All of humanity's problems stem from man's inability to sit quietly in a room alone.
Governance of the Self
Time, attention, energy, identity — the four resources that compose into who you become. The disciplines that keep an operator human as the instruments multiply.
Chapter 11 — Governance of the Self
Uddhared atmanatmanam, natmanam avasadayet —
let a man lift himself by his own self;
let him not lower himself;
for this self alone is the friend of oneself,
and this self alone is the enemy of oneself.— Bhagavad Gita, 6.5 (c. 200 BCE)1
All of humanity's problems stem from man's inability
to sit quietly in a room alone.— Blaise Pascal, Pensées, fragment 139 (1670)2
A small refusal
Two months ago I was about to publish a piece — a short blog post on a technical topic adjacent to my work — that the model had drafted and I had only lightly edited. It was Tuesday evening. I was tired. The post was scheduled to go out in an hour. I was about to click publish and move to the next thing on the list.
I stopped, for a reason that took me a minute to name.
The post was technically correct, the prose was clean, the framing was reasonable. There was nothing wrong with it. But there was a sentence in the second paragraph that I would not have written. Not in the sense that it was a mistake — it was fine — but in the sense that it was the model's voice and not mine. It had a particular kind of tidy three-clause structure that the model defaults to when it has not been pushed. Reading it again at 9 PM on a Tuesday, I could feel the model in it.
I unscheduled the post.
I rewrote that one paragraph from scratch. I re-read the whole. I rescheduled it for the morning. The whole exercise cost forty minutes I did not really have.
The next morning I read the published version and, this time, I recognized myself in it.
The forty minutes were the most important forty minutes of the week, and I want to be honest about why.
Without that pause, the post would have gone out under my name, with a sentence in it that was not mine, on a public platform, where readers — some of whom I will never meet — would have absorbed something written in a generic voice and attributed it to a specific person. The damage would have been small. Each post like that, by itself, is small. The aggregate damage of a creator who lets a thousand of these slip through over five years is the gradual erasure of their voice — into the average, the smooth, the placeless.
This is not a hypothetical risk. It is happening, at scale, right now, to creators who used to have voices and have not built the discipline to keep them in the era of cheap drafts. The voice is the most precious thing a creator carries. It is also, structurally, the single thing the models cannot manufacture and the easiest thing to lose by carelessness.
This chapter is about the discipline that prevents that loss.
It is also, more broadly, about the architecture of being someone — having attention to spend, time to honour, ethics to enact, identity to protect — in a moment when all four of these are under unprecedented pressure from instruments of unprecedented amplification.
The Stoics had a name for this. They called it governance of the self. We are about to need it more, not less.
The four resources
A useful frame, when thinking about personal governance, is that you are running four distinct, finite, ethically charged resources, and the way you govern each of them shapes the others.
Time is the most obvious. You have an hour, a day, a year, a life. Time spent on one thing is not spent on another. Most people have a vague sense of time as a resource and spend it accordingly poorly.
Attention is the most under-recognized. Attention is the rate at which you can consciously process the world. It is finite, it is costly to direct, it can be fragmented or focused, and the cost of fragmentation is large. Attention is what advertising harvests, what social media exploits, and what the contemplative traditions have spent millennia training. Time without attention is dead clock-time; the same hour can be expansive or wasted depending on what you brought to it.
Energy is the resource that determines what kinds of work you can do at any given hour. Some hours of the day, you can do creative work; in other hours, you can do administrative work; in still others, you can do nothing useful at all. Energy follows physiology — sleep, food, movement, stress, the nervous system. The serious operator runs their hard work in their high-energy hours and accepts that not all hours are equivalent.
Identity is the resource the others compose into. Identity is the accumulated, particular who you are that emerges from how you spend the previous three resources. Identity is shaped by what you make, who you spend time with, what you read, what you refuse, what you say in public, what you say to yourself in private. It is the slowest of the four to change and the most consequential.
The personal CoE of Chapter 9 is the architecture for deploying AI. The personal governance of this chapter is the architecture for deploying yourself — the four-resource discipline that runs underneath everything.
The two architectures interact. Bad time governance corrupts attention; bad attention governance erodes energy; bad energy governance distorts identity. And vice versa. The four resources are coupled, and a discipline that strengthens one usually strengthens the others.
I will walk each of the four briefly. Then I will return to what AI does to all of them, which is the heart of this chapter.
The governance of time
The Stoic and contemplative literatures converge, almost word for word, on a small set of principles about time.
Seneca's De Brevitate Vitae — On the Shortness of Life — argues that the complaint about life being short is wrong; life is long enough; we have squandered most of it. The Buddha's discourses on the rarity of human birth and the urgency of practice land in the same place. Marcus Aurelius's repeated reminders to himself — do not act as if you would live ten thousand years, you may die at any moment — are the same instruction in different vocabulary.
The lesson is unanimous. Time is finite, much of it has already been spent on things that did not matter, and the response is not panic but care.
Practical time governance for the modern operator is, again, less complicated than the productivity literature implies. A small set of disciplines does almost all the work.
Decide your work. Pick a small number of things — three or four projects in any given season — that you are seriously trying to make happen. Do not try to make ten things happen. The cost of distributed effort is more than linear, and the cost of starting a new project before the previous ones are complete is paid in months of lost momentum.
Schedule your work. Hold blocks for the things that matter, in the energy windows where you can do them. The novel gets the morning. The administrative work gets the afternoon. The reading gets the evening. The exact pattern is yours to find. The discipline is to have one.
Refuse meetings that are not your work. The single largest theft of time in a modern professional's life is meetings that should have been emails or never have happened at all. The refusal is uncomfortable; it costs social capital; it is also one of the most reliable ways to recover the time on which a serious creative life depends.
Honour rest. Time spent not working is not wasted; it is the substrate from which work emerges (Chapter 7, again). A schedule with no rest is a schedule that produces increasingly bad work.
These are unfashionable disciplines because they are obvious, and because they are slow to compound. They also, after five or ten years, separate the people who get serious work done from the people who don't.
The governance of attention
Attention is the new currency, and it is being mined.
The economics of the present internet are built on the harvesting of human attention by algorithmic systems whose objective functions reward engagement. Engagement, in those systems, is mostly indistinguishable from outrage, anxiety, or addiction. The systems are not malicious. They are optimizing what they were told to optimize. The result is a world in which the average human now spends multiple hours a day in a state of fragmented attention being handed to them by machines.
This is one of the under-acknowledged crises of the present moment. The damage is not principally to time — although that is bad — but to attention itself. A mind trained for years to fragment its attention every ninety seconds in response to a notification loses the capacity to remain with anything for ninety minutes. The substrate of deep work, deep relationship, and deep contemplation is being slowly eroded in millions of people who do not know it is happening.
AI complicates this picture.
On one hand, AI is, increasingly, the most powerful tool the attention economy has ever had. Recommender systems, ad targeting, persuasive content generation — these are getting much better, much faster, and the risk of asymmetric warfare against an unprotected human attention is real.
On the other hand, AI also offers, for the first time, the prospect of agents that work for the user rather than the platforms — assistants that filter, summarize, draft, refuse on the user's behalf, and let the user spend their attention on what they actually care about. Whether the era unfolds in the first direction or the second depends, in significant part, on whether ordinary people develop the disciplines of attention governance that allow them to wield the new tools rather than be wielded by them.
The disciplines are, again, smaller than the productivity literature implies.
Reclaim the morning. The first ten minutes after waking are the highest-leverage attention window of the day. Phone in another room. Notebook by the bed. The first inputs of the day are the inputs of the day.
Practice silence daily. Ten minutes, eyes closed, slow breath. The attention you train in those ten minutes is the attention you have available for the other twenty-three hours and fifty minutes. The contemplative literature is not optional for the modern operator; it is the foundational training for the substrate that everything else runs on.
Run an information diet. Decide, deliberately, what comes in. Choose a small number of regular sources — slow ones, where the information is dense and the framing is honest — and reduce or eliminate the high-noise channels. The default setting of take whatever the platforms send you is the equivalent, for attention, of eating whatever is on the conveyor belt at the food court. The personal CoE includes a dietary policy for inputs.
Audit your apps quarterly. Which apps justify their place on your phone? Which are providing value? Which are mining you? Delete the second category. Re-evaluate the first.
These four practices, run for a year, change the substrate of a person's available attention more than any productivity technique on offer. They also become harder to do as the AI-driven platforms grow more sophisticated, which means the bar of necessary discipline is rising, not falling.
The good news is that the same era that produces the attention-mining systems also produces the tools to defend against them. The defense is not technological alone. It is the contemplative disciplines, restated for the present, applied with seriousness to the new conditions.
The governance of energy
Energy governance is the governance of the body.
Most of what is written under this heading in the productivity literature is right but obvious. Sleep eight hours. Exercise three to five times a week. Eat real food, not too much, mostly plants. Spend time outside. Spend time with people you love. Avoid chronic stress where you can; metabolize it where you can't.
I will not repeat the literature in detail. I will offer one observation that the productivity literature understates.
Energy is the resource through which every other resource flows. Time means nothing if you have no energy to use it. Attention degrades sharply with energy. Identity, over years, is shaped by the choices you make on low-energy days more than by the choices you make on high-energy days, because the low-energy days are more numerous and the choices accumulate.
The single most important energy intervention is sleep, and we covered it in Chapter 7.
The single most under-discussed is the management of stress. Modern professional life produces a low-grade chronic stress that the human nervous system was not designed for and does not handle well over decades. The cumulative effect is real and measurable, in cortisol levels, in cardiovascular risk, in cognitive degradation, in mood, in relationships, in the slow blunting of joy.
The disciplines that metabolize stress are mostly old. Walking. Movement of any kind. Time in nature. Time with people who love you. Slow breathing. Silence. The contemplative practices. Music. Prayer, if it belongs to your life. Service to others.
These are not luxuries. They are the maintenance of the substrate from which everything else runs. The operator who treats them as optional is running the engine without oil and will fail later, often badly, in ways that are catastrophic to projects, relationships, and health.
In the AI era specifically, there is a new energy challenge. The instruments are so capable that the temptation is to do more — start more projects, run more loops, ship more outputs, keep more late nights. The amplification of output capacity, without a matching amplification of recovery capacity, is a recipe for the early burnout that I have already begun to see in some of the most prolific operators I know.
The discipline is to let some of the new bandwidth become rest. The cheap leverage is precisely what allows you to do the same work in less time, and to give the saved time back to the body. The operators who will still be making in twenty years are the ones who use the leverage that way. The ones who use it to push harder will not be making in twenty years, or will be making at a steeply diminished quality.
The Renaissance is long. The studio works on the master's behalf for decades only if the master is still alive and present.
The governance of identity
Identity is the slowest resource and the most under-attended.
In any given week, your identity does not noticeably change. In any given decade, it changes profoundly. The cumulative pressure of the choices you make about how to spend the other three resources — time, attention, energy — produces, gradually, a particular person.
The question of identity governance is: who are you becoming, and is that the person you would choose?
Most people do not ask this question. They drift. The drift produces, over years, a person who is the median of their environment — their workplace, their social media, their consumed content, their casual habits. This is not a moral failing; it is the path of least resistance, and the path of least resistance is well-paved by an attention economy that profits from your default settings.
The discipline of identity governance is to ask the question deliberately, regularly, and honestly.
Once a year, on a quiet morning, write a single page on it. Who are you becoming? Is that the direction you would choose? Where, in the last year, did you drift? What did you absorb that you would not have chosen? What did you make that you are proud of? Whose opinions of you have started to matter more than they should?
The page is private. It is not for the internet. It is for you and your conscience.
In the AI era, identity governance has a particular shape that is worth naming.
The risk specific to this era is voice contamination. The constant exposure to model outputs — drafts you read, replies you accept, summaries you skim — slowly trains your own voice toward the model's. This is not metaphorical. Linguists studying the spread of LLM-style writing in 2024-2026 have already begun to identify a recognizable register that is migrating into human prose by osmosis. People who use the models heavily and uncritically begin, over months, to sound like the models.
The discipline against this is, again, simple. Read good prose. Write your own first drafts before consulting the model. Edit the model's drafts thoroughly enough that what remains is yours. Check your published work, periodically, for the markers of model-voice — the tidy three-clause sentences, the smooth abstract nouns, the indistinct general claims. If you find them, the discipline has slipped, and the work is to recover it.
Identity, in the era of cheap amplification, is a discipline. It is also, increasingly, the differentiator that will separate the consequential operators from the volume producers. The volume producers will sound like each other and like the models. The consequential operators will sound like themselves, because they have done the work to keep the substrate of who they are intact.
This is the most worth-protecting thing you have. Protect it.
A few personal policies
I have, over the past few years, settled into a small set of policies that govern my own AI use. I want to share them not as prescriptions for the reader, but as examples of the kind of thing personal governance produces. The reader's policies will be different, and that is the point.
I do not use AI to write to people I love. The condolence message, the apology, the message to my children when I am away — these I write by hand, on the principle that the inefficiency is the point. The friction of writing it myself is what makes the message what it is. Outsourcing the friction outsources the meaning.
I do not publish under my name what I would not have written. The Tuesday-evening rewrite from the opening of this chapter is not an exception; it is the rule. Anything that goes out under Frank Riemer has been read by me carefully enough that I would defend every paragraph as mine.
I declare collaboration when it matters. When AI has done significant lifting in a piece of work — drafts, code, analysis, illustration — I say so. Not in every sentence. In the appropriate place. The discipline of telling the truth about what is yours and what is the collaboration is a key ethical discipline of this era, and the failure to do it corrodes the public discourse fast.
I run a quarterly refusal-list review. Once every ninety days, I look at the things I have refused to do with AI and ask whether the refusals are still right. Sometimes they have softened — something I was once unwilling to do, I now think is fine. Sometimes they have tightened — something I was once willing to do, I now think is corrosive. The list is alive.
I take regular fasting weeks from the heaviest tools. Once or twice a year, for a week, I work without AI on the work that matters most. The point is not asceticism; it is recalibration. The week reminds me what I sound like, what I think like, what I produce when the instrument is unplugged. The fasting weeks make the working weeks better.
These policies cost me time and a small amount of efficiency. The benefit, over years, is that the work that ships under my name is recognisably mine, the relationships that matter to me have not been mediated through models, the substrate of my voice is intact, and I can defend in the company of my own conscience the way I have used these instruments.
Your policies will be different. The discipline of having them is not.
The unexamined use
Socrates said the unexamined life is not worth living. The reformulation for the present is small.
The unexamined AI use is not worth shipping.
Every operator, in this era, has a private negotiation to run between the leverage offered by the new tools and the substrate of the person they want to remain. The leverage is real. The substrate is real. Neither has to be sacrificed for the other. But the negotiation has to be conducted, deliberately and honestly, by the operator themselves.
The personal CoE of Chapter 9 is the architecture of deployment.
The personal governance of this chapter is the architecture of integrity.
Both are required. Either alone is incomplete.
The good news, again, is that the disciplines are old, the costs are small, and the compounding is real. The operators who run them well will not, in twenty years, look back on the AI era and feel that something was taken from them. They will look back and see that something was given, that they had the discipline to receive it well, and that the giving made them more themselves.
This is the bargain at its best.
Hand-off
The architecture is in place. The studio is built. The instrument is wielded. The integrity is intact.
What remains is the only thing that has ever mattered.
The transmission.
What we have learned, made, suffered, and become — passed to the people who will carry it forward when we no longer can. The teacher to the student. The parent to the child. The creator to the audience. The elder to the apprentice. The species, slowly, to its successors.
In Chapter 12, the closing chapter, we walk into the room of transmission. The reason for the work. The thread that links every Golden Age to the Golden Age before it. The benediction for the next person who picks up the instruments after we set them down.
The book closes there.
The Golden Age, however, is just beginning.
We are walking in.
Footnotes
- Bhagavad Gita 6.5. The verse continues: for whom the self has been conquered by the Self, the Self is the friend; but to him who has not so conquered, the Self is hostile, like a foe. Translation adapted from Eknath Easwaran. ↩
- Blaise Pascal, Pensées, fragment 139 in the Brunschvicg numbering (fragment 136 in Lafuma). The Pensées were unfinished at Pascal's death and published posthumously in 1670. ↩