Some games have capped outcomes. You can play perfectly and still only win a predictable amount. Other games have uncapped outcomes, where a single result can dwarf everything else combined. Most people spend their entire lives playing the first kind without realizing the second kind exists.

This distinction comes down to two distributions that govern almost everything: normal distributions and power laws.

The Two Distributions

In a normal distribution, outcomes cluster around an average. Extreme results are rare and bounded. Height is normally distributed: most people are close to average, and no one is five times taller than anyone else. If you measure more, your average converges and stabilizes.

In a power law, there is no meaningful average. A small number of outcomes account for the vast majority of the total. Income follows a power law: some people earn 10x, 100x, or 1,000x more than others. If you measure more, the average keeps shifting because one new outlier can dominate everything you’ve measured so far.

  Normal Distribution Power Law
Outcomes Cluster around an average Span orders of magnitude
Extremes Rare, bounded Rare, but unbounded
The average Useful predictor Misleading (dominated by outliers)
More data Average stabilizes Average keeps shifting
Winning strategy Optimize consistency Maximize persistence

Where You See Each

Normal distribution games are everywhere. Working an hourly job, ranking up in a competitive video game, filling restaurant tables night after night. The outcomes are proportional to effort. You grind, you get a predictable return. There’s a ceiling.

Power law games look completely different. Venture capital: Y Combinator calculated that 75% of their returns came from just 2 out of 280 startups they funded. Publishing: most books flop, but one bet on a story about a boy wizard turned Bloomsbury into a global brand. YouTube: less than 4% of videos reach 10,000 views, but those videos account for over 93% of all views.

The pattern repeats: most attempts produce little, but the rare wins are so large they make everything else irrelevant.

Why Power Laws Exist

Power laws emerge when small causes can cascade into massive effects. In physics, this happens at critical points, where systems become maximally unstable and a tiny perturbation can ripple through the entire system.

Forest fires work this way. Most lightning strikes burn a few trees. But when the forest is dense enough, one identical lightning strike can trigger a fire that burns across an entire state. The cause is the same. The outcome is not.

The 1988 Yellowstone fire burned 1.4 million acres, 50 times more than all fires over the previous 15 years combined. There was nothing special about the spark. The forest was simply in a critical state.

Earthquakes follow the same pattern. The physical process behind a tiny tremor you can’t feel and a catastrophic quake that levels a city is identical. The difference is whether the stress cascades along the fault line or dissipates locally. You can’t predict which one it will be. The system is inherently unpredictable at the critical point.

The same dynamics show up in networks. Barabasi found that the internet follows a power law: a few sites have thousands of times more connections than most. New nodes are more likely to connect to well-connected nodes, creating a snowball effect. This preferential attachment is why early advantages compound: the more connected you are, the more connections you attract.

The Decision

If you’re playing a normal distribution game, consistency wins. Show up every day, optimize the small things, grind out incremental improvements. The returns are proportional and predictable. There’s nothing wrong with this, but the ceiling is real.

If you’re playing a power law game, persistence wins. Most of your bets will produce nothing. That’s not failure, that’s the expected distribution. The strategy is to keep making intelligent bets, because you can’t know in advance which one will be the outlier. You only need one.

The costly mistake is spending all your time on normal distribution games, optimizing for a 10% raise or a slightly better ranking, while ignoring power law games where a single outcome could change everything. The opportunity cost is invisible because you never see the power law game you didn’t play.

The even costlier mistake is suppressing small fires. The US Forest Service spent a century trying to prevent all fires, which just made the forest denser and the eventual megafires more catastrophic. In your own life, avoiding all risk and variability doesn’t eliminate the power law. It just ensures that when the big event comes, you’re unprepared, or worse, you’re never in the game at all.

Playing Power Law Games

Knowing the theory is one thing. Actually shifting how you think and act is another. Here’s what changes when you start treating life as a power law game.

Input vs. Outcome

Normal thinking: output is proportional to input. Work 10% harder, get a 10% raise. Study four hours instead of two, get a better grade. The goal is efficiency: best return per hour.

Power law thinking: outcomes are non-linear. 99% of efforts yield nothing. 1% yield 1,000x. Working harder on the wrong thing is useless. Finding the right thing is the only thing that matters. The goal is optionality: maximize exposure to positive outliers.

The fear shifts too. Normal players fear wasting time on something that doesn’t pay immediately. Power law players fear missing the magnitude, being steady but capped.

Time vs. Equity

Normal path: sell time for money. Career ladder. Junior to senior to manager. The variance is low. You won’t make $10M next Tuesday, but you won’t make $0 either.

Power law path: seek leverage. Code, media, capital. These work for you while you sleep. A salary only works when you’re awake.

The barbell strategy: keep a boring, low-effort income source to survive (capped downside), while aggressively pursuing high-variance projects (uncapped upside). Prefer equity, royalties, or products over salary.

Instead of consulting for $200/hr, build a SaaS tool that might fail completely or scale to $20k/month with zero marginal cost.

Networks

Normal strategy: maintain a stable circle of similar peers. Coworkers, local friends. Comfortable, predictable, low new information.

Power law strategy: send cold signals. Emails, DMs, published work. One introduction to a super-connector is worth more than 1,000 coffees with peers. Publish your work publicly, because the internet has fat tails. Bill Gates might see your blog post, but he will never see your internal memo.

Casper, the researcher from the Veritasium video, read a line in a book: “One idea could transform your entire life.” He wrote underneath it: “Send an email to Veritasium.” Four weeks of silence. Then a reply that changed his career. Same mechanism as the lightning strike. The cause was small. The outcome was not.

Skills

Normal path: deep specialization in a predefined niche. “I am a tax accountant for mid-sized retail firms.” The risk is obsolescence. If the niche shrinks, your value drops.

Power law path: talent stacking. Combine 2-3 skills that don’t usually go together. Being top 1% in one skill is brutally competitive (normal distribution competition). Being top 25% in three different things and combining them creates a monopoly of one (power law value).

Failure

This is the most critical shift. Normal players view failure as a net loss of resources and social standing. “I wasted six months on that.” Power law players view failure as the cost of discovery. They expect 9 out of 10 projects to fail.

The VC approach to life: don’t try to ensure every Saturday night is “pretty good.” Have five terrible weekends exploring weird hobbies to find the one passion that defines the next decade.

The Algorithm

Cap the downside. Ensure you can’t go to zero, financially or socially. Keep a safety net.

Maximize shots on goal. Take as many small, intelligent risks as possible. Write more posts, launch more tiny projects, send more cold emails.

Cut losers fast. If something shows linear or diminishing returns, kill it. Don’t fall for the sunk cost fallacy.

Let winners run. When something starts working exponentially, drop everything else and double down.

The phrase that stuck with me: be persistent, not consistent. In a power law world, the person who makes 100 smart bets and fails 99 times will outperform the person who made one safe bet and succeeded.

What You Learned

âś“ Normal distributions reward consistency; power laws reward persistence
âś“ Power laws emerge from critical systems where small causes cascade into massive effects
âś“ Seek leverage (code, media, capital) over trading time for money
âś“ Stack skills instead of hyper-specializing; top 25% in three things beats top 1% in one
âś“ Failure is the cost of discovery, not a loss. Expect most bets to fail.
âś“ The algorithm: cap downside, maximize shots, cut losers fast, let winners run

Resources: