Learning Losses

Learning losses are losses taken deliberately in practice, where they cost nothing, to expose and fix weaknesses before an opponent can exploit them in competition. The idea rests on two related concepts: the losing quota, the baseline amount of losing everyone must go through to improve, and the competency trap, the fragile comfort of a game built on one winning method.

In education, "learning loss" means something unrelated: academic progress students lose, for example over long breaks from school. This page covers the competitive meaning.

Where the idea comes from

Edge: The Art of Ruthless Competition gives learning losses a chapter in its third part, Play for Blood, and opens it with a paradox:

"A paradox of playing for blood is that you must be prepared to lose all the time."

Edge, Play for Blood: Learning Losses

There's a catch: the losing has to be rational and calculated, done where it teaches the most and costs the least. Practice fits both conditions. Nothing is at stake, and you can choose exactly which weakness to put under pressure.

"Remember: being the best doesn't mean you never lose, it means you lose less than everyone else."

Edge, Play for Blood: Learning Losses

Why losses teach more than wins

Winning tells you your current approach worked once. A loss tells you where it breaks. Avoiding every loss in training means throwing away the most useful information you can get about your own game.

"A win is a confirmation of your current worldview, while a loss is a correction - which is more information-rich by default."

Edge, Play for Blood: Learning Losses

The losing quota

"Every game has what I like to call a losing quota, which is some baseline amount of losing everyone has to endure in order to get better."

Edge, Play for Blood: Learning Losses

Nobody's quota is the same size: it depends on the game and on how quickly a player learns. Skipping it isn't possible, though, so the real question is where it gets paid. A competitor who keeps losing real matches is usually not filling enough of the quota in practice: not exploring, not taking risks, not finding the holes before an opponent does.

Beginners are the exception, because some of their quota has to be filled in real competition. An old saying from the game of go captures it: "Lose your first 50 games as fast as possible." Early competitive losses buy experience quickly. After that, competitive losing is something to avoid, since the scene judges results whatever you learned.

The competency trap

"A competency trap is what's created when you become so specialized that your entire game revolves around a single winning method."

Edge, Play for Blood: Learning Losses

A competency trap feels safe, because the routine has always paid off. It is fragile for the same reason: an opponent who takes away that one method takes away everything. (Researchers who study how organizations learn use the same term for companies that keep refining an old winning approach long after it stops fitting.) Learning losses are the way out, because experimenting with weaker parts of your game means losing more often in practice.

Learning losses across competitive games

Grappling and MMA. World-class grappler and MMA fighter Garry Tonon is Edge's model. He deliberately starts in bad positions in training and fights to escape instead of playing static defense. By his own estimate he gets tapped at least 15 times a session. On fight night, very little is unfamiliar, which gives him the confidence to attack.

Wrestling and jiu-jitsu. Start rounds from the positions you hate, roll with partners who beat you, and ban your best takedown or submission for a week. Every tap or takedown you give up in the room is one you won't give up in a bracket.

Chess. Play the openings you avoid, and play practice games against stronger opponents. A loss in an unfamiliar structure at the club costs nothing; meeting that structure for the first time in a rated game costs points.

Poker and other high-variance games. Correct play can still lose for long stretches. Edge compares this to baseball, where the best hitters earn huge salaries while failing more than half the time. Judging yourself by the size of long-term payoffs, rather than the count of short-term wins, keeps those stretches from pushing you off a sound plan.

Esports. Scrims are where teams can test risky strategies and lose cheaply. A team that only runs its comfortable plays in practice discovers its weaknesses on stage, where opponents who have studied its footage are waiting.

How to use learning losses

1. Separate the two kinds of losses. Learning losses happen in practice and are welcome. Competitive losses happen in real matches and are to be avoided.
2. Handicap yourself on purpose. Take away a strength for a session or a week and see what breaks.
3. Train with people who beat you. Partners who expose your weaknesses are doing you a favor. Only training with people you can beat is how competency traps form.
4. Detach your ego from practice results. A tap in the gym means nothing in the grand scheme. Treat it as data.
5. Check your quota. If you keep losing real matches, you probably aren't losing enough in practice.

Related concepts

Losing quota | Competency trap | Gym monster | Autopilot | Variance | Mental toughness for young athletes

Frequently asked questions

What are learning losses in competition?

Losses you take on purpose in practice to find your weaknesses, such as starting in bad positions or banning your favorite move. They cost nothing because nothing is at stake, and they teach more than wins do.

What is a losing quota?

The baseline amount of losing every competitor has to go through to improve. Nobody escapes it. The only choice is whether most of it gets filled in practice, where losses are free, or in competition, where they cost results.

What is a competency trap?

A competency trap forms when a competitor becomes so specialized that their entire game depends on one winning method. It feels safe because it has worked before, but it is fragile: an opponent who shuts down that one method shuts down the player.

Is learning loss the same as learning losses?

No. In education, learning loss means academic progress students lose, for example over long breaks from school. Competitive learning losses are deliberate practice losses used to improve.