of things, you free yourself to focus entirely on the input: living an interesting life, having real conversations, building stuff, reading widely, thinking carefully. That’s where the best content comes from. Not from a laptop. Not from a content calendar. Certainly not from an LLM. From being a person who’s paying attention to the world.

What automation should actually do

I think there’s a version of this that’s almost within reach. The pieces exist. Transcription is solved. Theme extraction is getting better. Content generation from raw conversation is functional, if imperfect. Scheduling tools exist. Image generation exists. The integration layer between all of these is the remaining gap, and it’s closing.

The critical thing is what the automation is for. Most people think about content automation as a way to produce more. More posts, more frequency, more platforms. That’s the wrong frame entirely. The point of automation isn’t to produce more content. It’s to remove the creator from the parts of the process that drain them, so they can invest that energy into the parts that actually matter.

Record a conversation. Walk away. Let the machine handle everything between the raw idea and the published post. Come back to the conversation when you have something new to say.

That’s the thesis. Automate the machine so the human can be more human.

The trap I’m trying to avoid

I sort of made a vague commitment to myself recently. At some point this year, I’m just going to pack a bag and go for a very long walk. Genuinely. And maybe create content doing that, like Liam Brown does, just to see where it goes.

The irony isn’t lost on me. The thing I’m most drawn to, the thing that worked for me before, the thing that’s working for other people right now, is the opposite of sitting in front of a screen optimising engagement metrics. It’s going outside and doing something real.

And yet here I am, scrolling through identical LinkedIn posts at midnight, getting annoyed at people I’ll never meet, spending emotional energy on platforms that are designed to extract exactly that from me.

The whole reason I’m interested in content automation isn’t because I want to scale. It’s because I want to escape. I want to build a system that lets me walk away from the screen and still have my ideas reach people. Not more people. Just the right ones.

The uncomfortable middle

I’m not pretending this is figured out. The pieces are close but not connected. The current tools are functional but flawed. And there’s a real question about whether fully automated distribution loses something essential, some quality of presence or intentionality that people can sense.

But I keep coming back to Liam Brown. His content works because he’s present in the experience, not in the distribution. He’s fully there on the trail, fully there with his camera, and then the platform stuff is just the container. The experience is the content. Everything else is plumbing.

That’s the split I’m trying to make. Be fully present in the living and the thinking. Automate the plumbing. See what happens.

A couple of people have messaged me recently saying they’re glad to see me writing again. That’s encouraging. It means the ideas are landing somewhere. Now I just need to make sure the act of getting those ideas out into the world doesn’t become the thing that stops me from having them in the first place.

Subscribe now

There’s a pattern I’ve watched play out enough times to trust it, even though it runs against every instinct most of us have.

Someone you care about is making choices that are visibly, predictably destructive. You can see the trajectory. You know where it ends. And because you care, you want to intervene - to say the right thing, remove the obstacle, engineer a moment of clarity on their behalf.

It almost never works. And understanding why it doesn’t is one of the more useful things I’ve come to appreciate about how people actually change.

The person who avoids discomfort isn’t lazy - they’re overwhelmed

The surface-level read on self-sabotage is always the same: they don’t care enough. They’re not disciplined. They’re choosing comfort over growth.

But the more I’ve observed these patterns - in people close to me, in myself at various points - the more convinced I am that the opposite is usually true. The person who bails on a training session and immediately reaches for comfort food isn’t someone who doesn’t care. They almost certainly care to an unbearable degree. The pizza and the indulgence aren’t indifference. They’re anaesthetic.

What’s actually happening is a relief-seeking behaviour. The feeling of not being good enough, of falling short, of confronting the gap between where you are and where you think you should be - that feeling becomes so acute that the only available response is to shut it down. Escape it. Numb it with whatever’s at hand.

The problem is that this works, briefly. And because it works briefly, it reinforces itself neurologically. Each cycle of discomfort → avoidance → temporary relief makes the avoidance pathway stronger and the tolerance for discomfort weaker. It’s not a character flaw. It’s a feedback loop. And once you see it as a loop rather than a choice, you start to understand why willpower-based interventions - “just push through it,” “just stop doing that” - are so consistently ineffective.

The compulsion to escape intensifies with every successful escape. That’s the mechanism. And it’s difficult to interrupt from the outside.

Why intervention backfires

This is the part that’s hardest to accept, especially if you’re someone who sees problems clearly and wants to spare others the pain of learning the hard way.

Trying to prevent someone from making mistakes - even when you can see the mistake clearly, even when you’re right - can challenge their sense of agency and ego in ways that produce the opposite of what you intend. Instead of gratitude, you get entrenchment. Instead of course correction, you get defiance. Not because they’re ungrateful or stubborn by nature, but because the act of being rescued carries an implicit message: *you are not capable of navigating life yourself.*

For someone already struggling with feelings of inadequacy, that message is devastating. And the ego’s response to devastation is almost always to dig in, not to open up.

I’ve seen this enough times to believe it’s close to a rule: premature intervention entrenches the behaviour it’s trying to correct. The timing matters enormously, and the person doing the intervening almost never gets to choose the timing.

The rock bottom problem

Alcoholics Anonymous figured something out decades ago that most well-meaning friends and family members still resist: meaningful change almost always requires the individual to first recognise, on their own terms, that they’ve lost control. Not to be told they’ve lost control. Not to intellectually agree that things aren’t going well. To feel it - in their bones, in a way that can’t be rationalised away or numbed with a quick hit of comfort.

That’s what “rock bottom” actually means. Not the worst possible outcome. Just the point at which the pain of staying the same finally exceeds the pain of changing. And that threshold is different for every person, which is why you can’t engineer it from the outside.

The AA framework is instructive beyond just addiction. The steps that follow the recognition of powerlessness include surrendering the illusion of total control, and then - and this is the part I find most interesting - redirecting attention toward helping someone else. That redirection is doing something psychologically precise: it dissolves the ego’s grip by shifting focus away from the self. When you’re helping someone else, the constant internal monologue of self-judgement quiets down. Not because you’ve conquered it, but because you’ve stopped feeding it.

It’s an elegant mechanism. And it only works if the person arrives at it voluntarily.

The oxygen mask problem

There’s a related failure mode that gets thrown around all the time, but only ever as a truism, a cliche, a throwaway and rarely connects with the target at the level of behavioural change. It’s the one that affects the person watching, not the person self-destructing.

When you’re pouring energy into trying to help someone who isn’t ready to be helped, you’re draining a finite resource. The resentment that builds - and it does build, even if you don’t want it to, even if you feel guilty about it - isn’t a moral failing. It’s a signal. It’s telling you that you’re neglecting your own needs in service of an intervention that isn’t working.

Referencing the cliché deliberately: you have to put your own oxygen mask on first. Not because you matter more than the other person, but because if you don’t, you eventually become a burden yourself. You lose patience. You lose empathy. You lose the very qualities that made you want to help in the first place. And then you have two people struggling instead of one.

I’ve noticed that the resentment often isn’t really about the other person at all. It’s a reflection of frustration with yourself - with times when you’ve done versions of the same thing, when you’ve made things harder for yourself through your own avoidance or stubbornness. Watching someone else repeat patterns you recognise might be uncomfortable precisely because the recognition is too intimate.

What’s actually left

So if intervention backfires, if advice goes unheard, if engineering a crisis is both manipulative and unlikely to work - what’s left?

Two things, as far as I can tell.

The first is presence without agenda. Being available, genuinely, for the moment when the person is ready. Not hovering. Not hinting. Not holding your help over their head as a reminder of how much you’ve offered and how much they need it. Just being someone they know they can turn to when they reach their own threshold, who has always been there, patiently ready. That requires patience on a scale that most people underestimate.

The second is setting your own example. This sounds passive, and in a way it is. But it’s also the only form of influence that doesn’t trigger the ego-defence response. You can’t argue someone into changing. You can’t logic them out of a feedback loop they’re neurologically wired into. But you can live in a way that demonstrates the alternative is real and attainable. Not performatively. Not with the intention of making them feel bad about their own choices. Just by doing the work on yourself, consistently, visibly.

The people who eventually pull themselves out of destructive cycles almost always point to a moment of recognition rather than a moment of intervention. They saw something - in someone else’s life, in their own reflection, in a consequence they couldn’t rationalise away - and something shifted. You can’t manufacture that moment. But you can be the thing they see when it arrives.

The unsatisfying takeaway

The thesis here is simple and uncomfortable: the most caring response to someone you love who is self-destructing is often to step back. Not to abandon them. Not to stop caring. But to stop trying to be the mechanism of their change.

Everything else - the advice, the carefully worded conversations, the attempts to remove obstacles from their path - is more about your need to help than their need to change. And recognising that distinction is itself a form of the same work you’re hoping they’ll do: confronting an uncomfortable truth about yourself rather than reaching for the easier story.

It’s not satisfying. It doesn’t feel heroic. But the evidence, both from formal recovery frameworks and from simply watching how people actually change, points in the same direction.

Set your own example. Be available. Wait.

Subscribe now

There’s a pattern that keeps surfacing across industries, and it starts with subtraction.

When electric powertrains began replacing combustion engines, the obvious gains were immediate: better torque, lower maintenance, reduced emissions. The numbers were decisive. But something was also lost. The mechanical intimacy between driver and machine: the vibration through the steering column, the resistance of a clutch, the audible feedback of an engine responding to throttle input. All of it smoothed out into a silent, frictionless glide.

The car became more capable and less interesting at the same time.

I’ve been thinking about this pattern because the same transition is beginning in software. As AI takes over more of the interface layer, interpreting intent, anticipating needs, collapsing multi-step workflows into single interactions, the traditional UI is being hollowed out in the same way. And the consequences will be strikingly similar.

What electrification actually removed

The mistake most people make when discussing electric vehicles is framing the transition purely in terms of performance metrics. But driving was never purely about numbers. A well-engineered combustion car involved a constant physical dialogue between human and machine. Gear selection was a decision. Engine note was information. The mechanical linkages between your inputs and the car’s response created a feedback loop that was rich, textured, and satisfying to master.

Electrification didn’t just change the powertrain. It removed an entire layer of interaction. And in doing so, it revealed something: much of what people valued about driving had nothing to do with getting from A to B efficiently. It was the texture of the journey itself.

This is why classic cars have become more valuable, not less, since electrification accelerated. A 1960s Porsche 911 isn’t a rational transport choice. It’s a deliberate rejection of optimised efficiency in favour of something that feels like it was made by humans, for humans. The human premium, the value placed on handcrafted, tactile, non-commodity experience, rises in direct proportion to how much the mainstream alternative strips that experience away.

It’s not the first time this has happened. When manufacturing automated textile production, handmade fabrics became luxury goods. When digital photography made images effectively free, skilled photography became more prized, not less. When streaming commoditised access to music, live performance became the premium experience. The pattern is consistent: automation commoditises the output and elevates the craft.

The same pattern in software

For thirty years, software design has been an exercise in organising complexity. Navigation menus, settings panels, dashboards, toggle switches: all of it exists because humans needed structured ways to communicate intent to machines. The interface was the translation layer.

AI changes this fundamentally. When a system can interpret natural language, understand context, remember preferences, and anticipate what you’re likely to need next, the traditional interface becomes redundant in the same way a gearbox becomes redundant when there are no gears. The translation layer dissolves.

This is, in many respects, a genuine improvement. Most UI is compensating for limitations, not adding value. Nobody loves navigating a settings menu. The cognitive overhead of learning a new application’s interface conventions is pure friction. Removing it is real progress.

But here’s what’s being overlooked: as AI commoditises the functional interface, the experiential interface becomes the only remaining differentiator. When every product can understand what you want and deliver it efficiently, the question stops being “can this tool do the job?” and becomes “how does this tool make me feel while doing it?”

The logic chain runs like this: AI removes the translation layer, then functional UI commoditises, then experiential design becomes the sole differentiator, then craft becomes the premium. This isn’t a cultural reaction. It’s market mechanics.

The template already exists

There’s a useful case study in the automotive world. Ferrari’s forthcoming electric vehicle, the Luce. It features an interior designed by Jony Ive (of Apple design fame) and purposefully resists the trend toward the exclusively forward looking, screen-dominated, minimal cockpits that has come to define the EV interior. Instead of stripping things down to a tablet bolted to a dashboard, the design blends digital screens with tactile materials and visual language that draws on decades of Ferrari heritage.

It’s modern and familiar simultaneously. And that simultaneity is the point. The design isn’t rejecting the new technology. It’s insisting that capability and character aren’t mutually exclusive. You can have the electric powertrain’s performance without surrendering the feeling that this object was considered, crafted, and intentional at every level.

This is the template for what happens next in software: adopt the capability, but preserve the character.

Why familiarity is a feature, not a limitation

One instinct, when thinking about the likely direction of AI-native interfaces, is to pursue radical adaptability. Interfaces that reshape themselves entirely based on what you’re doing right now. A dashboard that’s different every time you open it because the system has determined what’s most relevant in this moment.

It’s compelling in theory. In practice, it collides with something deeply human: people want consistency. They want to know where things are. They want the comfort of recognition, the sense that an environment is theirs because they’ve developed a spatial and rhythmic relationship with it.

We navigate physical spaces through landmark recognition and muscle memory. We navigate digital spaces the same way. The feeling of mastery over a tool comes partly from the tool staying still long enough for you to learn it. The most efficient interface in the world fails if it disorients the person using it.

The better approach isn’t radical adaptability. It’s considered evolution, changing what genuinely needs to change while preserving the structural familiarity that gives people a sense of place. This is what good architecture does: a renovated building retains its character even as its systems are modernised. The bones stay. The experience updates.

What this means for anyone building products now

The implication is counterintuitive but increasingly clear: as AI handles more of the functional workload, the competitive advantage shifts decisively toward design, craft, and intentionality.

This cuts against the prevailing narrative, which treats AI integration as primarily a capabilities arms race. More features. More automation. More intelligence. But if every product has access to roughly equivalent AI capabilities (which, given the pace of model commoditisation, is the likely trajectory), then capabilities converge and cease to differentiate.

What remains is the quality of the experience itself. The weight of a button. The pacing of a transition. The typeface. The tone of voice. The sense that a human with taste and conviction made decisions about every detail, rather than optimising for metrics or deferring to defaults. These are the elements that can’t be automated, precisely because they require human judgement about what feels right, a category of decision that sits outside algorithmic optimisation.

This doesn’t mean craft always wins. In price-sensitive markets, or where the task is purely functional, a frictionless-but-generic AI interface may be entirely sufficient. The human premium is not universal. It’s a premium, which means it commands a price, and not every user will pay it. The opportunity is in knowing which users will, and building for them deliberately.

Where this leads

The human premium is going to become one of the defining market dynamics of the next decade, not just in cars and software, but across every domain where digital technology is commoditising the baseline experience.

In content, where algorithmically generated material is already flooding every platform, the writers and thinkers who maintain a distinctive voice will command a disproportionate premium. In architecture, in food, hospitality, in education: anywhere the default is becoming automated, the deliberately human alternative becomes more valuable, not less.

The mistake would be to see this as a temporary reaction. It’s structural. The more frictionless the commodity layer becomes, the wider the gap between commodity and craft. And that gap is where the value accumulates.

The builders who understand this early, who adopt AI’s capability without surrendering design’s character, are the ones who’ll define what quality looks like in a post-scarcity interface world.

The texture is the product. It always was. We’re just about to be reminded.

Subscribe now

There’s a mistake I’ve seen creators/founders/builders/entrepreneurs/leaders make more than any other. More than hiring the wrong person, more than misjudging a market, more than running out of cash. It’s more fundamental than all of those, and it underpins most of them.

It’s the failure to understand how many assumptions are baked into an idea - and then having no system to kill those assumptions one by one.

I’ve seen this everywhere I’ve worked. Across automotive, cleantech, FinTech, VC, PE, Web3, and now in the AI space. The pattern is always the same. A team generates a promising thesis. The thesis gains internal momentum. People start building toward it. And buried inside that thesis are four or five assumptions that nobody has explicitly named, let alone stress-tested. When one of those assumptions turns out to be wrong - and it always does - the entire edifice wobbles, and the team is left wondering what happened.

But there’s something worse than backing the wrong thesis. It’s what happens when an organisation accumulates multiple theses simultaneously and lacks the discipline to kill any of them.

The Accumulation Problem

Most organisations I’ve encountered don’t suffer from a shortage of ideas. They suffer from an inability to eliminate them. The ideas pile up. Each one has merit. Each one has a champion. Each one has a plausible case for why it could work. And because no single idea is obviously wrong, the organisation tries to keep all of them alive.

This is especially acute in crypto/Web3, where the strategic landscape can completely shift quarterly and the absence of proven revenue models means almost everything is speculative. You’re not choosing between a known good option and a known bad option. You’re choosing between five uncertain options, each with a different risk profile, each requiring different resources, and each defended by someone who genuinely believes in it.

The result is predictable. The organisation enters what I’d describe as an endless validation loop. Teams attempt to validate each thesis. Every thesis surfaces problems. Because every thesis has problems - that’s the nature of operating in uncertainty - the validation process never produces a clean winner. And so the loop continues. More conversations. More analysis. More “what about this angle?” meetings.

Eventually, people get exhausted. The energy drains out of the process. And the attitude becomes: “We need to do something, so let’s just do something.”

That’s not strategy. That’s surrender dressed up as decisiveness.

Why This Happens

The root cause, in my experience, is that most teams confuse exploring an idea with validating an idea. Exploration asks: “What could this become?” Validation asks: “What would have to be true for this to work, and can we prove or disprove those things quickly?”

Exploration is energising. It’s creative. It generates excitement. People enjoy it. Validation is tedious, difficult and uncomfortable. It requires you to take your favourite idea and try to destroy it. It requires you to name your assumptions explicitly - not the comfortable ones, but the uncomfortable ones. The ones that sound like: “This only works if institutions actually care about X” or “This only works if we can convince custodians to do something they’ve never done before.”

Most teams skip the explicit assumption-naming step entirely. They move straight from “this is a promising direction” to “let’s start building toward it.” And when you haven’t named your assumptions, you can’t kill them. They just sit there, invisible, load-bearing, and untested, until the structure collapses.

The Morale Tax

There’s a second-order consequence that rarely gets discussed: the impact on the broader team.

When senior leadership cycles through strategic directions without resolution or well-founded conviction - or worse, when they shield the rest of the organisation from the fact that it’s happening - the effect on morale is corrosive. People aren’t stupid. They sense the drift even when they can’t name it. They notice that priorities shift without explanation. They notice that last quarter’s urgent initiative has quietly been deprioritised. They notice the gap between the stated values (transparency, alignment, focus) and the lived reality (ambiguity, shifting goalposts, hedged bets).

Shielding teams from strategic uncertainty might feel protective, but it often produces something worse than anxiety: it produces cynicism. People stop investing emotionally in any given direction because they’ve learned that it will probably change. The organisation retains bodies but loses belief.

The alternative isn’t radical transparency about every internal debate. It’s having a process rigorous enough that, by the time a direction reaches the broader team, it has survived genuine scrutiny. The team doesn’t need to see the sausage being made. They need to trust that the sausage was actually made well, not just assembled from whatever happened to be lying around.

Share

What Assumption-Killing Actually Looks Like

It’s less glamorous than it sounds. The discipline is straightforward, even if the execution requires real honesty.

For any thesis under consideration, you explicitly list the assumptions that must be true for it to work. Not the obvious ones. The ones buried two or three layers deep. The ones that start with “we’re assuming that...” and end with something that makes the room go quiet.

Then you rank those assumptions by two criteria: how critical they are (if this assumption is wrong, does the whole thesis collapse?) and how testable they are (can we get signal on this within weeks rather than months?).

Then - and this is the part most teams resist - you design cheap, fast tests for the most critical and most testable assumptions. Not pilots. Not MVPs. Just: what’s the fastest way to get evidence that this assumption holds or doesn’t?

When an assumption fails its test, you kill the thesis. Not reluctantly. Not “let’s revisit it next quarter.” You kill it, free up the resources, and move to the next one. The emotional difficulty of this is exactly why most teams don’t do it. Killing an idea feels like failure. In reality, it’s the opposite. It’s the only way to concentrate resources on the ideas that have survived genuine scrutiny.

The Crypto-Specific Version of This Problem

Crypto amplifies the accumulation problem for a specific structural reason: the absence of revenue clarity. In industries with established business models, strategy debates are bounded by economics. You can model the revenue case. You can look at comparable companies. You can run the numbers.

In much of crypto, there is no revenue, or the revenue model is speculative. Blockchains are commoditised. Token prices are decoupled from fundamentals. The industry has spent years in a mode where developers build for other developers and nobody asks where the money comes from, because token appreciation masked the absence of a business model.

When that mask slips - as it does in every downturn - organisations suddenly have to confront questions they’ve been deferring. And they discover that they have five or six strategic directions, each speculative, each requiring significant resourcing, and no framework for choosing between them. The result is exactly the paralysis described above, except with the added pressure of needing to survive a market that’s no longer forgiving.

The teams that navigate this well are the ones that accept a hard truth early: in the absence of proven revenue models, your strategic process matters more, not less. You can’t rely on market signals to tell you which direction is right. You have to do the unglamorous work of naming assumptions, testing them, and killing the ones that don’t hold up. The process is the competitive advantage.

Building the Muscle

This isn’t a framework you implement once. It’s a muscle you build over time. It requires a culture where attempting to kill or killing an idea is celebrated rather than mourned. Where the person who identifies the fatal assumption is thanked, not resented. Where the question “what would have to be true for this to work?” is asked reflexively, not reluctantly.

I’ve been guilty of the opposite more times than I’d like to admit. Falling in love with a thesis, ignoring the wobbly assumptions underneath it, pushing forward on momentum rather than evidence. Every significant mistake in my career traces back to that pattern in some form.

The discipline isn’t complicated. Name your assumptions. Test the critical ones fast. Kill what doesn’t survive. The hard part is doing it honestly, repeatedly, when the ideas you’re killing are the ones you most want to be true.

To help with this challenge, F3, my venture studio, created a free tool designed to pressure-test whether a business idea addresses a real problem and if the proposed solution is credible. We call it ‘Gary,’ a name that’s a confusing nod to both Y Combinator CEO, Gary Tan, and a young illegal immigrant in the classic sitcom, Only Fools and Horses, who is found in Denzil’s lorry by Del and Rodney. They hide him in their flat and nickname him “Gary,” as it’s the only word he says.

Gary operates as a structured validator and thinking partner (ours, not the Only Fools and Horses one), not a brainstorming assistant - he won’t come up with ideas for you and he won’t research and validate ideas for you either. His sole objective is to challenge assumptions and design tests to extract evidence for any given idea. While Gary should not be considered the final word on assumption testing and idea validation, over 40 users have found it a useful tool for establishing problem legitimacy, customer segment precision, solution credibility, evidence of demand and moreover, gaining a practical understanding of what it really means to try and kill and idea to make it stronger.

Gary is free to use, and you can find him here. Give it a try and let me know what you think in the comments!

Subscribe now

I’ve been noticing something in myself lately that I suspect isn’t unique to me. After weeks of relentless screen time - calories restricted, training hard, eyes burning from too many hours staring at displays - I hit a wall. Not a dramatic one. Just a quiet reluctance to open another tab, scroll another feed, engage with another piece of content that felt like it had been generated rather than made.

The timing coincided with a conversation I had with an old friend who’s started a company producing ultra-premium handmade leather goods. Everything bespoke. Everything physical. Every stitch placed by a human hand. And something about that conversation crystallised a thesis I’d been circling for a while.

I initially called it the “analogue backlash.” Then I caught myself. Backlash is reactive. It’s negative. It frames the movement as being *against* something. What I’m actually observing is something more interesting - it’s a movement *toward* something. A revaluation. A rediscovery.

So I’m calling it what it actually is: an analogue renaissance.

The saturation problem

We’ve reached a peculiar inflection point with digital content and AI-generated material. Not the inflection point that technologists talk about - the one where the models get smarter and more capable. The other one. The one where the average person starts to feel, on some barely conscious level, that everything they’re consuming has been touched by something inhuman.

That feeling doesn’t arrive as a fully formed thought. It arrives as fatigue. As a vague sense of distrust. As the creeping suspicion that the LinkedIn post you just read was written by the same system that wrote the last forty LinkedIn posts you read. The words are fine. The structure is competent. But something essential is missing - some quality you can’t quite name but whose absence you can absolutely feel.

This is what happens when authenticity becomes scarce. It doesn’t disappear overnight. It gets diluted, gradually, until one day you realise you can’t remember the last time you encountered something that felt genuinely, irreducibly human.

Scarcity flips the value equation

Here’s the thing about scarcity: it doesn’t care about your intentions. The moment something becomes rare, it becomes valuable - whether or not anyone planned it that way.

For decades, the scarce resource was information. Access. Distribution. The entire digital revolution was built on making those things abundant. And it succeeded, spectacularly. But abundance in one dimension creates scarcity in another. When everyone can publish, curation becomes valuable. When AI can generate a thousand articles overnight, the handwritten letter becomes remarkable. When every brand has a polished, algorithmically optimised presence, the one that feels rough and real and human stands out like a bonfire in a field of fairy lights.

This is not a sentimental argument. I’m not suggesting we all retreat to typewriters and carrier pigeons. The analogue renaissance isn’t anti-technology - it’s a correction in what we’ve decided to value. Technology made distribution free. In doing so, it accidentally made authenticity expensive.

My friend’s leather goods company is a useful illustration, not because leather goods are inherently special, but because of what the business represents. Every product takes hours to make by hand. There’s no scaling that with AI. There’s no shortcut. The constraint *is* the value proposition. And the customers arriving at his door aren’t Luddites - they’re people who spend their entire working lives in digital environments and are desperate to own something that carries the unmistakable weight of having been made by a person who cared.

Share

Credibility as the new luxury

The analogue renaissance extends well beyond physical products. I think it’s reshaping how we evaluate credibility itself.

Consider what “credible” used to mean in a content context. It meant well-researched. Well-sourced. Authoritative. Those qualities still matter, but they’re no longer sufficient. Because AI can produce well-researched, well-sourced, authoritative content at scale. The baseline has shifted. Credibility now requires something that machines cannot yet replicate - the texture of lived experience. The specific, ungeneralisable detail that could only come from someone who was actually there, actually thinking, actually wrestling with the idea rather than predicting the next likely word in a sequence.

This has real consequences for anyone operating in the ideas space. The people who will compound trust and authority over the coming years are the ones whose work carries that unmistakable human signature - not because they refuse to use AI, but because the core of what they produce is rooted in genuine thought, genuine observation, genuine experience. The tools don’t matter. The origin does.

I keep coming back to this framing: it’s not about being anti-AI. It’s about being pro-reality. The two positions might look similar from a distance, but they lead to very different places. The anti-AI position is defensive, nostalgic, and ultimately futile. The pro-reality position is forward-looking. It says: *given that synthetic content is now abundant and free, where does that leave the things that are irreducibly real?* The answer is: it leaves them more valuable than they’ve been in decades.

The cultural undercurrent

I don’t think this is a niche observation. The signals are everywhere once you start looking.

Vinyl record sales have been climbing for years - not because vinyl sounds better (the debate is endless and largely beside the point), but because people want to hold the thing. They want the ritual. They want the object that exists in physical space and cannot be deleted by a terms-of-service update.

Independent bookshops are thriving in cities where Amazon delivers in hours. Farmers’ markets are busier than they’ve been in a generation. There’s a quiet, persistent migration toward things you can touch, smell, verify with your own senses. And critically, this migration isn’t being driven by people who reject technology. It’s being driven by people who are saturated by it.

The renaissance framing matters because it captures the generative quality of what’s happening. A backlash burns itself out. A renaissance builds something. The original Renaissance wasn’t a rejection of the medieval period - it was a rediscovery and revaluation of classical ideals through a new lens. That’s precisely what’s happening now: a rediscovery of analogue values - craftsmanship, presence, tangibility, human authorship - through the lens of a society that has lived through the digital revolution and come out the other side asking what it lost along the way.

Where this goes

I don’t have a neat prediction for how this plays out at scale. I’m suspicious of anyone who does. But a few things seem directionally clear.

First, the premium on human-made will continue to climb. Not because human-made is inherently superior in every dimension, but because it will become the credible signal in a world drowning in synthetic output. “A person made this” will carry the same weight that “organic” or “fair trade” carries now - a marker of values as much as quality.

Second, the brands and individuals who figure out how to be genuinely, verifiably authentic will have an asymmetric advantage. Not performed authenticity (if there’s one thing I’m sure of, this movement will produce a lot of grifters) - that’s just marketing with a different coat of paint. Actual authenticity, grounded in real experience, real craft, real thought. The kind that’s expensive to fake and impossible to scale through automation.

Third, and this is the part I find most interesting: this creates an entirely new relationship between technology and craft. The most compelling future isn’t one where you choose between the digital and the analogue. It’s one where technology handles distribution, logistics, and access, while the human handles the part that actually matters - the making, the thinking, the creating of things that carry weight precisely because a person chose to invest their finite time and attention in them.

The analogue renaissance isn’t a retreat. It’s a rebalancing. And for anyone paying attention, the opportunities inside that rebalancing are substantial - not because authenticity is trendy, but because scarcity always, eventually, dictates value.

Subscribe now

There’s a concept in quantum mechanics that most people encounter as a thought experiment about a cat in a box, and then never think about again. Superposition - the idea that a particle exists in multiple states simultaneously until it’s measured. The cat is alive and dead until you open the lid.

It’s become a cultural shorthand for uncertainty, which is a shame, because the actual problem underneath is considerably more interesting than the metaphor suggests. And considerably more unsettling.

The standard account goes like this: a quantum system exists in superposition - multiple possibilities coexisting - until a measurement occurs, at which point the system “collapses” into a single definite state. Before measurement, the particle doesn’t have a position. It has a probability distribution. After measurement, it has a position. The act of looking changes what’s there to be looked at.

This is not a thought experiment. It’s empirically verified, repeatedly, across decades. The double-slit experiment demonstrates it cleanly: photons behave as waves when unobserved and as particles when detected. The measurement changes the outcome. Not metaphorically. Physically.

And this is where things get genuinely uncomfortable for anyone who assumes there’s an objective reality sitting underneath everything, independent of whether anyone’s paying attention to it.

The gap between the maths and the world

The wave function - the mathematical object that describes a quantum system in superposition - is spectacularly good at predicting outcomes. It tells you with extraordinary precision what you’ll find when you measure. In terms of practical utility, quantum mechanics is the most successful theory in the history of physics.

But here’s the thing that doesn’t get discussed enough outside of physics departments: the collapse of the wave function is a mathematical description, not an observable event. Nobody has ever watched a wave function collapse. You set up a system, you measure it, and you get a result consistent with the predictions. But the moment of transition - from probability to actuality, from superposition to definite state - is not something anyone has directly observed or can explain mechanistically.

This is not a minor gap. It’s the central unsolved problem in the foundations of quantum mechanics: what constitutes a “measurement”? What counts as an observation? And why does the act of observing appear to determine reality rather than simply reveal it?

Physicists have been arguing about this for a century. The Copenhagen interpretation says don’t worry about it - the wave function is a calculation tool, reality is what you measure, and asking what happens “between” measurements is not a meaningful question. The Many-Worlds interpretation says there is no collapse - every possible outcome happens, each in its own branching universe. Decoherence theory explains why quantum effects disappear at macroscopic scales but doesn’t fully resolve what happens at the moment of measurement.

None of these is settled. None commands consensus. The maths works. The interpretation remains an open question.

Where consciousness enters the frame

The uncomfortable territory - the place where physics starts blending into philosophy whether anyone wants it to or not - is the role of the observer.

In most of physics, the observer is irrelevant. Gravity works whether you’re watching or not. A chemical reaction proceeds regardless of who’s in the lab. The universe operates according to laws that don’t require witnesses.

But quantum mechanics introduced something that doesn’t behave that way. The measurement problem specifically asks: does something have to be “looking” for reality to resolve into a definite state? And if so, what qualifies as a “looker”?

A detector? A camera? A human being? Does consciousness play a role, or is it entirely mechanical - any physical interaction sufficient to count as observation?

This is where people get dismissive, and understandably so. The history of “consciousness causes collapse” is cluttered with new-age appropriation, people who watched one documentary about quantum mechanics and decided it validated their spiritual beliefs. The serious version of the question has suffered from its less rigorous enthusiasts.

But the serious version remains serious. John von Neumann, one of the most rigorous mathematicians of the twentieth century, argued that the chain of physical interactions in a measurement has to terminate somewhere - and the only candidate for that termination point is the conscious mind of the observer. Eugene Wigner, a Nobel laureate in physics, took a similar position. These weren’t mystics. They were people who followed the logic of quantum mechanics to its conclusions and found consciousness waiting there.

The counterarguments are strong. Decoherence provides a mechanism by which quantum superposition effectively disappears at macroscopic scales through interaction with the environment, no consciousness required. Most working physicists operate perfectly well without invoking consciousness at all. The question is whether “operates perfectly well” and “fully explains what’s happening” are the same thing. They might not be.

Subscribe now

The construct of reality

The deeper question that all of this points toward: is there an objective reality independent of observation? Or does observation participate in constructing what we call reality?

This sounds like the kind of question that belongs in a philosophy seminar, not a physics lab. But quantum mechanics forced it into the physics lab whether the physicists liked it or not. The measurement problem isn’t a philosophical abstraction bolted onto science. It emerged from the science itself.

If superposition is genuinely what’s happening - if particles truly exist in multiple states until observed - then the idea of an objective, observer-independent reality sitting underneath everything becomes difficult to maintain in its simplest form. At minimum, reality at the quantum level appears to be relational: it depends on the interaction between system and observer. What exists is a function of what’s being measured by whom.

This doesn’t mean reality is subjective in the everyday sense. Your desk is not flickering between existing and not existing based on whether you’re looking at it. Decoherence ensures that macroscopic objects behave classically. But the substrate underneath - the fundamental layer of physical reality - does not appear to be a fixed thing that’s simply there, waiting to be discovered. It appears, at least partially, to be constituted through the act of observation.

And that changes the nature of the question from “what is reality?” to “what is the relationship between consciousness and reality?” - which is a question that physics alone cannot answer, because physics has no theory of consciousness. It can describe every interaction between particles with extraordinary precision. It cannot tell you what it’s like to be the thing doing the observing.

Why this matters outside of physics

The reason this isn’t just a puzzle for specialists is that it touches the deepest assumption most people carry without examining: that there’s a world out there, fixed and definite, and that your job is to perceive it accurately.

If quantum mechanics suggests anything, it’s that the relationship between observer and observed is more entangled than that. Not in the mystical sense. In the empirical sense. The data shows that measurement affects outcome. The question is how far up that principle scales, and whether consciousness is somewhere in the chain or merely a spectator to a process that would proceed identically without it.

That question doesn’t have an answer yet. It may not have one for a long time. But it’s worth sitting with, because the assumption it challenges - that reality is independent of the mind observing it - is the kind of foundational belief that shapes everything downstream from it. How you think about knowledge. How you think about certainty. How you think about the difference between what you observe and what’s actually there.

The cat in the box was always the least interesting part of the thought experiment. The interesting part is the question it was designed to ask: does opening the box reveal the cat’s state, or does it create the cat’s state?

After a century of quantum mechanics, that question remains genuinely open. Which, depending on your tolerance for ambiguity, is either deeply unsatisfying or the most interesting thing about being alive in a period where the foundations of reality are still being negotiated.

Share