[Syd Thrift]

I wrote this gigantic rambling essay on how I think AI could be "improved" (read: made even more sloppy) and I wanted to include something about probability but the thing was already too long. So instead I'll put it here.

The Stand Up Maths guy, Matt Parker, did a thing with the woman who does the Numberphile YouTube channel, Hannah Fry, where they demonstrated Bayesian probability by putting an X on a table, and having Matt toss beanbags onto the table to try and figure out where the X was. He couldn't see the table and was only told a. where his beanbag landed, and b. if it was to the left or right and up or down from where the X was. I'm having problems finding that video - I guess it's pretty old - but trust me, it's out there! If you go to either of their YT pages you'll probably see it.

Anyway, it strikes me that this is basically what baseball, or at least the OOTP implementation of baseball, is. For any given hitter, you have, say, a weirdly shaped, loaded die, and rather than being told what the given chances are of that die landing on any particular number, all you get is the face it came up as. On any given day, the "die" that comes with a .240 hitter could go 0 for 5 or it could go 3 for 5. Over time, you hope that the result set you get is indicative of the shape of the die, but even with everything else being equal it's really not: over the course of a 550 at bat season the standard deviation for batting average is IIRC around 30 points. So if you roll that one misshapen die 550 times and that die's real, actual value is it should give you a hit 27% of the time, 1/3rd of the time you make 550 rolls it will give you hits 24% of the time or less or 30% of the time or more.

But on top of that, everything isn't always equal. On any given day a player could be facing a pitcher they're especially bad at getting hits against (this is probably much less of a factor in OOTP than in real life but it is a small factor... although handedness is potentially a kind of big one). You'd hope that this evens out over time but depending on the size of the sample it may not. Some hitters do very poorly against certain pitcher types - a guy with a low Avoid Ks for instance tends to be just plain destroyed by a pitcher with high Stuff because of the way the game engine works, where it determines the Three True Outcomes before it looks at anything else (FWIW I think this is a pretty accurate approximation of how baseball works from a probabilistic standpoint - of course, real life baseball, like real life everything else, is based on... a lot of real life stuff). A guy who does nothing but pinch-hit likely faces more quality pitching than a guy who plays in the lineup every day. A guy fighting an injury will make his "die" come up with fewer hits.

And then on top of that the die can change over time, due to aging and development but also just due to random chance. In real life of course it's never ascribed to random chance - a pitcher improves because he worked on his mechanics, a hitter suddenly stopped hitting for power because word got around the league that he couldn't hit the curve and so he started seeing nothing but curves - but in video games you can't literally account for everything you could possibly account for and so you need randomness instead. And also, crap's random in real life way more than people think it is but humans are wired to hate randomness as an explanation.

Anyway, what I'm getting at here is that this is why baseball interests me from a nerdy stathead perspective. You never truly "know" if a player is as good as they're hitting at any given point in time. I mean, very, very improbable right now that Mike Trout is only average at hitting baseballs, but there's still a tiny, tiny chance that that is true. And for guys like, say, Evan White... I am reasonably sure that he's not a .144 hitter, but can I really and truly be positive of that? No, no I can't, but the only way I can really know for sure is by putting him in the lineup every day.

So... as it applies to playing and watching baseball... I think people should look at this as a feature rather than as a bug. Embrace not knowing for sure if your .105 hitting second baseman is just in a slump or if they're really that bad. Turn off the things that tell you definitively if they are. Those are cheats, essentially. You don't get that in real life, not in baseball but also not in whatever it is you actually do. In baseball you do get scouts and scouting reports who tell you they think a guy has a swing that projects him to Ken Griffey Jr. or a fastball that hits 99 on the gun, but those same scouts told GMs they should draft guys 25 years ago that because they had a good "baseball face", and who is to say that stride mechanics or FB velocity aren't the baseball face of 2021?

Play the game (and live your life) however you want to, of course, but I feel like people in general just spend way too much time trying to figure out what is a sure thing and what isn't instead of accepting that there will always be risks and chances. That doesn't mean you should do stupid, risky things - that's part of "grokking" probability, understanding that a 0.1% chance is super low but if the stakes are high enough and the consequences of doing the other thing aren't very large at all, there's no reason to even take that tiny chance - but it absolutely means that you need to get comfortable with uncertainty. Sometimes stuff will just not go your way and life will suck. Refusing to take a low-risk, high-reward move because of this isn't a good way to go about life, though. And conversely if you're lucky enough to have had some good things go your way, it's good to have the humility to accept that at least some random chance was involved to get you there (even if you, say, also had to work hard to get where you did).

(sorry if I'm speaking in generalities here - I personally like to be much, much more specific, generally, but I also feel like there are very particular political rabbit-holes that the above paragraph could go down and rather than do that I'll leave you to think about what these situations and this thinking means for yourself)

So, you know, trust science and analytics and all of that... but also accept that not really and truly knowing deep and underlying truths is part of baseball and of life. And given that it's always going to be there and won't go away no matter what you do, try to have fun in the sea of improbability. The world is chaotic. Embrace the chaos!