Reading the game

Most games are games of risk, not true uncertainty.

Hello friends,

tl;dr: Most games are risky, not uncertain — they have clear, unambiguous, rigorously enforced rules and winning conditions. That’s why they don’t prepare us well for important real-world situations of uncertainty. To prepare for true uncertainty, it’s crucial to a) learn how to read the game (i.e., distinguish between risky and truly uncertain situations) and b) build routine exposure to well-designed situations of true uncertainty into everyday life.

Here in Marseille, the hordes of summer tourists have begun to dissipate but they have been partly replaced by Large Rugby Fans in Team Kit. Fortunately, the LRF’s natural habitat is the Fan Village at the Vieux Port instead of the city beach — which has finally returned to an enjoyable level of democratically crusty congestion.

Midweek at Catalans in September 2023 — democratically, but tolerably, busy.

I sometimes get into conversations with people about uncertainty (which is /= risk). Almost every time, I will assert that we increasingly rarely put ourselves in situations where we explicitly confront true uncertainty and learn to live productively with it. Almost every time, my interlocutor will crow in triumph: “You’re wrong! People play games all the time. And the final outcome of games is always unknown!” Or something to that effect.

The idea, as far as I can tell, is that playing games where the final outcome isn’t known in advance is supposed to be adequate prep for real world situations of true uncertainty like climate change or entrepreneurship (or war or terrorism).

To me, this doesn’t make sense.

It’s true that people play games because their final outcomes aren’t known in advance. They sometimes become very good at playing these games and use them to make money or become famous or whatever. But this doesn’t make most games situations of true uncertainty.¹

Nearly all the games I can think of have two properties which make them closer to being situations of formal risk, not of true uncertainty:

Property 1: Game rules are unambiguous, bind all players, and are enforced. These rules determine how the game is played — and players are kicked out or penalized for violating them, even if the violation is accidental. 

Property 2: Game winning conditions are defined in advance, unambiguous, and easily observable. Players all know in advance what it takes to be considered to have won.

When a game combines both properties, players know all possible (i.e., allowable) actions and outcomes in advance, and they also know the relative value of possible outcomes in advance. Property 1 reduces not-knowing about actions and outcomes, and even causal not-knowing, while Property 2 reduces not-knowing about value.

Games with both Properties 1 and 2 can therefore be “solved,” even if the solution requires computational capacity far in excess of what a human possesses (e.g. Go, which is at least PSPACE-hard). This makes them much more risky than they are truly uncertain. Most games have these two properties.

Some games partially suspend Property 1 by making some rules ambiguous within clear boundaries. Poker, which permits bluffing (pretending that you’re playing better cards than you actually have), is a good example of a game where the rules allow players to strategically introduce some not-knowing about actions. Property 2, though, always seems to hold: Players know in advance and agree on what the win condition looks like.

Petanque also (sometimes) partially suspends Property 1 — but Property 2 holds.

Here is the problem with playing games to train for reality: No important real world situation of uncertainty conforms to Properties 1 or 2, let alone both of them.

Whether it is the uncertainty of climate change (on which see more below), or the uncertainty of launching a company based on still-emerging technology/use-cases, it is never true that all possible actions, outcomes, and definitions of value are unambiguous and known in advance. These different types of non-risk not-knowing make real world uncertain situations hard to deal with (much harder than risk situations) but they also create space for innovative and creative solutions which don’t exist yet.

I’ve written before about the importance of recognizing whether the game being played is mostly risky or mostly uncertain — the importance of reading the game correctly — and how dangerous it is to play truly uncertain games as if they are only risky. It’s both easy and common to mix up risk and uncertainty (for more about these mix-ups, read this and this).

Learning to play the meta-game of reading the game correctly is a prerequisite for figuring out appropriate strategies for these situations of true uncertainty which are different from strategies appropriate for situations of formal risk.

In the meantime, well-designed routine (even daily) exposure to true uncertainty is a form of Mithridatism in an increasingly uncertain world: idk is a tool for progressively exposing users to situations of true uncertainty designed for learning and growth.

The fog of time

Episode 9 of my monthly Interintellect series is about how futurity affects not-knowing, and what we can do about it. Our collective difficulty with thinking and acting on climate change (just one of many Big Deal wicked problems) show why it’s important to understand how the future affects not-knowing in the present.

Fog on the beach at Ericeira, February 2020. A frequent topic of conversation during that trip — nearly the last trip I took out of the UK before the world shut down — was where to go and what to do in the event of global catastrophe.

Climate change was once comfortably a few generations away. This magical thinking is becoming less believable as increasingly unpredictable, extreme, and unseasonable weather (such as high heat, intense cold, drought, excess rainfall, and elevated ocean temperatures) becomes normal. Yet we still don’t have good ways to think about what to do about it in the present. Climate change is full of true uncertainty of various kinds (not just risk), and the future-ness of climate change compounds this not-knowing — we urgently need better ways to think about how to act on truly uncertain futures.

So, this Thursday, we’ll discuss five levers by which futurity changes different types of not-knowing, and the actions we can take to influence these futurity levers (see the diagram below). The discussion is open to all and is part of a broader investigation of not-knowing.

More information and tickets here; you can also find the pre-reading (a 6-8 minute read) here.

See you soon,

VT

1. I’m sure there are games which represent situations of true uncertainty — i.e., where the game is designed around one or more of incompletely known actions, outcomes, causation, and/or value. I can’t think of many at the moment (other than something like the fictional game Azad in The Player of Games where the uncertainty is a major #plottwist) so please send pointers to actual games like this. Update (19/10/23): The Glass Bead Game (in the eponymous book) and Calvinball (from Calvin & Hobbes) are two other fictional games like this.