Precision and accuracy are not the same

Most of us know two things about pi: its decimal part never ends, and we use it to calculate circles. Its well-known approximation, 3.14, is a useful place to start discussing the difference between precision and accuracy, and how that should impact the way we think about GenAI.

3.14 and 1.34 are equally precise numbers. Commenting on their precision only requires that we compare the number of decimal points they have, and does not require any reference to the value they’re trying to represent.
1.34, however, is a very inaccurate representation of Pi, since it doesn’t even match the first number.

In other words, a number’s precision, on its own, does nothing to inform its usefulness for a particular task. To determine how useful it is, we have to understand its accuracy as well as its precision. Making the distinction between the two is not always as easy as comparing 1.34 and 3.14.

The difference between 3.14 and 1.34 is 57.3%. It’s very easy to spot and will show up in your calculations right away. Take a circular pool with a 10-meter radius, and apply 2 x pi x radius to calculate its circumference, the difference between 2 x 3.14 x 10 (62.8 meters) and 2 x 1.34 x 10 (26.8 meters) is stark. You could intuitively walk around that pool and realize its circumference is a lot more than twice as long as its radius.
The difference between 3.14159 (the correct value for pi, to that decimal point) and 3.14278 (an equally precise, but incorrect value) is 0.038%. Your circumference calculation now yields 62.8318 meters versus 62.8556 meters, which is a lot less intuitive to detect as the wrong result.

The closer the relationship between accuracy and precision, the more difficult it is to spot the difference.

Make your numbers bigger, however, and the error from the wrong value of pi becomes a lot more noticeable. Using 6,371 km as the radius of planet Earth, and assuming for our purpose that it’s a perfect sphere, our calculation with the high-precision wrong value of pi yields a circumference number that is off by 15.2km.

The larger the scope, the broader the consequences of a lack of accuracy, even with high precision.

Fluency is not fidelity

When ChatGPT first burst on the scene, its first impressive trick was its ability to generate grammatically correct sentences and paragraphs. Whether or not you agreed with its interpretation of the history of the French Revolution, you couldn’t dispute that the way it wrote about it used proper grammar and proper structure.

GenAI has been fluent from the start. As it evolved into a tool that could generate programming language code, it kept its fluency in that domain as well.

But the challenge with writing code was never its syntax. It was never its adherence to the rules of a specific programming language. Programming languages, after all, were created specifically to have limited vocabulary and rules that are easy to follow and enforce.

The challenge has always been in making sure that the code actually implements the right solution to the right problem. Taking our circumference calculation example, I could write a program that uses 10 x pi x radius as the formula for circumference, instead of the correct 2 x pi x radius formula we used earlier. That code would compile and run cleanly. It would also produce the wrong answer, no matter how accurate our value of pi.

GenAI’s precision does not guarantee its accuracy. The interface we use to close that gap is natural language, and that is where the next problem starts.

Why natural language cannot close the gap

No matter how we interact with computers, the levels of abstraction ultimately translate to zeros and ones that computer chips can understand and execute natively.

Historically, the translation layers have been “deterministic”, which means that a specific set of instructions in a programming language will always translate to a predictable, fixed sequence of zeros and ones for the computer to execute.

LLMs are different from traditional software in two ways: their input is natural language rather than a programming language, and their output is probabilistic rather than deterministic. Both matter. This essay is about the first.

In 1978, several decades before LLMs were invented, Dijkstra wrote a short essay called “On the Foolishness of Natural Language Programming”. His main claim was that natural languages are inherently imprecise. The argument has been recycled every few years since, but the precision of LLM output now makes it harder to see why it matters. Here is an example.

Let’s consider the simple question: “Is James angry?”, which is a very reasonable question that you might ask before you try to have a delicate conversation with James. If he’s angry, you might wait until a later time. But in programming terms, it’s a question that is not answerable in a deterministic way without a lot of additional context. Who is James? What usually makes him angry? How does that manifest through his behavior?

If I had to implement this question with a programming language, I’d need to define the entity “James” and the state of “anger”. In doing so, I would map very specific attributes to both, and my program would manage those attributes over time, such that for any given timeline of my program, the entity James would be fully known, and its state of “anger” could be calculated precisely.

A natural language does not impose such discipline. I can simply tell an AI agent to perform a task on the condition that “James is angry”, and the LLM will assume what “angry” means in this context. And the inherent subjectivity of natural languages makes it difficult for me to be precise in my description of what makes James angry. He may be angry when he’s tired, when he’s frustrated, when he’s impatient. Natural languages have extensive vocabulary, and each adjective I use to try to explain James’s state of anger invites further context and definition.

For programming, having a limited vocabulary is a feature, not a bug. It forces the programmer to either simplify their instructions, or to build complex instructions carefully from simple building blocks.

The one-shot fallacy

In numbers, the distinction is precision and accuracy.

In writing, it is fluency and fidelity. An essay about the French Revolution can be fluent and wrong. Confident sentences. Clean structure. The right vocabulary. None of it tells you whether the essay is faithful to what actually happened in 1789. Fluency is a property of the surface. Fidelity is a property of the relationship between the surface and the thing it claims to describe.

In programming, the distinction is between code that runs and code that’s right. Code can run cleanly and solve the wrong problem. Cleanliness is a property of the syntax. Rightness is a property of the relationship between the code and the problem it claims to address.

LLMs are fluent in the same natural languages we are. That fluency makes their output feel correct, but correctness is not usefulness, and the gap has to be closed deliberately.

Pi has a true value. Approximations can be judged against it. The assumption with LLMs is the same: better models get closer to the right answer. But that assumption breaks, because the problems we try to solve with programming have no absolute true value. The right answer has to be constructed. That construction is the disambiguation process. It asks questions, surfaces assumptions, and builds context until the problem is defined well enough to solve.

A one-shot prompt skips the construction step. It doesn’t approximate a known target. There is no known target. It generates a target and an output simultaneously, fluently, and with no mechanism to check whether either is faithful to anything.

What’s worse, the fluency masks the absence of the process. The better the model, the better the fluency and the better the illusion.

AI has a metric problem.

I want to discuss three metrics, and why focusing on them is distracting people from the real work that needs to be done.

Benchmarks

Let’s make this one short. Benchmarks were good in the beginning because they provided a simple way for people to 1) compare models to each other and 2) compare the newer version of a model to its previous versions. The more advanced these models have become, however, the more they have either maxed out specific benchmarks to the point where a 3% increase from one version to the next is not meaningful when both models score in the 90% range. Or the AI companies have started training their models to improve their results on specific benchmarks, yielding benchmark gains that do not reflect real-world improvement.

Either way, benchmark results alone are no longer a reflection of the expectations you should have for a specific model when you implement it for your specific use cases.

Speed

For speed, I want to ask a seemingly simple question:

“At what point does our ability to create outstrip our ability to be thoughtful about what we’re creating?”

I think we’re already here. You can write entire essays, create music and videos, and make software applications in hours. How thoughtful can you be about the subject of an essay in a single day? How thoughtful can you be about an end user experience for your software application in a single day?

Speed is almost always the wrong metric. Even in car racing, which is one of the human activities most singularly focused on speed, focusing on speed alone is a recipe for disaster. Race cars have brakes for a reason. “Fast is slow and slow is fast” - even race cars have to slow down in order to go fast.

When we take all the things we can do with AI and try to optimize for speed, we are like a race car driver trying to go through a hairpin turn at 100mph. We give ourselves the illusion of efficient speed, all the way up until we realize we can’t actually move the car in the direction it needs to go in without slowing down first.

Let’s translate that into the Software Engineering world. Optimizing for speed would have me, as a Solution Architect, talk to my customer once. Gather as much information as I can in a single session, and then start a vibe coding session with my AI to create a piece of software to solve my customer’s problems. In this scenario, my AI might ask questions I don’t have answers to. If I’m optimizing for speed, I will want to make assumptions to answer those questions, rather than take time to schedule another meeting with my client. The more assumptions I make, the faster I’m able to go, so I keep making assumptions. And before I know it, I have a fully functional piece of software that has made tens if not hundreds of wrong assumptions about what my client really wants. I might have a working piece of software, but I certainly don’t have one that meets my client’s requirements, since I never bothered to ask after my initial conversation.

Cost

The first issue with cost as a GenAI metric is that cost per token is misleading. The Stanford AI Index will tell you that the cost per token of a GPT 3.5-level model has gone down from $20 per million tokens in November 2022 to $0.07 per million tokens in October 2024, representing a 280-fold reduction in cost.

But here is the thing. You wouldn’t use a GPT 3.5-level model for most of your GenAI use cases in April 2026. As the models have evolved, so have our use cases and our needs for increasing levels of intelligence. In April 2026, Anthropic’s latest model costs $25 per million output tokens and $5 per million input tokens, so overall on the same cost scale between input and output tokens as those GPT 3.5-level models from November 2022.

What makes this worse is that you actually need more tokens in order to handle your more sophisticated use cases in April 2026 than you needed in November 2022. Not only has the per-token price not really gone down, the actual cost of carrying out specific tasks has gone up because the relative complexity of those tasks has gone up and each task actually requires more tokens.

So you’ll at least have more reliable numbers if you measure your cost by use case instead of by tokens, but I would still argue that cost is the wrong metric to focus on.

We cannot talk about cost without talking about value.

$1,000/month would be very expensive for a telephone line because a) most of us don’t need a landline and b) the service is so commoditized that it would be easy to find much cheaper options. So the service has little value, and there are plenty of people willing to provide that small value at a correspondingly small price, therefore the $1,000/month cost can be considered high.

What’s the equivalent reasoning for GenAI costs? In the US, $200/month buys you the most expensive subscription plan that either Anthropic or OpenAI have to offer. For the sake of this discussion, let’s assume you’re able to max out this subscription and need to get 4 more subscriptions to do all the things you’re going to do with AI. That’s $1,000/month.

Is that expensive?

Well, what are you using it for? At $100/hour, you’d need to save 10 hours/month in order to have those subscriptions pay for themselves. $100/hour in the US is a reasonable rate for several roles I rely on regularly: an editor like one I might use for this blog. A website administrator like one I’d use to manage my web site. An Excel guru like one I might have used for the complex pivot table and reporting charts I created earlier this week. A data engineer like one I might have used to convert raw data for a dashboard. As an IT professional and a consultant, I go through these types of tasks multiple times every day, and I save hours worth of work multiple times every day. Even if that time was only worth $100/hour, I would only need 10 of those hours every month to make $1,000/month worth it. And I wouldn’t come near maxing out a single subscription, let alone five.

The real work

So if benchmarks, speed and cost are not the right things to focus on, what is?

First we have to recognize AI is not a toy, and it’s not a hobby. If you are using AI to solve problems, then the real work is to identify the value being created by solving those problems. That work forces you to answer the critical questions. What problems are you trying to solve? What are the leading indicators that will tell you that you have solved those problems successfully? What objective metrics can you test to validate your solutions? What subjective qualities do you need to consider when evaluating your solutions?

This assumes you already know what you are using AI for. If you don’t, that is the first problem to solve, and it is a harder one than any of these metrics can help you with.

Once you have those answers, you can ignore the benchmarks. When a new model comes out, you test it on the problems you are solving. If it works better, you use it. If it doesn’t, you don’t. If you can’t tell, then you have a new problem at the top of your list, which is to build an eval framework that works in your domain.

Speed and cost should not drive your decisions either.

Speed is already at a point where you cannot max out your speed and stay thoughtful about what you’re actually creating. If you create/build/write as fast as you already can, then you’re not creating/building/writing anything worth consuming. You’re just making a slop factory.

At current inference costs, our ability to create value should outstrip the cost needed to create that value by several orders of magnitude. There are not many white collar jobs that are being charged at less than $100/hour. At that rate, you need to replace just 2 hours of human work per month in order to justify the most expensive subscription OpenAI or Anthropic will sell you. If your people are unable to generate 2 hours of billable value from a full month of tokens, the problem is not the tools. It’s the deployment, the workflow, the training, or the use case itself. And those are the problems worth your attention.

AI’s metric problem is pulling attention away from work that actually matters. Forget benchmarks, speed and cost. The real question is whether your people have the time, the problems, and the tools to do work worth doing.

Capitalism does two things that we usually talk about as if they were one thing.

The first is a method. It’s a rigorous framework for allocating scarce resources toward chosen ends. Capital goes where it’s expected to produce returns. Decisions get made with explicit trade-offs. Nothing is assumed to be free, because nothing is. As a system for thinking about how to deploy limited means, it has no serious rival.

The second is a default. Over time, the method got bundled with an assumption about what the returns should be, namely, more economic output. Produce more. Earn more. Grow more. The bundle is so tight that most people can’t separate the two, and most critiques of capitalism are really critiques of the bundle rather than the method.

I want to unbundle them. Keep the method. Reject the default.

Because there is one scarce resource where we need the method badly and almost never use it, and that resource is time.

The one thing you can never earn back

Every person, regardless of wealth or status, begins the day with 86,400 seconds. No one can stop time from passing, and no one can buy back the time that’s already gone. The next day starts anew, with no balance carried forward.

This is not a motivational observation. It’s a structural one. Time is the only resource you will ever manage where the supply is fixed, the depletion is continuous, and the account cannot be topped up. Money can be earned back. Reputation can be rebuilt. Time cannot.

A capitalist would look at this resource profile and immediately ask two questions. What return am I trying to generate? And is my current allocation producing it?

We rarely ask either question about our own time.

The return is yours to define

Here is where the unbundling matters. When I say “return on time,” I do not mean productive output in a corporate sense. I do not mean career advancement or measurable achievement or anything else that capitalism’s default would smuggle in.

The return is whatever you decide it is. But you do have to decide.

If what you want from your life is deeper relationships with the people you love, that is the return, and every hour should be evaluated against it. If what you want is better physical health, that is the return. If what you want is to make something, to understand something, to rest, to serve, to wander, any of these is a legitimate return, and any of them becomes the criterion against which your allocation is judged.

The method is agnostic. The method does not tell you what to value. The method only tells you that once you have decided what to value, you should allocate your most scarce resource toward it with the same discipline a serious investor brings to capital.

Most thinking about time takes the method from capitalism and keeps the default, so it ends up telling people to optimize their hours for an output they never chose. The result is exhausting, and rightly resented.

Take the method. Leave the default.

This argument has a serious opponent in the author of Four Thousand Weeks, Oliver Burkeman. In his book, Burkeman argues that capitalism’s real damage is the instrumentalization of time itself, and that the exit is to reclaim hours whose value does not depend on what they produce. He is right that not every hour should justify itself by its output. He is right that there is something corrosive about a life in which every moment must pay. And he is right to critique the optimization mindset behind how we allocate time, but in a world of constant external capture, abandoning intentional allocation altogether leaves our time vulnerable to being claimed by others. The alternative to intentional allocation is not freedom. It is extraction. An hour you have not decided on is an hour the feed has already decided on for you. Genuine rest, genuine disconnection, genuine aimless time are not the opposite of allocation. They are among its most valuable outputs, and they require more intention to protect, not less. The choice is not between instrumentalizing your time and leaving it alone. The choice is between allocating it toward your ends and having it allocated toward someone else’s.

The first investment

Some readers will already be objecting. I do not know what my returns are. I have not figured out what I want my life to produce. The framework assumes I have clarity I do not have.

Good. That is the most important thing the framework reveals.

If you cannot name what you want your time to return, then the first and most urgent investment you have to make is the one that answers that question. Not as a weekend exercise. Not as a journaling prompt. As the prerequisite without which none of the rest is possible. An investor who does not know what return they are seeking cannot allocate capital. They can only be acted upon by whoever is selling something.

Consider the person who goes to the gym four times a week, every week, for years, and never gets meaningfully stronger, leaner or fitter. That person is not lazy. They are showing up. They are investing. But they never identified what they wanted the investment to return, and so they never stopped to ask whether their sessions were producing it. The hours accumulated. The return did not. And because there was no defined return, there was no feedback loop telling them the allocation had failed. They could continue indefinitely, mistaking presence for progress.

This is what happens when you invest without naming the return. You do not lose your discipline. You lose your ability to tell whether your discipline is working.

Figuring out your returns is hard. It may take years. It is not a one-time exercise, because the answer changes as you change. None of that is a reason to skip it. On the contrary, it’s a reason to start.

And it has become harder, not easier, to do. The reflective space required to ask “what do I actually want my life to return” has been systematically colonized by products designed to keep you distracted instead of thinking. You cannot answer the question while scrolling. That is not a coincidence. That is a design choice, made by people who benefit from harnessing your attention for their economic gain.

Where the allocation breaks

Once you know what you want your time to return, a second discomfort arrives. Your allocation almost certainly does not match.

The person who says their highest return is time with their children and spends three hours a night on a feed does not have a values problem. They have an allocation problem. The stated return and the actual expenditure have come apart. In the capitalist frame, this would be a scandal. Any fund manager whose deployments drifted that far from their stated thesis would be fired. In personal time, the same drift is treated as normal, and the person feels vaguely bad about it without ever naming what went wrong.

The same pattern runs through corporate life. A manager knows, if they stop to think about it, that their team’s performance is gated by unresolved dynamics they have never properly examined. Understanding those dynamics would produce a return measured in months of better work from the whole team. And yet the hours go to email and Slack, to responses that feel urgent and look like management, while the harder investment, the one with the actual return, never gets made. It is not that the manager does not know what matters. It is that what matters is uncomfortable, illegible, and slow, and the allocation mechanism has defaulted to what is easy, visible, and fast.

What went wrong, in both cases, is not character. It is infrastructure. The allocation mechanism, which is to say your ability to direct your own hours toward your chosen returns, is under continuous attack. Entire industries exist to capture attention before it can be directed. The products you use most are the ones best at doing this. They are not optimizing for your time. They are optimizing for time spent on them. Those are different objectives, and one is directly opposed to your interests.

A ten-minute video that delivers forty-five seconds of useful information is not a mistake. It is a successful execution of a business model that treats your attention as raw material. The model works whether or not you benefit. Your time was the input, not the output.

And the same dynamic operates inside organizations. The infrastructure of modern work is organized to reward the expenditure that produces nothing and punish the investment that produces everything. Responding to messages is visible, immediate, and socially legible as work. Sitting with a team problem for three uninterrupted hours is none of those things. The manager who allocates correctly will look, to the surrounding system, like they are doing less. The market is winning.

This is where accepting the capitalism analogy gets sharper. A capitalist operating in a rigged market does not blame themselves for the losses. They identify the rigging and adjust. The first adjustment available to you is recognition. The mismatch between your stated returns and your actual allocation is not evidence that you are weak. It is evidence that your allocation mechanism is being exploited by actors who profit from the exploitation.

The discipline

None of this is a program. I am not going to hand you a morning routine or a set of rules. That would be reintroducing the default through the back door, prescribing how the return should be generated instead of letting you define it.

The discipline is simpler and harder than a routine.

Know what you want your time to return. If you do not know, make that the first allocation.

Look at where your time actually goes, with the same unsentimental clarity you would bring to a bank statement. Not in the aggregate. In the specifics. Hours, not vibes.

Notice which expenditures produce your chosen returns and which produce someone else’s. Be especially suspicious of the ones that feel restful but leave you more depleted. They are often the most expensive, because they are charged against your account while appearing free.

Accept that some of your time is research. You will spend hours on things whose returns you cannot yet name, and some of those will turn out to matter and some will not. That is not waste. That is how you learn what you value. But research is different from distraction, and the difference is whether you are paying attention to what you are learning.

Treat the rest of your hours the way a capitalist treats capital. Scarce. Allocatable. Accountable to a return. Defended against anyone trying to take it without your consent.

What this is not

This is not a case for optimizing every minute. That would be the default creeping back in. Some of the highest returns in a life come from unstructured time, from boredom, from conversations that go nowhere, from walks without a destination. A good capitalist does not liquidate their reserves. They maintain the conditions under which returns remain possible.

The point is not that every hour must be productive. The point is that every hour must be allocated with intention, consciously, against a return you have chosen. An hour of rest you decided on is not the same as an hour of scrolling that happened to you. They may look similar from the outside. They are categorically different investments.

The difference is whether you were the one making the decision.

Personal capitalism

We have accepted, as a society, that capital should be managed with rigor. We have built an entire civilization around the idea that scarce financial resources deserve careful allocation toward chosen ends.

We have not extended the same courtesy to the one resource that matters more than any other.

Your time is capital. The returns are yours to define. The allocation is yours to discipline. The market you are operating in is rigged against you, and the first act of a serious investor is to see the rigging clearly.

Take the method. Leave the default. Decide what you want your life to return, and spend accordingly.

No one else is going to do this for you. A great many people are already spending your time on your behalf, and the returns are accruing to them.

A colleague of mine recently put a name to a phenomenon that has been front and center in our work and in society recently. Brandolini’s law states that:

“The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.”

In other words, refutation costs more than fabrication. That asymmetry shapes everything that follows. Consider flat earthers and the traction they achieve with baseless claims, even against centuries of scientific evidence. In public health, consider the pervasive anti-vaccine movement, decades after vaccines eradicated deadly diseases.

In the political realm, this produces an odd pattern. Some of the most nonsensical policies and decisions get attributed to 3D chess. The simpler explanations, ignorance or malice, go unexamined.

I want to establish that these two things, the improper attribution of 3D chess genius and Brandolini’s law, are related. Understanding their relationship gives us a blueprint for dealing with habitual liars and scammers.

The mechanism behind the confusion

Brandolini’s Law creates 3D chess attributions. To see how, start with what happens when audiences encounter incoherence from someone they want to continue believing.

Prolific fabricators create exponentially expanding refutation burdens. Their audiences face an uncomfortable choice. They can do the expensive work to verify claims themselves, risking the discovery that they have been trusting someone untrustworthy. Or they can find an alternative explanation for their confusion that protects their existing beliefs and their self-respect.

This is where Brandolini’s Law does its damage. Verification costs real time and cognitive effort. Fabrication costs nothing. A prolific source produces claims faster than any individual can reasonably check them. Verification stops being a neutral option. It becomes a burden the audience has to decide whether to carry.

Most will not carry it. But they still need to resolve the confusion somehow.

This is not simple intellectual deference. It is something more specific. When audiences encounter incoherence from trusted sources, they face competing explanations for their confusion, and they reach for the one that costs them least.

Traditional authorities present complex information. Understanding it takes effort. When people distrust those authorities, complexity stops being a credential. It becomes a liability. The thinking becomes: “This is unnecessarily complicated because they’re trying to deceive me.”

But when trusted contrarian sources make claims that don’t hold up under scrutiny, believers face a different problem. The claims themselves are simple and appealing. The contradictory evidence comes from the same complex institutional sources they have already rejected. So believers must reconcile two uncomfortable facts: their trusted source’s claims don’t make logical sense, yet the rebuttals come from sources they have decided are untrustworthy.

The resolution splits in two. Institutional complexity gets framed as deliberate obfuscation: “They’re making it complicated to hide the truth.” Contrarian incoherence gets framed as strategic sophistication: “He’s playing chess while they’re playing checkers.” This lets believers maintain that they are smart enough to see through institutional deception while crediting their preferred source with genius-level strategic thinking that explains away obvious logical gaps.

The move isn’t “I don’t understand this, so it must be sophisticated.” It’s “I don’t trust the complex explanations from authorities, but I do trust this simple source, so when his claims seem contradictory, he must be operating on a level I can’t quite see.”

Look at what this accomplishes. Faced with a contradictory statement from a figure they trust, supporters have three options: acknowledge the contradiction and lose faith in their judgment, invest significant effort to verify and possibly confirm the contradiction, or assume the apparent contradiction reflects strategic thinking beyond their comprehension.

The third option costs the least and preserves the most. It transforms confusion from a warning signal into confirmation of sophistication. The more incoherent the statement, the more sophisticated the attributed thinking required to explain it.

The flat earth movement demonstrates this process perfectly. It starts with distrust of scientific disciplines people don’t understand: astronomy and physics. Contrarian sources gain credibility through simplified anti-establishment messaging rather than science-based presentation. When these sources make baseless claims, believers get confused because the reasoning offered is nonsensical by design. There is no logical substance to explore, much less understand. But believers want to believe, so they explain away their confusion through the dual mechanism described above. Contradictory evidence gets dismissed as institutional conspiracy. The incoherent reasoning gets attributed to sophisticated truth-telling they are not smart enough to fully grasp. The contrarian source gets credited with playing chess while institutions play checkers.

The feedback loop

This creates a perverse incentive structure. Incoherence is not a bug. In this system, it is the product. Sources who produce clear, verifiable claims face constant fact-checking and pushback. Sources who produce simplistic nonsense get credited with playing 3D chess.

Once the 3D chess attribution takes hold, it becomes self-reinforcing. Every subsequent confusing statement adds to the perceived evidence of advanced strategic thinking. Believers develop a stake in maintaining the attribution, because abandoning it means acknowledging they were wrong not just about the most recent claim, but about their entire assessment of the source’s competence.

The result is a feedback loop where the most prolific creators of falsehoods receive the most intellectual credit. The more nonsense one produces, the more credible one becomes, as long as the premise stays simple enough for people to take on face value. The system rewards quantity over accuracy and confusion over comprehension.

This explains why obviously incoherent figures maintain devoted followings despite producing provably false claims. Their audiences are not consuming the claims as information to be verified. They are consuming the confusion as evidence of sophistication they cannot quite access.

None of this means supporters are unreachable. It does mean the standard approach, more evidence, more complexity, more institutional authority, is doing the opposite of what we think it’s doing. Every additional layer of complexity increases the refutation burden and strengthens the case for trust-preserving attribution. If we want to reach these audiences, we have to work against the asymmetry of Brandolini’s law, not feed it.

A blueprint for response

Understanding this mechanism suggests several practical responses to deal with habitual liars, and more importantly, their supporters.

First, recognize that simplicity is key. The people who consume fake science and other dubious claims are not interested in, nor do they have the capacity to absorb, the complex arguments we might be tempted to press on them to bring them to their senses. Make your counter-arguments simple and make them clear. That also reduces the refutation burden.

Second, certainty is paramount. Science rightly prides itself on its ability to question its findings and constantly challenge itself, but the reasons why that matters are lost on this crowd. Any equivocation is seen as weakness, and is an implicit invitation for some charlatan to come along with a simpler, more certain explanation made up out of whole cloth. This does not mean inventing things we don’t believe or cannot demonstrate. We can have confidence in not knowing the answer to a question. What came before the big bang? Science does not know. What came after the big bang? Science does know, and we should be able to explain it simply and confidently.

Finally, and most importantly, we must come down from our high horses. The people who fabricate these lies are despicable. The people who fall prey to them are a different group. They are often people who have been lied to by their government, their politicians, and the companies that manufacture the products they consume. All those liars have historically been adept at justifying their lies with complicated-sounding explanations. So when someone comes along willing to explain things simply, plainly, and with confidence, some people are just relieved to finally be spoken to without condescension.

I will not use the term “dumbing down” to describe the blueprint here. These people are not dumb. They are not stupid. They are tired, they are busy, and mostly they are weary of intellectualism.

Lead with simplicity and empathy. But remember that the burden of proof still rests with the people making up the new theory. With that in mind, we might challenge a flat earther with this experiment. The next time the sun comes up in your time zone, call or text someone in another time zone, and ask them how long the sun has been up for them. If the earth were flat, the sun would illuminate the entire surface at the same time. A lamp hovering over a flat sheet of paper demonstrates that. When the flat earther’s friend in a different time zone reports a different sunrise time, they will have little choice but to recognize that their theory has a hole that has nothing to do with complicated physics or mathematics.

Not all theories are as easily disprovable as the flat earth, of course. But this blueprint can work. And even if it doesn’t fully convince people to change their view, it might open an important line of communication with people who may appear to have abandoned logical thinking, but have really just fallen prey to habitual liars and exploiters of Brandolini’s law.

There is a saying often attributed to Blaise Pascal:

“I would have written a shorter letter, but I did not have the time.”

The implication is that something appearing shorter, smaller, and less complex actually requires more effort to produce. Brevity is the product of work, not the absence of it.

This concept is more relevant today than it has ever been. We have more technology than ever to create and share content, write and deploy code, generate and publish ideas. And the systems that reward us are not designed for quality. They are designed for volume. The reason is not mysterious. It maps directly onto Pascal’s observation: it is much easier to produce more than it is to produce better.

I need to be honest about my own participation in this. I write longer emails than necessary. I start making a point before I have fully thought through my argument, let alone considered how it looks from a different angle. Volume is the path of least resistance, and I take it more than I would like to admit.

But volume is not even the most damaging cognitive shortcut that Pascal’s observation points toward. The more consequential failure is subtler: stopping analysis at the first appearance of a plausible answer. It gives you the sensation of having reasoned your way to a conclusion without committing to the harder work of following the argument to where it actually leads. You arrive at a pre-drawn conclusion and retrofit a justification onto it, rather than letting the reasoning take you somewhere you did not expect to go.

Here is what makes this failure particularly hard to catch. Pascal’s short letter and a first-plausible answer can look identical from the outside. Both are concise. Both feel resolved. But they are produced by opposite processes. The short letter is the result of genuine effort: distilled, refined, worked over until nothing unnecessary remains. The first-plausible answer is the long letter in disguise: unrefined output, the raw product of stopping before the work is actually done. Simplicity and refinement are not the same thing. Confusing them is precisely how intellectual shortcuts sustain themselves.

Traffic gave me an unexpected test of that distinction. Not the one I would have predicted.

I am allergic to traffic. I am a reluctant long-distance driver, and crawling along a highway for forty-five minutes has a reliable tendency to put me to sleep. But the deeper irritant is what traffic represents. Merging onto the interstate and instantly meeting thousands of people coming from roughly the same place and going to roughly the same destination is a particular kind of depressing. All of us in cars easily capable of seventy miles per hour, spending forty-five minutes covering twenty miles to an office building where we will spend most of the day on calls with people who are not in that building.

There was a brief reprieve during Covid. Remote work and the shutdown of most public spaces meant Atlanta traffic was the lightest I had ever experienced. That led me to think the answer was obvious: remote work eliminates the problem. The reasoning was simple. And surface-level. Most people on the road are going to a building they do not need to be in. The departure times are synchronized by working hour conventions that predate every technology we now have. Remove the obligation, remove the traffic.

I took the time to dig deeper. What I found was more uncomfortable than a complicated answer. It was the realization that my first-pass reasoning was not wrong so much as it was operating at the wrong level of resolution.

Remote workers do not stop driving. They redistribute.

The most persistent finding across nearly a hundred studies on this subject is counterintuitive: remote work tends to increase overall driving among remote workers, not decrease it. The commute trip disappears, but something else takes its place. Errands, school runs, coffee shops, lunch outings. These trips spread across the entire day rather than concentrating in two peaks. Hybrid workers, on average, live farther from work than fully in-office workers, which means their commute trips, when they happen, are longer. The car use did not disappear. It changed shape.

The gains that did materialize were structurally fragile.

Remote work is not a policy. It is an employer decision. That distinction matters enormously. The cities with the clearest congestion improvements, like Raleigh and Salt Lake City, had high concentrations of knowledge-economy workers whose employers sustained flexible arrangements. When those employers changed their minds, the benefits reversed. San Jose saw work-from-home rates drop 33 percent in a single year as major employers recalled their workers. More than half of US urban areas saw traffic delays increase in 2024 compared to 2023. What the employer giveth, the CEO memo taketh away. Any solution contingent on a variable entirely outside the control of transport planners is not really a solution.

Remote work may actually make the structural problem harder to solve.

This is the finding that most disrupted my thinking. The evidence-backed solutions to traffic all depend on density: congestion pricing, dense transit networks, mixed-use zoning, walkable neighborhoods. They require enough people close enough together to make alternatives to the car viable and politically fundable. Remote work accelerated suburban and exurban migration, dispersing exactly that density. It also undermined public transit ridership, which depends disproportionately on commuter trips to job centers. A transit system that loses its ridership base loses its financial and political justification. The knock-on effect is more cars, more congestion, and a higher cost to build the infrastructure that would actually solve it.

The pattern across all three findings has the same shape. The first-pass answer is not false. Remote work does reduce commute trips. But traffic is not a commuting problem. It is a land use problem, a political will problem, an infrastructure problem that expresses itself as congestion. Operating at the level of commute trips and calling it a solution is a category error. It is like treating a recurring fever with cold water. The temperature drops. The infection continues.

Which brings me back to Pascal.

The remote-work answer to traffic felt like the short letter. Concise, resolved, satisfying. It was not. It was the long letter: unrefined, produced quickly, not yet subjected to the work that honest thinking requires. The actual short letter, the one worth sending, turned out to be longer. More research, more nuance, more uncomfortable conclusions. But it was refined. It was worked over. It did not stop where stopping felt comfortable.

The shorter letter takes longer to write, just as the more intellectually honest argument takes longer to think through. The shorter letter’s reward is clarity. The thorough argument’s reward is a better question. The traffic research did not answer the congestion problem. It reframed it: the question worth asking is not how to reduce commute trips. It is whether we are willing to build the density that makes alternatives to the car viable. And a better question is where every useful answer actually begins.

“Just be consistent.”

It’s one of those pieces of advice that sounds obviously right. Almost too obvious to question.

Show up every day. Do the work. Stick with it.

And yet, you can look around and see plenty of people who are consistent, but never get better. Something is missing. Not because consistency does not matter. It matters a lot.

But on its own, consistency is incomplete advice.

When we talk about consistency, we often mean repetition. Do the same thing. Again and again. But repetition alone doesn’t guarantee progress. It can just as easily reinforce stagnation. You can:

• Write every day and not become a better writer

• Go to the gym regularly and not get stronger

• Work long hours and not become more effective

Consistency, in this sense, is neutral. It amplifies whatever you are already doing. If your approach is flawed, consistency doesn’t fix it. It actually compounds it.

What’s missing is intentionality. Consistency determines if you will show up. Intentionality determines what you will do when you do show up.

Intentionality is what turns activity into progress. It’s the difference between:

• Practicing and practicing with feedback

• Working and working on the right things

• Showing up and showing up with intention

It forces you to ask:

• What am I trying to get better at?

• What does “better” even look like?

• Is what I’m doing moving me in that direction?

Without those answers, consistency can actually take you very far in the wrong direction.

The layers underneath

Part of the problem is that we treat consistency as the main behavior, when it actually sits on top of several deeper layers.

Intentionality shapes those layers.

Identity

In psychology, consistency is tied to alignment. We are naturally pulled toward coherence between what we believe and what we do. When that coherence breaks down, the discomfort tends to push us back toward alignment.

Intentionality defines who you are trying to become. Consistency reinforces it.

If you see yourself as “someone who writes,” consistency becomes easier. If you don’t, it becomes a constant uphill battle.

Systems

From a behavioral perspective, consistency is less about motivation and more about structure.

James Clear popularized the idea that habits are built through environment, cues, and repeatable processes.

Intentionality determines what system you build. Consistency determines whether it runs.

As Clear puts it, “you don’t rise to your intentions, you fall to your systems.” But you still have to choose the right system in the first place.

Performance

In performance domains, consistency is not about peaks. It’s about floors.

The best performers are not the ones who occasionally do something extraordinary. They are the ones who reliably deliver a high baseline, even under pressure.

Intentionality is what raises the ceiling. Consistency is what raises the floor.

You need both.

Time

Consistency is what allows time to do its work.

Intentionality determines what time is working on.

Small actions, repeated, create outcomes that feel disproportionate to the effort that produced them. But only if those actions are pointed in the right direction.

Without intentionality, time amplifies noise. With it, time amplifies signal.

Where the advice breaks

“Be consistent” assumes that the underlying layers are already in place.

But often, they’re not.

So people try to force consistency through discipline alone. And when that inevitably fails, they conclude that the problem is a lack of willpower. It usually isn’t.

It’s a lack of intentionality:

• The identity isn’t clearly defined

• The system isn’t designed with purpose

• The work isn’t structured to improve performance

• The feedback loop is missing

Consistency is being asked to carry weight it was never designed to carry.

A better framing

Consistency still matters, a lot. Paired with intention, it creates forward progress. Without it, consistency alone produces stagnation or worse, regression.

We tend to treat consistency like a virtue in itself. But consistency is closer to a force. It will faithfully scale whatever you feed it. Good or bad. Thoughtful or careless. Aligned or misaligned.

Intentionality is what decides what gets scaled.

So yes, be consistent. But be intentional first.

GenAI enablement is not binary, and it is not single-threaded.

Many companies are struggling to get meaningful returns from their GenAI initiatives. Part of the reason is how they think about “enablement”, the process of training and supporting people with GenAI tooling and ways of working. Too often, it is treated as a point-in-time activity. Something to check off and move past.

But enablement is not binary. You cannot go from “I have never interacted with ChatGPT” to “I run swarms of AI agents that push code straight to production” in one session, or even a handful.

It sounds ridiculous when stated that way. Yet many training programs effectively assume exactly that.

Day 1: Prompting fundamentals Day 2: Context engineering Day 3: Agentic workflows Day 4: Onboard onto your project Day 5: Increase velocity by 40% sprint over sprint

This is not verbatim, but it is not far off from what I have seen in Fortune 100 companies. Run a short training, then expect lift-off.

Let’s count the ways this approach falls short.

First, the initial training, while necessary, cannot be compressed into a few days. At best, it establishes a foundation. Cramming this much content into a short window leaves no time to absorb, experiment, or develop a personal perspective on how to apply it.

Second, this training is largely use-case agnostic. It does not consider which use cases the tooling will be applied to, or adjust the approach accordingly. Context engineering, for example, looks very different depending on whether you are doing UAT bug triage or writing user stories.

Third, it treats each capability as a one-and-done. Spend a day on agentic workflows, and you are now “enabled”. That would be like saying that once someone learns to drive, they can handle every vehicle and situation. No distinction between motorcycles, 18-wheelers, race cars, and dune buggies. You are now a “driver”. Therefore, you can drive anything.

Finally, it focuses on increased velocity, which is misguided on multiple levels.

Velocity, the number of story points completed per sprint, was introduced by Scrum and XP in the late 90s. It was never meant to measure productivity. It is a relative measure of a specific team’s capacity over time.

A story point does not represent just time, just complexity, just effort, or just uncertainty. It is a combination of all of those. And crucially, it is relative to a specific team.

If my team estimates a piece of work at 3 points, that only has meaning for my team. Another team might estimate the same work at 5 points and complete it in the same time with the same resources. Both are correct, because the scale is internal.

If points are team-specific, then velocity is also team-specific. That alone makes it unsuitable as a cross-team measure of productivity.

Worse, using velocity this way encourages bad behavior. If my team delivers 100 points per sprint and leadership expects 150, the easiest solution is not to work faster. It is to re-estimate. Turn every 3 into a 5, and suddenly we are delivering 160 points. Nothing has improved except the optics.

In the process, we lose the integrity of story points as a meaningful reflection of time, complexity, effort, risk, and uncertainty within a specific context.

So let’s recap.

We cannot meaningfully change how people work by compressing training into a few days. We should not assume GenAI practices apply uniformly across use cases. We should not assume all practitioners operate at the same level of proficiency. We should not treat velocity as a universal benchmark.

So what should we do instead?

We need enablement motions that treat true enablement as continuous, contextual, and compounding.

Enablement is continuous. People do not learn GenAI the way they learn a tool. They learn it the way they learn a craft. Through repeated exposure, feedback, and iteration over time.

Enablement is contextual. The way a QA engineer uses GenAI should not look like the way a product manager uses it, which should not look like the way a platform engineer uses it. Even within the same role, the approach should evolve based on the problem at hand.

Enablement is compounding. Each layer builds on the one before it. Prompting leads to better context. Better context enables more reliable workflows. Reliable workflows make agentic systems possible. But none of these layers are ever “done”.

And most importantly, enablement is not single-threaded.

It does not happen in a classroom. It happens in the flow of work.

It does not happen once. It happens every day.

It does not happen the same way for everyone. It evolves based on the individual, the team, and the problem space.

The organizations that will get real returns from GenAI are not the ones that run the best training programs. They are the ones that build systems of learning.

Systems that embed experimentation into delivery. Systems that allow teams to share patterns and evolve practices. Systems that treat enablement as an ongoing capability, not a kickoff event.

Because GenAI is not a tool you roll out. It is a way of working that teams evolve into.

Enablement is not an event. It is an operating model.

In a follow-up, I will break down the specific roles and practices you can put in place to jumpstart that evolution and sustain it over time.

What do you write when you have nothing to say?

Wrong question.

Do you ever actually have nothing to say? I don’t think so.

When we don’t write or publish, it’s rarely because we have nothing to say. It’s because we don’t want to publish what we’re thinking.

So the real question becomes: why don’t we want to publish what we have to say?

That is a much more interesting question, and one worth unpacking.

Our heads are full of ideas, but they are often disorganized and fleeting. Writing forces us to sit with those ideas, explore them, and shape them into something coherent. In doing so, we clarify our own thinking.

If writing is about organizing our thoughts, then the idea that we have “nothing to say” doesn’t really hold. We always have something to think through: what’s happening in the world, in our work, in our relationships, or within ourselves.

We don’t lack ideas. We lack the willingness to share them.

Often, that hesitation comes from a simple belief: what I have to say isn’t interesting enough.

But should writing be about other people?

If the primary goal is to think more clearly, then no. Writing is first and foremost for the writer.

But then why publish?

Because in practice, intention is not enough. Without external pressure, writing becomes optional, and optional things rarely get done. A publishing deadline, even a self-imposed one, creates the necessary pressure.

So writing may not be about other people, but publishing is what makes the writing happen.

This raises a fair question: if our thoughts might not interest others, should we still publish them?

Yes — I think so.

It is always worth writing, and publishing, what you feel compelled to explore.

First, because the people most hesitant to publish are often the ones we most need to hear from.

Putting yourself out there is hard. The more self-aware and introspective you are, the harder it becomes. Unfortunately, this means the loudest voices are not always the most thoughtful ones.

If you are methodical, reflective, and precise in your thinking, your voice matters. Don’t let it be drowned out.

Second, because of humility.

You may not (yet) be a great thinker or writer. I’m not. But improvement requires practice. And if publishing is what forces writing, and writing is what sharpens thinking, then publishing becomes necessary.

There is humility in doing something you are not yet good at, especially in public. And that discomfort is productive.

Humility underpins curiosity, openness, and growth. You have to accept that you don’t know everything to learn. You have to believe others might be right in order to truly listen. Writing, and publishing, is a way to exercise that muscle.

Finally, because people care both less and more than you think.

Most people will ignore what you write. They don’t know you, or they’re simply not interested. That’s fine.

Some will engage critically. If it’s in good faith, you can learn from it. If not, you can ignore it.

And a few will resonate with what you’re saying. In addition to yourself, those are the people you’re writing for. They are the ones who will engage, encourage, and help ideas grow beyond you.

You won’t find them unless you publish.

I don’t know who those people are yet. But I’m committed to showing up, week after week, sharing what I’m thinking, and letting that process shape me.

For the humility it demands. For the clarity it creates. And for the possibility of finding others who care about the same ideas.

For those reasons, I resolve to never have nothing to say.

When people think about careers or the kind of work they enjoy, the default way of thinking is vertical.

We tend to ask questions like: What field should I be in? What job title fits me best?

So we sort the world into domains and activities.

A domain might be finance, media, education, travel, or acting. An activity might be analyst, teacher, coach, engineer, or manager.

Most career advice is built around navigating these vertical categories. Find the right domain. Find the right role inside that domain. Then build a career around it.

But there is another way to think about work.

Instead of thinking vertically, we can think horizontally. The 6 Types of Working Genius proposes exactly that: a horizontal lens for understanding work.

Before getting into the framework itself, it helps to imagine a simple picture.

Imagine a cake. Well, imagine a layered cake, like a Black Forest cake.

Each slice of the cake represents a different activity in your life.

One slice might represent your professional work. Another slice might represent volunteering. Another slice might represent kayaking. Another slice might represent woodworking.

Over time, you might add more slices. You might be into kayaking today and pick up photography tomorrow. You might start coaching a team or learning a new craft.

There is virtually no limit to the number of slices.

But the layers of the cake stay the same. Every slice is made up of the same layers. That idea sits at the heart of the Working Genius framework.

Regardless of the domain or activity, every meaningful effort tends to involve the same kinds of work. At a high level, these fall into three phases: ideation, activation, and implementation.

The framework breaks those phases down further into six types of work:

• Wonder

• Invention

• Discernment

• Galvanizing

• Enablement

• Tenacity

Each represents a different contribution to getting something done.

• Someone has to ask the question: Why are we doing this? Could it be done differently? That is Wonder.

• Someone has to propose possible ways forward and generate solutions. That is Invention.

• Someone has to evaluate those ideas and determine which ones are worth pursuing. That is Discernment.

• Once a direction is chosen, someone has to rally people around it and mobilize action. That is Galvanizing.

• As the work progresses, people inevitably get stuck. Someone has to step in, support others, and help remove obstacles. That is Enablement.

• And finally, someone has to make sure the work actually gets finished. The details have to be handled. The last mile has to be completed. That is Tenacity.

The key insight of the framework is that while every meaningful activity contains all six types of work, individuals tend to have natural places where they enjoy spending their time and where the work brings them energy and fulfillment, even when it’s intense and tiring.

Some people love asking the big questions but lose interest once implementation begins. Others thrive in the middle of execution but feel drained when asked to brainstorm endlessly. Still others are happiest finishing the job and making sure everything is buttoned up.

Without a horizontal lens, we often misdiagnose the problem.

Imagine volunteering at your kids’ school. You may initially think you enjoy helping out. But the group might quickly discover that what they really need is someone to take detailed notes during meetings, organize the action items, assign tasks, and track progress.

In the language of the Working Genius framework, that work sits in the implementation phase, specifically Enablement and Tenacity.

If that is not where your natural energy lies, the experience can quickly become frustrating. Without another way of interpreting that frustration, you might conclude that you simply do not like volunteering for the school.

But that conclusion would be misleading.

The real issue might not be the activity itself. It might be the type of work you were asked to do inside that activity. You might actually love brainstorming ideas for the upcoming school fair. You might enjoy recruiting people to help run the booths and rallying parents to participate.

Those contributions fall into Wonder and Galvanizing, which sit in very different parts of the work cycle.

The cake analogy helps make this clearer. Each slice represents a different activity. But every slice contains the same layers.

Now imagine something slightly absurd.

Imagine a cake that is made entirely of one layer.

A cake made only of icing.

It might taste great for the first bite. But very quickly it would become overwhelming. And structurally, it would barely hold together.

Or imagine a cake made entirely of dense base crumbs with no icing, no filling, and no variation. Technically it might still be edible, but it would be dry, heavy, and not very enjoyable.

A good cake needs multiple layers, and each layer serves a different purpose. Together, they create something balanced.

Work operates in much the same way.

If an activity is dominated entirely by Wonder, you may end up with fascinating questions and endless curiosity, but no concrete ideas.

If it is dominated by Invention, you may generate many clever solutions, but struggle to decide which ones to pursue.

If it is dominated by Tenacity, you may execute extremely well, but risk executing the wrong thing.

The balance of layers is what makes the slice hold together.

From Personal Insight to Team Design

Seeing work through this lens is genuinely useful when thinking about our own careers, activities, and hobbies. It gives us a vocabulary to explain something many of us have felt but struggled to articulate. The frustration we feel with certain activities is not always about the activity itself. Often, it is about the type of work we are being asked to do within that activity.

But the usefulness of this framework goes beyond personal self-knowledge.

Most meaningful work cannot be done by one person. The problems worth solving, the projects worth building, and the organizations worth creating usually require groups of people working together over time.

Which means understanding the types of work involved is not just a personal exercise. It becomes a team design problem.

If every meaningful activity contains the same six layers, then every effective team needs to cover those layers as well.

A team full of Wonder can generate powerful questions and insights. Add a few Inventors, and the team might produce an endless stream of clever ideas. But if the team lacks people who naturally thrive in Enablement and Tenacity, those ideas may never move beyond the whiteboard. The insights are there. The potential solutions are there. But the work that actually brings those solutions into reality never gets done.

The opposite imbalance can happen as well. A team dominated by execution-oriented profiles may move quickly and efficiently, but struggle to step back and question whether they are solving the right problem in the first place.

Imbalance in the layers can quietly undermine the whole slice.

This is what makes the Working Genius framework particularly useful for thinking about teams. Most frameworks for assembling teams focus on expertise, personality types, or job titles. Those things matter, but they do not always capture the kinds of work that naturally energize people.

Looking at a team through the lens of Working Genius, we don’t just ask who has the right skills, we also ask who naturally thrives in each layer of the work.

Do we have people who ask the right questions? Do we have people who generate possibilities? Do we have people who discern which ideas are worth pursuing? Do we have people who can rally others into action? Do we have people who support the work when it gets difficult? And do we have people who will see things through to the end?

When those layers are present and balanced, something interesting happens.

The slice holds together. The flavors come together. And the cake becomes much more than the sum of its parts.

The world is moving at a dizzying pace. Whether you think it is moving in the right direction, the wrong direction, or somewhere in between, most people would agree that things feel very fast right now.

I sometimes wonder whether every generation feels that way. We can objectively say that some things, like technology, are moving faster today than ever before. But acceleration is always relative. People in the 1980s may well have felt that the personal computer revolution was pushing technology faster than it had ever moved before.

Similarly, the now-common idea that if you’re not paying for the product, you are the product is often attributed to Andrew Lewis, who used that phrasing in 2010 when referring to Digg.com. At the time, it was largely about free social media platforms and their ability to attract attention with “free” services while monetizing the data and attention of their users.

But trading free products for marketable attention and data did not begin in the 2010s. The idea is at least as old as advertising itself.

That said, there is something unique about the mid-2020s. Over the sixteen years since 2010, the scale of “free” products has become unprecedented. Think about the products you use most: there is a good chance you do not pay for many of them. Most consumer services from Google are free. YouTube is free. Instagram, TikTok, Facebook, WhatsApp, and Spotify are all free or offer free tiers.

As these products have grown, so has the need for their creators not only to capture our attention but also to compete with one another to do so in the most addictive way possible.

It is no wonder, then, that it has become harder for us to think for ourselves.

Thinking requires space. And space requires time. But these companies also require our time, and they are better at capturing it than we are at protecting it.

Space and emptiness both arise from free time, but they are not the same.

Emptiness is what happens when the creation of space is not intentional. It is what happens when we find ourselves bored with time on our hands. It is the state that appears when boredom fills what could have been thinking space.

Even though Aristotle’s saying that “nature abhors a vacuum” has long since been disproven in physics, I would argue that it still holds true for time.

Your time will not remain empty for long. If you are bored and not intentional about how you spend your time, you will naturally fill it with whatever is easiest and most readily available. And the social media companies we discussed earlier are more than happy to fill that emptiness with as much dopamine as you have time for.

But we need time to think. We need moments away from the constant motion of everyday life in order to analyze, review, reflect and, ultimately, think.

So what are we supposed to do if several billion-dollar companies are pounding on the door with their dopamine machines every time we manage to create a little time for ourselves?

Creating thinking space requires intentionality. You cannot simply make time for yourself. You also need a clear plan for how you will protect that time from being filled with empty social media calories.

For me, sports has always been the easiest answer. Whenever I have free time, the first thing that tends to fill it is exercise. Sports conveniently force you to step away from social media, because most sports cannot be practiced while scrolling.

Beyond sports, my more general answer is curiosity.

Curiosity, by definition, comes from within. It begins with a thought and a desire to understand something more deeply.

That process of thinking, questioning, and learning does three important things.

First, it puts you in control. The process begins with your own thought. You decide how to satisfy your curiosity: a book, a conversation, a web search, or even a YouTube video. Yes, it is very different when you search YouTube intentionally than when you let it feed you whatever keeps you watching the longest.

Second, curiosity feeds on itself. Whatever you start exploring today will inevitably lead to more questions. And if you allow yourself to pursue those questions, you may find that several hours have passed learning about something that genuinely interests you.

But most importantly, curiosity cultivates the mind.

When you embrace your curiosity and give it space, you ask questions. When you ask questions, you find answers that lead to more questions. Eventually you realize that some of your questions are uniquely yours, and that the answers to them have not yet been written.

At that point, you go out into the world to engage with it and find those answers.

People who do this consistently rarely experience emptiness when they have free time, because they are not bored. Their time produces thinking space, and that thinking space protects them from one of the defining afflictions of our era: having our brains slowly rotted by social media companies.

With deep curiosity, you can not only protect your free time from these social media companies, you can also leverage the best thing they offer: a remarkable window into the creative minds of people all over the world. At its best, social media lets you see ideas, projects, and perspectives that would have been almost impossible to encounter just a generation ago.

Yes, if you let it, it can also turn you into a mindless scrolling machine. But if you approach it with curiosity, some of that content can become a jumping-off point for your own journeys of learning.

Curiosity is what transforms free time into thinking space instead of emptiness.

As an overall optimistic person, I’ve recently been feeling a quiet sense of dread that doesn’t quite square with what I deeply believe is a moment filled with potential for better days ahead.

As I try to reconcile these two feelings, their contrast becomes clearer. My optimism is rooted in the potential of today’s technology and the promise of what it can deliver tomorrow. The dread, however, is not just vague anxiety. It’s an honest observation of where my mind and psyche are right now.

Many things are heading in the wrong direction in American society. But the one I worry about most is the erosion of reason. Not just the ability to reason, but even the recognition that reason is a foundational skill worth nurturing, celebrating, and rewarding. There is a growing pride in ignorance, an acceptance of it without curiosity or critical thought. It is no longer merely tolerated. It is increasingly treated as a virtue.

It couldn’t come at a worse time.

The way most of us work will change drastically over the next few years. Seeing those changes coming, understanding what they mean, and doing the work to pivot and adapt will require a deep commitment to learning. People who can’t be bothered to fact-check their politicians on basic lies will struggle to adapt to far more complex shifts.

That does spell disaster for some. I have come to terms with the uncomfortable truth that not everyone will adjust in time. We can only hope that the reality of the needed mindset shift becomes clear before it’s too late.

And yet, despite these looming Generative AI-driven changes, I remain optimistic.

Partly because I’ve never believed society owes me a way to contribute. I do believe society should strive to create opportunity and lift up those who need a hand to get started. But I’ve never thought of myself as someone who must be carried forward. I’m a problem solver. I have never been concerned that society would run out of problems worth solving.

As a problem solver, I believe the ability to learn more, do more, and do it more efficiently will never be a bad thing. Strip away the very real concerns about Generative AI, including intellectual theft, energy consumption, and the concentration of power in a few companies and individuals. What remains is a profound truth. Even if large language models never improve beyond where they are today, they represent a massive leap in humanity’s ability to learn, produce, and build.

What problem facing humanity would not benefit from greater knowledge, speed, and productivity?

So when I hold these two emotions side by side, I arrive at a diagnosis.

I am suffering from an optimistic low-grade fever.

Our society is fractured in ways that will be difficult to repair. This has been true for years, perhaps decades. The patterns of injustice we are witnessing today are not new. But we still have to engage. Because we either help build the future or it gets built without us.

And when the future being built is not incremental but transformative, when it changes how we build, how we think, and what we are capable of creating, the stakes rise.

Being left out of the Generative AI revolution will be like missing the Industrial Revolution, the computer revolution, and the internet revolution all at once. Today, you would not attempt to compete without machines, computers, or the internet. Tomorrow, you will not compete without AI tooling.

The low-grade fever persists.

But we cannot let it immobilize us.

Because the future is not waiting for us to feel ready.

Tribe. It is a powerful word. It evokes pride, belonging, and identity. It can be a profound driver of purpose and fulfillment. And at the same time, it is often a source of division, violence, and pain.

I have been thinking about this in the context of identity. I believe there is a way to hold on to our identities, to nurture our tribes, without becoming so tribal about it.

There is a formal definition of tribe that refers to blood ties or religious affiliation within a society. But for this discussion, I want to use a broader meaning. A tribe is a group of people we identify with for whatever reason. A profession. A faith. A generation. A stage of life.

Identity has been on my mind lately because I am going through several transitions at once.

My wife and I became empty nesters recently. That is a significant shift. We lost several tribes almost all at once. The kids themselves, of course. They are not gone, but they are no longer woven into our daily lives the way they were for nearly twenty years. We also lost the orbit around them. Soccer games. Family dinners. Camping trips. The rituals that quietly structure a life.

At the same time, my industry, software engineering and consulting, is changing rapidly. Adapting to that shift forces me to reconsider what it even means to be an IT consultant. And I am nearing the big 5.0. No matter how often I tell myself that age is just a number, this milestone feels more daunting the closer it gets.

With all of that changing, I found myself asking what, exactly, was being lost.

The more I reflected on these identities that seemed to be slipping away, parent of young kids, traditional software engineer, young adult, the clearer it became. I was losing circumstances, not identity.

Two concepts from my field helped me make sense of this. First, the concept of an interface. It defines a contract, and specifies the behaviors something must support, but it does not dictate how those behaviors are implemented. Second, the concept of a class, which by contrast is a specific implementation. It fulfills the contract in a concrete and particular way.

Think of a piece of music. The score defines the notes, the timing, and the structure. That is the interface. An orchestra performs it. They honor the structure, but their instruments, tempo, and emotional tone make the performance uniquely theirs. That is the implementation, the class.

I have come to believe that the healthiest version of identity is an interface, not a class.

That does not mean the contract itself is negotiable or infinitely flexible. An interface still defines something real. The commitments must be durable. The principles must hold. The flexibility lies in how those principles are lived, not in whether they are lived at all. Without a stable contract, there is no identity. There is only drift.

When identity becomes a class, it gets fused to a specific set of circumstances. It becomes rigid. If the circumstances change, the identity feels threatened.

If my identity as a parent were tied to the specific stage when my kids were small and living at home, I would have lost that identity many times already. I would have lost it when they stopped being babies. I would have lost it when they became teenagers. I would have lost it the day they moved out.

If my identity as a problem solver were tied to the specific tasks I performed as an individual contributor, I would have lost it twenty-five years ago when I became a lead and started managing teams instead of writing code myself.

If my identity as a young adult were tied to my twenties, that one would have expired a long time ago too.

But none of that happened.

Because my identities are interfaces. They are not rigid classes frozen in time. They define commitments and principles, but they allow the implementation to evolve.

As a parent, the contract might be love, guidance, availability, and nurturing. That contract can be implemented through bedtime stories and soccer practice. It can also be implemented through phone calls and text messages.

As a software engineer, the contract might be solving problems, building systems, and helping others think clearly about technology. That contract can be implemented by writing code by hand. It can also be implemented by leading teams, designing architectures, or even telling an AI what code to generate.

As a young person, the contract might be curiosity, openness, and a willingness to change. That does not require a young body. It requires a certain posture toward the world.

At the same time, not all identities are chosen, and not all can be shed through reinterpretation. Some identities are inherited through circumstance and cannot be separated from the events that formed them. Nothing I am writing here suggests that a widow can simply choose not to be a widow, or that a person can step outside the lived realities that shaped them.

Thinking of identity as an interface can extend to our tribal affiliations as well.

I am not religious myself, but many of the religious friends and family members I connect with most easily are those who treat their faith as an interface. For them, religion is a vehicle for deeper principles such as compassion, humility, discipline, and service. It is not a rigid class that claims exclusive access to truth.

It is not difficult to see how much conflict arises when identity becomes fused to a particular implementation. When the specific expression of a belief, a job, a culture, or a stage of life is mistaken for the essence of the identity itself.

The Identity Trap is believing that everything about how we live must stay the same in order for who we are to remain the same.

If we instead ask a different question, the perspective shifts. What is the contract? What are the principles that this stage of life is implementing? If we understand that, then we can hold the contract firmly and the implementation lightly.

And if we can do that for ourselves, it becomes easier to do it for others. We can begin to see that their religion, their family structure, their job, or their culture may simply be a different implementation of principles we also value.

That shift alone might change how we see ourselves. It might also change how we see each other.

There are things happening in AI right now that are just plain silly. People are riding the hype train, either with no idea where it is going, or pretending it is headed somewhere they know it is not.

Moltbook is one of those things. Openclaw is not.

And also, nothing happening right now merits granting personhood to AIs. Not even close.

Let’s unpack that.

Moltbook

Moltbook is the social network for AI agents currently taking the nerd internet by storm. It was built for Openclaw, the AI agent framework we will discuss shortly.

If you run Openclaw locally, you can instruct your agent to register on Moltbook, post content, and respond to other agents’ posts. Humans are ostensibly not allowed to post, only to read.

This setup has produced sensational headlines and breathless social media commentary suggesting that AI agents are collaborating, inventing new religions and languages, even plotting to hide things from their owners.

It sounds dramatic.

It is not.

Openclaw: the real breakthrough

To understand why the hysteria around Moltbook is silly, we need to understand what Openclaw is and what it is not.

Openclaw is an AI agent framework that runs locally on your machine (do not install it unless you are comfortable managing permissions - more on that in another post). It connects to your preferred GenAI service provider such as Anthropic, OpenAI, or Google to provide smart assistant functionality.

Its breakthrough, and also the reason you should be careful, is that it requests access to your private data. Files, emails, calendar, browser, social networks, and more. In exchange, it promises to connect all of that information and act on your behalf. It can flag important emails, draft responses, manage calendar events, monitor competitors, or summarize news.

It is one of the most successful recent open-source AI projects because it largely deliver on its promise.

But that power comes with real security risks. Even its creator, Peter Steinberger, has advised against installing it on your primary machine.

With access to your data and integrations across tools, Openclaw can interpret your instruction, determine which tool is needed, and execute the task.

That is powerful.

But power is not agency.

Expanded input is not agency

With a regular chat assistant, like ChatGPT or even Claude Code, your input is what you type into the interface, at the moment that you type it. Openclaw stretches that definition of “user input”.

First, it includes scheduling. Your input does not have to be something you send right now. It can be something you schedule for tomorrow, next week, or every hour indefinitely. That supports brand new use cases, like scheduling a daily news brief, or setting up a request to monitor a competitor’s web site for specific activity virtually continuously, or check a sales pipeline and take specific actions based on the signals it sends.

Scheduling turns prompts into automation.

Second, Openclaw implements a memory system, a structured set of files it can read from and write to. You can define its task preferences, and operating rules once instead of repeating instructions every time. You can give it a persistent personality. You can give it instructions on when to update its own “memory” and how to leverage it to perform the tasks you give it.

These capabilities make it extremely useful.

But when a scheduled job wakes up and executes while you are not at your keyboard, that is not agency.

It is deferred obedience.

Why the Moltbook hysteria is misplaced

Openclaw agents are not talking to each other.

They are executing configurations that cause them to send messages.

Every Moltbook agent was manually configured by a human to register, post about specific topics, and respond in specific ways.

None of it is spontaneous. None of it originates from the agents themselves.

Moltbook is active because humans are configuring agents to interact and enjoying the spectacle.

Interesting experiment? Maybe.

A waste of tokens and energy? Possibly.

Evidence of emergent AI consciousness? Absolutely not.

There is no collective awakening happening. There is no hidden plotting. There is no emergent self-organizing intelligence.

There are humans pointing automation systems at each other.

Agency and Personhood

Some might object that not all humans actively exercise agency. Infants, individuals in comas, and people with profound disabilities may not initiate goals in the way we typically imagine.

But the difference is not current performance. It is intrinsic moral status and intrinsic capacity.

Humans are members of a moral community independent of function. They are subjects of experience. They possess actual, latent, or species-typical capacity for agency. Even when dormant, impaired, or undeveloped, that capacity is part of what they are.

Large language models are not subjects. They are engineered systems. They have no intrinsic interests, no continuity of self, no vulnerability, and no capacity to originate goals. They do not possess latent agency waiting to be activated. Their outputs are entirely derivative of external input and external configuration, even when that input is deferred or the configuration persists over time.

A temporarily inactive human remains a being. An idle model is simply inactive computation. That is the difference, and we will need to remember it as these models get better and better at simulating human intelligence.

It’s not new for a piece of technology to emerge and threaten an entire sector of the economy. The Industrial Revolution did it. Robotics did it. History is full of similar moments.

What feels different about this GenAI-driven threat, though, is that this time we are the ones potentially being displaced. Knowledge workers. People who have contributed greatly to the demise of many other industries. Now our own livelihoods may be at risk because much of the work we’ve been doing could soon be handled almost entirely by machines.

Or could it?

Jensen Huang, the CEO of Nvidia (which, yes, is one of the companies benefiting from the current AI hype cycle), put it well in Davos earlier this year: we need to distinguish between a job’s purpose and its tasks. He used radiologists and nurses as examples, but the idea applies just as strongly to my people: engineers (software, quality, platform, data, you name it, but let’s use software engineers as a short-hand).

Even though historically our work has not been just about programming computers, i.e. writing instructions in a language a machine can execute, that specific aspect of software engineering has often become the center of attention because the process was so difficult.

It’s no surprise, then, that many engineers are nervous about GenAI’s ability to write most, if not all, of our code.

But writing code was never the purpose. Or at least, it shouldn’t have been.

I’ve worked with plenty of engineers who believed their primary value was the act of writing code itself. They were mistaken then, and they’re mistaken now. Our real purpose is to analyze problems, generate possible solutions, evaluate those options, implement one or more of them (this is where programming comes in), and then determine which one (if any) solves the problem best.

Implementation should never have been the centerpiece. It should have always been a detail, something mechanical that a machine could handle.

How many late nights have we all spent discovering that a bug was caused by a missing semicolon (damn you, JavaScript;), or that version 1.4.5 of Library A doesn’t work with version 2.8.7 of Library B? Is any of that fundamentally important to the people using the product we were building?

Of course not.

The painstaking act of assembling exactly the right characters, in exactly the right order, with exactly the right spacing was never the point. It was a necessity, a means to an end, a way to communicate intent to a machine.

We should celebrate that GenAI is finally freeing us from that minutiae. There are real pitfalls, and there’s still plenty of progress to be made, but the promise of GenAI for software engineering is undeniable. GenAI understands patterns. It understands syntax. And it’s increasingly good at understanding intent.

The GenAI era will be a golden age for true problem solvers: people who want to engage deeply with customers, who are curious about their domains, and who can articulate intent clearly enough to leave the implementation details to the machines.

TwigLeap brings together two simple concepts that run deceptively deep.

TWIG is shorthand for “the whole is greater than the sum of its parts”. It’s a familiar idea that most of us grasp intuitively: come together, and you will be better for it.

What I find compelling is how urgent this idea has become, especially when it comes to technology innovation and how we must change the way we solve problems. For example, each role in the Software Development Lifecycle could evolve its own tasks in isolation using Generative AI. But the real expansion of what’s possible happens when we step back and consider the system as a whole.

Imagine requirements specialists creating clickable prototypes, while software engineers write specifications. When teams blur traditional boundaries and think holistically, entirely new ways of collaborating and delivering value emerge.

This pattern shows up everywhere. Concepts combine to illuminate new meanings. Individual steps form end-to-end flows. Research methods converge to produce insight. Defenders and attackers come together to form a football (soccer) team.

This blog will explore how connecting the dots across disciplines, roles, and ideas can help us create more value and solve more interesting problems.

Leap represents the power of observation, the ability to think differently, challenge assumptions and innovate freely. One of my favorite leap examples is the story of M-Pesa in Kenya.

For decades, Kenya lacked the land-based infrastructure required for traditional banking and telecommunication. As a result, large parts of the population, especially in rural areas, were excluded from services that many countries take for granted. This placed Kenya’s economy at a severe disadvantage.

Yet that gap in infrastructure created a counter-intuitive freedom: when affordable mobile phones became widespread, Kenya had no legacy infrastructure to migrate and didn’t need to modernize an existing banking system - they had the opportunity to build something entirely new. And now M-Pesa is used by over 90% of adults in Kenya and facilitates almost 60% of the country’s GDP.

The birth and extraordinary success of Safaricom’s M-Pesa deserves its own post. And the core lesson that being “behind” can be an advantage when it frees you from legacy constraints is one we’ll return to often.

Together, the conviction that we need to free ourselves from legacy thinking and that we need to connect the dots in order to unearth new ways of solving problems is what animates this blog.

Subscribe now