prioritize the meta

One of my strongest convictions is that the most important things you can do in life don't look like progress. Preparation, reflection, recalibration. They operate not on your outcomes, but on the rate at which your outcomes improve, or even on the rate of that rate. I call these meta effects (I love the word "meta" from my debate days).

I want to formalize this intuition with two models. The first is geometric: your life as a vector in high-dimensional space, and the cost of misalignment. The second is kinematic: a toy problem that makes the dominance of higher-order derivatives viscerally obvious. Together, they make the case that in any long-term game, you should almost always be investing in the meta.

the alignment problem

Think of your life trajectory as a vector in a very high-dimensional space. There exists some vector also in that space that represents your most aligned self: the version of you that is doing the work you were "meant to do" — that you would be best at, that you would enjoy the most, that is most uniquely meant for you. You don't know this vector exactly and nobody does. But it exists as a target, and your job is to iteratively converge toward it.

Now consider a failure mode. You're 22, and deep, deep, deep down you know you want to be a scientist. You love asking hard questions. You love the feeling of discovering something new. You love the people that you've met in academia. But you go to a target school, join the consulting club, recruit for investment banking because the system makes it easy and the optionality narrative is compelling. You tell yourself you'll do two years, go to the GSB, HBS, or Wharton and reset. Instead, you go to the buy side. Then you tell yourself you'll reset after business school. You don't; you start a fund instead. You wake up at 55 running a PE shop, and something feels deeply wrong.

What happened here is not a single bad decision. It's the accumulation of momentum in a misaligned direction. And that accumulation of momentum in combination with an environment that makes each of these decisions very easy to make. This is exactly why this is the cliche example (although I have definitely met people within the financial services space who couldn't be more aligned with their job as well).

Each year, you add another velocity vector pointing toward wherever you're currently heading. These vectors compound. After year one, you're slightly off course. After ten years, you're not just further from your alignment vector. You're actively moving away from it at speed, with a career identity, financial obligations, and social expectations that all reinforce the trajectory. The total displacement from your aligned self isn't linear in time. It accelerates, because each year of misaligned momentum makes the next correction harder.

This is why course correction is cheapest when you have the least momentum. At a standstill, the cost of changing direction is near zero. And this is never more true than when you're at your youngest as there's no inertia to overcome. But at full speed in the wrong direction, realignment requires you to first decelerate, resist the world that has now organized itself around your current trajectory, and then re-accelerate in what is likely an opposite direction. The switching cost grows superlinearly over time.

So invest regularly in direction-setting. Not once, not at "transition points," but as a continuous practice. Spend time reflecting on whether what you're doing maps to where you actually want to go.

This is a meta effect: it doesn't move you forward, but it ensures that "forward" is the right direction.

the kinematics of leverage

The second model makes the quantitative case more explicit.

Imagine you have a race. The finish line is as far away as you can get in T months. You start at rest. Each month, you receive one token. You can spend that token in one of three ways:

• +1 to displacement: you teleport 1 meter forward.
• +1 to velocity: you permanently increase your speed by 1 meter per month.
• +1 to acceleration: you permanently increase your acceleration by 1 meter per month².

The question isn't which pure strategy wins. It's: at each month, which token should you spend?

This is a marginal analysis. Define R as the months remaining when you spend a token. The displacement you gain from each option is:

• A displacement token gives you exactly 1 meter. Always.
• A velocity token gives you R meters, because the added speed persists for the remaining time.
• An acceleration token gives you ½R² meters, because the added acceleration builds velocity across the remaining time, which builds displacement in turn.

The optimal play at any month is whichever option has the highest marginal return. So we find the crossover points. Acceleration beats velocity when ½R² > R, which simplifies to R > 2. Velocity beats displacement when R > 1.

Now set T = 960 months. Eighty years. A full human life.

| Months remaining | Optimal investment | | --- | --- | | R > 2 (months 1 through 958) | Acceleration | | 1 < R ≤ 2 (month 959) | Velocity | | R ≤ 1 (month 960) | Displacement |

958 months on acceleration. 1 on velocity. 1 on displacement. That's 99.8% of your life investing in the highest-order derivative.

Sit with that for a second. The math says the optimal month to start investing in direct skills and capabilities is month 959. The optimal month to start grinding out concrete outcomes is month 960, the very last month. Everything before that, every month from birth until roughly your 79th birthday, should be spent on acceleration.

Obviously this is a toy model. Life doesn't hand you one clean token per month, and the three categories aren't perfectly separable. But toy models are useful precisely because they reveal the shape of the answer even when the exact numbers don't transfer. And the shape here is unambiguous: the optimal allocation is almost entirely meta. Not 60/40. Not 80/20. 99.8/0.2. The only reason you ever stop investing in acceleration is that you're literally running out of time for it to compound.

If you're 25, you have roughly 660 months left. The crossover point where velocity beats acceleration is at R = 2. You are 658 months away from that threshold. Every instinct telling you to skip the meta work and just ship something, just close the deal, just get the credential, is asking you to allocate toward displacement when you have over half a century of compounding ahead of you. The math says that's almost always wrong.

what meta effects actually look like

So what does "investing in acceleration" translate to in practice?

The most obvious one is direction-setting — the alignment audit from Model 1. Regularly asking whether the direction you're heading is actually where you want to go.

Then there's your learning rate. Your ability to learn new skills is dramatically more cross-applicable than any individual skill. Learning how to learn, understanding your own cognitive patterns, reading across domains, building taste for what matters — this is pure acceleration.

Neuroplasticity sits even higher. Your ability to expand the dimensionality of your mental model is itself a higher-order derivative. Age, rigidity of worldview, and confirmation bias all erode it. Actively seeking orthogonal experiences, challenging your own assumptions, staying uncomfortable — these preserve the capacity to even benefit from acceleration.

And then health. Easy to dismiss as generic advice, but every single thing you do is downstream of your physical and cognitive capacity. Sleep, exercise, nutrition — these don't produce outcomes directly, but they multiply the output of everything that does.

the meta feels unproductive

There's an important psychological barrier here. Meta work almost never feels like progress. Reflecting on your direction doesn't ship a product. Improving your learning rate doesn't close a deal. Working on your judgment doesn't produce a deliverable.

The activities that operate on the highest derivatives are, by definition, the furthest removed from visible output. They feel like you're standing still while everyone else is sprinting. But the math is unambiguous: the person investing in acceleration will, given enough time, blow past everyone who skipped it in favor of raw displacement.

The kinematics model reveals something else worth noting. The dominance of acceleration depends on time horizon. In a 5-second race, the difference between strategies is modest. In a 100-second race, it's enormous. In a 1,000-second race, it's absurd. Life is not a 5-second race. It's a decades-long game, and decades are where the meta wins by orders of magnitude.

the synthesis

The two models say different things but point the same direction. First, make sure you're aimed correctly. No amount of acceleration helps if you're accelerating away from where you want to be. Then, once you have reasonable conviction in your direction, invest as high up the derivative chain as you can.

Life is a very long game. Prioritize the meta.