We don't need randomness

last edited 2026-06-24

No one really likes randomness. People think they like it because they think they can't have other nice things without it, but they're wrong. It's sometimes harder to design your game without randomness, but it's usually possible and better if you do.

The fundamental problem with randomness is that it's unfair; it arbitrarily advantages or disadvantages a player for something that isn't their fault.

Not all forms of randomness do this to the same extent. Random draw order in card games does it a lot, because you might draw the best card for your situation or the worst, and have no way to predict or control it. Some games like Cobalt Core have a system where enemies do random things, but show you what half a turn before, so you have a chance to counter them. But this is still unfair because your actions on the previous turn can leave you in a better or worse position to deal with certain enemy actions, and you can't predict which those will be more than half a turn in advance.

The only form of randomness that *doesn't* do this is pre-game randomness. For example, Chess960 is a Chess variant that randomizes the starting positions of the back row, but it's the same positions for both players, and it's chosen before any moves, so it's fair. So when I complain about randomness in games, I usually exclude pre-game randomness.

So, let's talk about all the claimed benefits of randomness and see that they can all be had without it.

Variety

Randomness helps keep every match fresh.

It does, but as mentioned above, Chess960 shows a way to do this without in-game randomness. Prismata and Dominion are two non-abstract games that use a similar approach, picking a random set of units or cards at the start of a match for both players to have access to.

Performance variance

In a completely deterministic game, if you're even a little bit more skilled than someone else you'll win nearly all the time. It'll be hard to find opponents you can have a balanced winrate against.

Prismata is again a great counterexample. This is a deterministic pure strategy game with no hidden information or any other "randomness by another name" mechanics, and it has extremely high performance variance. I won a couple of games against top players when I was average, and frequently lost to average players when I was in the top 20.

How to replicate Prismata's performance variance

Excitement

Uncertainty is essential for engagement; if you know what's going to happen it can't be exciting.

This is wrong because you don't know what's going to happen even without randomness. Try playing a game of chess and tell me if you know what it's going to look like 10 or even 2 moves later.

Personally, by far the most exhilarating game I've ever played is the deterministic perfect-information (and turn-based!) Go.

Comebacks

Randomness gives players a chance to come back after a mistake. This keeps the whole game interesting instead of making it get stale as soon as someone takes a lead.

This one is ironic because this is actually a problem mostly caused by randomness. The real solution to being behind is resignation; the only reason this argument even gets raised is because of games where you feel you can't resign because you might come back through luck. I used to play a card game called Spellweaver with a lot of randomness, and I had many experiences where I knew I was behind and was no longer enjoying the game, but I couldn't resign because there was one card in my deck that could turn around the game if I drew it; I had to hang on until the bitter end incase that happened. It was actually the randomness that was causing my misery. When I fell hopelessly behind in Go or Prismata I always just resigned and I was never dissatisfied beyond the inherent dissatisfaction of losing.

It's also possible to have a deterministic game where comebacks are possible and common. In fact strategy games are really the only genre that tend to lack comebacks without randomness. Fighting games usually have no randomness and comebacks are common.

I discuss comebacks more in my page on feedback loops.

Depth

This might be surprising, but there's actually a decent argument that randomness makes games deeper. Making the right choice a matter of probability can make it far harder to know; Spellweaver was actually a good example of this. There were a lot of common situations where the best choice depended on what cards you assume the enemy might have, and so you could never really prove what the best choice was. Top players disagreed. And when I tried to create my own card game, I learned the hard way that removing hidden information, intended to make the game more deterministic and fair, actually trivialized most of the choices and it just came down to deck matchups or draw luck.

But this doesn't mean games *need* randomness to be deep, just that it can be one way to slap a bandaid on an otherwise shallow game. The deepest game in the world is still Go as far as I'm concerned.

Aside: randomness prevents puzzles

A lot of turn-based strategy games have "puzzles", hand-crafted situations that challenge the player to find the optimal move. These are common in Chess and Go, and Prismata has many good ones. They're a good form of extra content to entertain players in a new way.

But randomness severely limits your ability to make these, because if some actions have random outcomes, you have to either:

1) let the puzzle solution not always work, and unintended solutions sometimes work

2) use a fixed RNG seed, which means everything that has a random outcome actually doesn't, and the player just has to use trial and error to find out what each action results in.

Both of these suck, so puzzles in games with randomness usually exclude any elements that could trigger it (for example, puzzles in card games only span 1 turn and you have no way to draw more cards). This also sucks because a lot of the most interesting situations in these games do involve drawing cards to get more options.