Lessons from weighted rock paper scissors
Lets play weighted rock paper scissors. Try to find a winning strategy.
Did you find a winning strategy?
The answer is no. You didn’t. The AI is playing a Game Theory Optimal (GTO) strategy which means you can only ever draw.
Run some simulations to check for yourself.
As always I think you should play around and draw your own conclusions, but here’s a few quick lessons:
- Think about “overall strategies” not individual instances.
- Small sample sizes lie. 100 is a small sample size.
- The “unexploitable” strategy is never the “best” strategy.
The first point pops up everywhere. In business I’ve found this translates to “all that matters is your processes not your individual actions”.