Originally posted at https://www.threemonkeymind.com/wgir/mario-kart/.

“Rationality techiques helped me to become less bad at Mario Kart” isn’t the most compelling elevator pitch for reading The Sequences, but it illustrates the usefulness of reevaluating the things you think you know every so often, especially when you’re making decisions based on the things you think you know.


Rationality gives you a mental toolkit to help you update your beliefs based on evidence. This is not always easy. Orwell once remarked that to see what is in front of one’s nose needs a constant struggle and it’s useful to have mental techniques, habits, and sayings to help you model the world as best as you can.

I recently got back into playing Mario Kart 8 on my own Switch. This wasn’t my first time playing Mario Kart, though. Back in the day, I managed to get first place in all Mario Kart cups at the highest difficulty. I’d played about three hours of Double Dash, and I’d played a bit of Mario Kart 8 at a friend’s house. When my friends play multiplayer, they all race at 200cc, so I tried to pick a character and loadout that was up to the task[1]. Because I didn’t know any better, I opted for Yoshi in the Standard Kart, Standard tires, and the Super Glider. I remembered reading that bikes were “more advanced”, so I didn’t try them initially. After a couple rounds of doing abysmally poorly, I tried one of the bikes thinking they’d be, at the very least, very different. After all, the motorcycle was very different from the cars in Stunt Race FX. I was right — I went from “abysmal” all the way up to “mediocre” just by changing my vehicle type. A massive improvement!

Several months later, I was sitting at home when I decided I wanted to get gold trophies for each of the cups in the single-player version of the game. I figured that I’d start out with the most basic loadout, but in the interim I forgot how well I did on the bike. Remembering that “bikes were for advanced players”, I ended up picking the same racer and loadout: Yoshi, Standard Kart, Standard tires, Super Glider. With this loadout, I was only mediocre. While getting 5th place is definitely better than trailing everybody else in 12th place, I wouldn’t be able to get a gold Flower Cup trophy anytime soon if all I could do was 5th place on the first circuit in the cup.

While I didn’t expect to dominate Flower Cup, I consistently did much worse than I hoped. My biggest problem was oversteering and understeering while sliding. The kart never seemed to do what I expected it to, especially since I was still getting used to the new post-slide boost mechanic.

At this point I decided to try the “more advanced” Standard Bike instead as I remembered doing a bit better with it at the friend’s house.

This seemed to be the change I needed; I took first place in Flower Cup’s first circuit, Mario Circuit. The bike behaved in ways I expected. I over- and understeered much, much less.


Now then. I had a belief of “bikes are more advanced”, but clearly this belief was keeping me from doing what I wanted to. The minimal-fanfare way to update your beliefs in this case amounts to shrugging your shoulders and reasoning “guess that half-remembered meme about bikes was way less useful for me than memories of how much better I am on a bike”. Generally, one doesn’t need to spend too much time mulling over one’s cognitive errors.

That said, sometimes it’s useful to pause a moment and mull over an error in cognition. After all, I’d like to make this error less often in the future. I like snowclones as much as the next guy, so I looked up the Litany of Tarski and amused myself by writing this:

If I am better at Mario Kart on a bike,
I desire to believe that I am better at Mario Kart on a bike.
If I am better at Mario Kart on a kart,
I desire to believe that I am better at Mario Kart on a kart.
Let me not become attached to beliefs I may not want.

From what I’ve experienced, both with Mario Kart and in other situations, the Litany of Tarski is handy to keep in mind when faced with conflicting information sources of imperfect reliability. It’s even more useful to keep in mind if there’s a chance you over- or undertrust a given information source, as I overtrusted what was probably a throwaway phrase in a GameFAQs guide.


  1. A racer who’s viable in slower circuits can be too slow to compete in faster circuits. In the original Mario Kart, for example, I chose Toad (slow top speed, easy to control), but eventually had to pick Yoshi (moderate top speed, accelerates quickly, somewhat harder to control) in order to keep up with the computer-controlled opponents. ↩︎

New to LessWrong?

New Comment