SpyParty and the Dunning-Kruger Effect

There is a well-known cognitive bias in psychology called the Dunning-Kruger Effect. The bias is named after social psychologists David Dunning and Justin Kruger who published a study in 1999 claiming that the less competent someone was, the less accurately they could judge their own level of competence. Put another way: part of not being good at things is not realizing it. Or, in bumper sticker form: You don’t know what you don’t know.

This has significant implications for many aspects of life, but the implications for a game like SpyParty, a game about deception and obfuscation, are profound, and are responsible for the single most important skill hurdle in a player’s development.

Your first couple hundred games are almost entirely about mechanics and knowledge: you learn to hit pads, you learn how the different missions work, you learn the minutiae of AI behavior. You learn what a green swap looks like, when to sip, and how long to wait before picking up statues. You never quite learn how the briefcase works, but you get closer. It’s a steady progression where knowledge accumulates, skill sharpens, and your ability to win increases more or less correlatively to the amount of time and effort you expend. Because there is little to do with reflex or aim or anything having to do with innate talent, whether or not to become an Intermediate SpyParty player is essentially a choice: if you try, and put in the time, you’ll get there.

Becoming an Expert SpyParty player is tougher. That the best players fail less often is obvious. What’s less obvious is that, when they do fail, they often know it. They can fail gracefully. They often sense when they’re behind, which means they can take a shot that is unlikely to win them the game, but still maximizes their chances of doing so.

At some point acquiring more arcane AI knowledge yields diminishing returns, and you play people good enough to take (and force) uncertain shots with regularity. And when you reach that point, self-reflection is required to improve. Whereas before you would simply ask yourself “did I see something?” you may have to ask yourself “how likely is it I missed something?” As the clock ticks down, or you hear “Banana Bread,” you have to accurately determine the likelihood that your opponent has gotten something by you, which means you need an objective assessment of your current skill level. To get better, you have to know how good you currently are.

A seasoned Sniper may miss a Purloin early in a game, for example, and spend most of the game knowing they’re behind. This may leave them in a position late where they have to take a shot that they’re, say, 35% sure about. This is a shot that is likely to lose them the game, but may still be the correct shot, if they’re confident that they’re behind and the Spy is likely to complete missions anyway. A shot can be likely to be wrong and still be the highest-percentage shot available.

In order to take these shots, and salvage games after mistakes, the Sniper needs a general sense of not just their opponent’s abilities, but their own. This leads to a paradox where not being that good, but knowing you’re not that good and making decisions as a consequence of that, is what makes you good. This is a game not just about deception, but about self-deception. Flatter yourself into thinking you would’ve caught something that’s already happened, and you’ll hold your shot as the spy completes their missions.

The header of the SpyParty development blog contains brief description of the game:

“Chris Hecker’s new espionage game about subtle behavior, performance, perception, and deception.”

I would add one word to this, and it might be more important than all of them:

Humility.