💾 Archived View for yujiri.xyz › game-design › performance-variance.gmi captured on 2023-09-08 at 16:30:54. Gemini links have been rewritten to link to archived content

-=-=-=-=-=-=-

Performance variance

last updated 2023-09-06

I've touched on performance variance in my randomness article and gave Prismata as an example of a game that has a lot of it with no randomness, but didn't give much advice on how to replicate it.

A little bit of randomness is not okay

I've gained some insight. Prismata achieves its performance variance by having a few decisions (the first few turns) meet a few criteria:

They have the highest skill ceiling in the game. A good player plays close to optimally for most of the duration of most Prismata matches, but even the best players often choose the wrong opening.

There are few options to pick from, so a guess based on a poor understanding of Prismata theory is reasonably likely to be correct.

They are the most important decisions in the game.

Combined, these factors mean that a weaker player has a significant chance of playing a better opening than a stronger player, and that if they do, they have a high chance of winning the game.

If you can have a few decisions in your game copy these traits, you can replicate Prismata's performance variance.

But, this combination of traits by itself is likely to entail a frustrating faux feedback loop since one early mistake can trump a match full of better decisions. Prismata has a solution for that too which should hopefully be copied at the same time: it's not obvious who's winning. I explain this in the feedback loop article.

Feedback loops