I've been doing capoeira for almost two years now, and the other day I sat down and tried to calculate how many hours of instruction I've actually received, and the number that came back was something like 120 hours. Two years of my life, and 120 hours to show for it. I found this troubling in a way I couldn't quite articulate at first, until I started doing more arithmetic and realised that a single quarter-year (90 days, minus sleep) contains roughly 1,440 waking hours. Which means I could have compressed two years of capoeira into a twelfth of a single quarter if I'd approached it differently. And this realisation has been gnawing at me all week, because I think it reveals something important about how we let ourselves exist in time, how we allow hours and days and years to slip past us as though they were abundant, as though attention were cheap, as though we had all the time in the world.
We don't, of course. But we act as if we do.
I want to be precise about what 1,440 hours actually means, because I think most people (myself included, until recently) have never really sat with the weight of it. The Foreign Service Institute estimates that a native English speaker needs 600-750 hours to achieve professional working proficiency in Spanish or French or Italian. Even Mandarin, which they classify as exceptionally difficult, requires about 2,200 hours for full fluency, though you can hold conversations at roughly half that. A single quarter-year, if you did nothing else, gets you halfway to reading Chinese newspapers and arguing with taxi drivers in Beijing.
Here's the same number dressed differently:
- 47.5 million words read at 550 words per minute (that's roughly 30 complete Worm-length web serials, back to back, which is more fiction than most people consume in a decade)
- 1,440,000 words written at a thousand words per hour (which is about fourteen full-length novels, or enough blog posts to establish yourself as a serious voice in almost any field)
- 288 five-hour Claude Code sessions (which, at current capability levels, is enough to prototype and ship dozens of software products that would have taken months or years to build just eighteen months ago)
I keep returning to these numbers because I've never poured 1,440 continuous hours into any single thing in my life, and I suspect most people haven't either. We let ourselves exist as if time is not a scarce resource that needs to be managed with the same rigour we'd apply to money or reputation or health. We operate at what I've started calling "real time," which is the default human pace: meetings that happen at the speed of talking, travel that eats hours without producing anything, conversations where you don't have your notes in front of you, situations where you wait. And I've become increasingly convinced that there's another mode available to us, a mode I've been calling quarter-time which means living as if each quarter-year were the unit of account instead of each year, as if 90 days were the natural rhythm of significant transformation rather than some arbitrary subdivision of the calendar.
I.
The thing about quarter-time is that it has nothing to do with speed in the conventional sense. It's about being greedy with updates.
Think about how learning actually works. Most improvement is punctuated. You operate under some coherent model of the world, and you do your reps, and you kick, and you perform a variation of something, and fresh data comes at you, and your model strains against the evidence. Eventually you hit a rupture point where the existing model is too inconsistent with reality, and it updates all at once. That rupture is where real learning happens. Everything else is just approaching the rupture, circling it, building pressure toward it. Ramanujan didn't become Ramanujan by doing more practice problems than everyone else; he became Ramanujan by having whatever bizarre neural architecture let him leap directly to the ruptures, skipping the tedious approach entirely. Most of us don't have that architecture. But we can at least ask better questions about our own learning: how do I get to the update faster? How do I clear the path to the rupture instead of wandering around in the fog for an hour before stumbling into it by accident?
One concrete way this plays out: managing the context in which learning happens. If you're learning about triangles for the first time and you erroneously believe that curved figures can also be triangles, you won't pay attention to what's actually essential, and you'll miss the side-side-side congruence theorem because you're imagining curved sides that could have the same lengths but different shapes, and the misconception blocks the rupture for hours or days until someone corrects you. So maybe the quarter-time version of learning triangles is: before you even start, make sure you have the right prerequisite facts loaded, clear the path to the rupture, don't let yourself be led astray for an hour before discovering you were confused about something basic.
Reading works the same way. Some naive person might think quarter-time reading means reading at 2,000 words per minute instead of 500, which is wrong, because humans have well-defined limits on processing speed. But there are other speed-ups available. The standard approach is to read, take notes, highlight, read again, process the same material multiple times, re-read a third time. All this scaffolding. But if you're actually thinking about the thing (grappling with it, treating the ideas as live, noticing where your model gets contradicted or updated), you can skip the scaffolding entirely and engage with the text right here, right now, word by word, tracking where your beliefs shift. One pass instead of three. That's quarter-time reading. I'm not claiming this is optimised for long-term recall. I don't have all the answers. But speed-ups like this are almost always available, and you're probably doing things in unnecessarily long and tedious ways without realising it, and allowing for that possibility is the first step.
II.
On the 24th of November, the cost of software production went to zero.
I mean this almost literally. Claude can write code now, and I don't mean toy code or "hello world" scripts, I mean real code, the kind that would have taken a senior developer days or weeks, the kind you would have decided wasn't worth building because your time was too expensive. A $100/month subscription gives you something like three million tokens, which translates to an almost absurd quantity of throwaway code and tests and research and prototypes you can spin up and tear down and spin up again. And most people don't know this yet. They haven't felt it. They're still operating as if building software requires expensive human attention at every step, like we're all still mass-producing Ford Model Ts by hand because nobody told us the assembly line got invented.
There's a countdown that started in November. People are slowly learning about this arbitrage opportunity, slowly figuring out how to spend tokens, slowly discovering that their expensive developer time can now be leveraged 10x or 100x. But the window is closing. Maybe 5-7 months before the arbitrage gets competed away, maybe less. This is the source of my urgency. Every hour I'm not spending tokens feels like an hour wasted. Every meeting I attend and every conversation that happens at the speed of talking and every time I travel somewhere instead of staying at my terminal (all of that is real-time activity, and real-time activity trades off against quarter-time activity, and the window is closing).
The distinction crystallises when you think about what each mode enables. In real time, you interface with people: conversations happen at the speed of speech, you travel, you don't have your context, you're not updating your documents, you're not spending tokens, time moves at 1x. In quarter-time, you interface with Claude: motion generates information, uncertainty gets resolved because you can just try things, you can run six research agents in parallel, you can prototype an idea in an afternoon that would have taken a month, time moves at 4x or more.
The uncomfortable truth is that people are slow. A lot of conversations could have been emails. Yes, some conversations genuinely demand face-to-face communication, the full modality, the real-time reading of expressions and tone. But most don't. "Here's a problem I have, blah blah blah" could have been async. I keep committing to meetings. I keep letting myself get pulled into real-time activities. There's social gravity to it, people expect you to show up and be present and operate at their pace, and every hour in real time is an hour outside of quarter-time, and the window is closing.
III.
Here is something absurd: I have earned real money doing forgettable work for other people. Didley squat, really. Corporate stuff. The kind of work that evaporates the moment you stop doing it.
And yet $100 from a single thing I sell on the internet remains an impossible dream.
This is absurd in a way that I find difficult to sit with. Tens of thousands of people earn $100 online every day. They sell stuff on Shopify and post TikToks and have tiny audiences that buy tiny products. It's not magic. It's not even hard, apparently. My granpa could probably figure it out if he weren't dead. So what's my blocker?
I have a friend who's not a particularly bright guy, but he's more extroverted than me, maybe higher energy. He comes from a formerly wealthy family and now has to take care of people. He started posting TikToks in September and now he gets 12,000 pesos per video. That's not a lot in the grand scheme of things, but it's something. He has an audience. He's actually selling his viewership. He figured something out that I haven't figured out. I keep skirting around this thing my entire life, like a dog circling a spot on the carpet seventeen times before lying down, except I never actually lie down, I just keep circling. There has to be some emotional block, some knot of avoidance that I haven't untangled yet, because this clearly isn't a knowledge problem.
What is a product, really? One type is a productised service: you help one person, then another person pays you to help them the same way, you notice the pattern, you systematise it. I've helped a bunch of people in wildly different ways across various domains. I'm probably as generalist as they come. Where are the patterns? What recurs? Another type is when you notice a problem and build a solution and sell access. But now that Claude can build the solution for you, the bottleneck has shifted. It's something else now. Noticing the right problem, maybe. Or having the stomach to actually ship.
I've heard this passed around as uncriticised startup wisdom: "all startups are equally hard." The original intention was probably something like: in terms of emotional difficulty and the kinds of problems that crop up, startups take all your bandwidth regardless of which startup you pick, so you might as well choose carefully. But I think people misread this. They hear "equally hard" and conclude that it doesn't matter what you build. Just pick something. Grind. But that's wrong, isn't it? There are physics to startups. Dynamics relative to the market, the timing, the users already there, the users who have yet to learn about your thing. Some startups have tailwinds. Some don't. Asking "which startup should I do?" while ignoring the physics is like asking "which direction should I sail?" while ignoring the wind. You can sail into the wind, technically, but you'll be tacking back and forth while the guy who checked the weather report glides past you.
There's a tension between building something you actually want to build versus something optimised to earn money. If you optimise for money, you learn the physics (you see what makes startups live and die), but you might hate what you're building. If you optimise for what you want, it's possible (likely, even) that not enough people want your thing, or you can't reach them fast enough, and it never develops legs. Maybe the answer is to treat the money-optimised thing as training wheels. Learn the physics. Then build what you want, informed by the physics.
Or maybe that's cope. Maybe I'm just avoiding the thing.
IV.
One consequence of thinking in quarter-years is that every hour becomes weighty. There's always this question: in this particular hour, what am I actually doing? Does it trade off against building more stuff and testing more ideas and evolving my setup?
I notice myself feeling increasingly guilty. The hours I live are not the hours I would live if I really had a quarter-year perception. I'm not maxing out my Claude subscription. I'm not hitting my weekly limit. I haven't reorganised my sleep schedule to maximise time with the tools. I attend meetings. I travel. I let myself get pulled into real-time activities.
If I lived every hour as if it mattered, if I truly operated in quarter-time, what would change? Maybe 40-50 more excursions or projects done with Claude Code in a quarter, and I mean done, finished, shipped. Possibly 200 if I truly maximised finishing things, if I went for low-hanging fruit instead of doing what comes naturally. The thought is dizzying: three months, 200 projects! What would that even look like?
Part of the problem is that I don't even know if building is the thing I want to be doing. There's a summit on Friday. I could go. I'd meet people, have conversations, maybe generate opportunities. But it would trade off against time with Claude. It's real-time activity. The social gravity pulls me in, and I hate it. But also: if I want to run a research group someday, people operate in real time. I can't escape it entirely. Reading summaries isn't quarter-time. Some things genuinely require presence.
God damn it.
I think the right move is probably to attend the event, do the panel, talk to a handful of people, then leave. Don't get pulled in wholesale. Protect the quarter-time hours.
One frame I keep coming back to is this: we're entering an r-selection era. In evolutionary biology, r-selection is the strategy of producing many offspring with little investment in each. K-selection is the opposite: few offspring, high investment. Most personal-development advice assumes K-selection. Pick one thing. Pour yourself into it. Compound over years.
But if the cost of production asymptotes to zero, r-selection becomes dominant. You can try many things and spin up prototypes in parallel and test ideas in the market quickly and see what clicks and double down on winners. This feels wrong to people with taste. It feels like giving up on craft. But I think the discomfort is misplaced.
Yes, a lot of exhaust will be produced. People will build things they realise they don't really want. They'll ship products that never find users. There will be slop. Mountains of slop. The slop hasn't fully saturated human activity yet, but it will, whether we like it or not.
The challenge is to produce your own slop before everyone else catches up, to treat quantity as a search strategy rather than a quality failure. Quantity has a quality of its own: more data, more feedback, more updates. And right now, the filtering function can come later. The priority is to generate.
But (and this is crucial) slop without coherence is just noise. If you're trying projects at random, you're not learning, you're not accumulating anything. The r-selection strategy only works if there's a through-line. A personal cinematic universe of your own software. Each project connected to the others. Each failure informing the next attempt.
This is where sovereignty matters. You now have to manage your own data and your own context and your own expressions of taste. Because if you put everything on external platforms (your messages, your emails, your half-baked prototypes), it's just there, forever. People will make backups. They'll use it against you. You lose control.
What you want is sovereignty over the data you've produced already and the data you keep producing. And by "data" I mean something broader than smartphone activity. I mean your moment-to-moment expressions of taste. How does a strawberry feel in your mouth versus another fruit? How does that differ from another person's experience? These are things even the best language models won't replicate for years. They're yours.
The personal cinematic universe is where your taste gets encoded. It's coherence across slop. It's the reason 200 projects in a quarter could actually compound into something instead of dissipating into noise.
So what does quarter-time look like concretely? I keep a running list of transformations I'm considering, and the list keeps growing.
Most of my email could be automated. I could fine-tune Claude to write in my voice, or voice-note my responses and have them formatted into something minimally recognisable. I know enough to avoid producing instantly-recognisable slop (the kind that makes you wince and think "a robot definitely wrote this"). This should save hours per week.
When I'm uncertain about something, I don't need to sit with the uncertainty anymore. I can ask Claude to interview me about the decision. I can run six research agents in parallel on technical topics I need to understand. Motion generates information. Uncertainty gets resolved. The cost of trying is now nearly zero, which means the old calculus of "is this worth investigating?" has completely changed. The answer is almost always yes now.
What's the quarter-time version of learning Mandarin? Can you compress 2,200 hours to 550 by using these tools strategically? I don't know the answer yet, but the question is worth asking. A lot of the constraint was on the supply side (tutors, materials, feedback loops), and those constraints are relaxing.
If I initiated more conversations (cold emails, landing pages, outreach to people with software-shaped problems), what would happen? Maybe an extension that watches the links I click, identifies problems people are having, suggests who to contact. "Can I solve this for you? Completely free." Capture some percentage of the value.
The pattern is: take any activity currently in real time and ask what the quarter-time version would be. Sometimes the answer is that there isn't one, because some things genuinely require presence and slowness. But often there's a transformation available that I just haven't found yet.
I keep wanting to end with something clean. A thesis statement. A call to action. But that would be dishonest.
The truth is I'm wrestling with this. I feel the urgency. I see the window. I understand, intellectually, that 1,440 hours is enough to do almost anything. And yet I still attend meetings. I still let myself get pulled into real-time activities. I still haven't earned my first $100. There's some irony in writing a 4,000-word essay about the importance of not wasting time, but alas, nobody's perfect.
I truly believe that three months is a lot of time. It's not infinite, but most projects that have ever been imagined can be prototyped or even fully implemented within that frame. And the question I keep asking, the question that won't leave me alone: what would radically change my life in three months?
I don't have the answer yet. But I'm going to keep asking until I do.