Fun With Graphs

Great work! I wonder why move time would indicate cheating? Does the bot take the same amount of time every turn?

For people, the more complicated the position, the slower the moves.

For bots, the move time depends only on the search depth setting.

I've looked at, IDK, 15 or so people.

It's easy to assume people like Hikaru are honest, because they have sooo much to lose (all that streaming money) vs very little to gain (a few rating points). And it's easy to know what cheaters look like because you can just analyze games from accounts closed for cheating.

So I didn't 100% believe B1Z was honest... but you'll all be happy to know B1Z's graphs look exactly like what I'd expect from an honest player

You'd be surprised... I've had a few people say stuff like "My wife is always bugging me when I play, so I'm sure it looks different"

But nope. It's kind of boring... basically everyone looks the same lol.

But sure, I'll do your blitz games.

Yeah, exactly. I like to think of it as a mental load... over 100 games a range of positions are sure to be represented, broadly falling into a few classes, each requiring a different amount of effort.

If a graph looks too far off, then we can say with some certainty that the player who played those moves wasn't experiencing the mental load expected of a human player

Could you do that for me?

Yay! I am surprised by the normality.

For me, I can assume B1Z is honest because cheaters typically stay incognito, or at the least are not one of the most active members on the site, but just getting ill-gotten rating points.

yeah, generally they do.

Here's a rapid game I played months ago (I was black) against someone who was banned for cheating.

Notice how white's time is usually in single digit time, while black's time varied by a lot.

if you go onto the chess.com official discord, and look for people rated above 2200 rapid, and they talk sh.t, say "1v1" me, and respond to chess questions with extreme generalizations, and say they don't like playing blitz because it's too fast, you will be able to find many cheaters.

This is your last 1000 blitz games.

More squished to the left because they're faster games of course... but the most common move time for (nearly) all players is 1 second... unless looking at extreme time controls like hyperbullet or 30|0 games.

And then of course it falls off exponentially from there.

---

Ok but I have to go for now. Spent too much time on this today.

Wow! I have several moves at over a minute long! Evidence of my spotty time management. 

Yeah, I agree with everything @b1zmark has said about cheaters, but that sorta topic will get this locked... which is fine, but I'll try to avoid it.

Eh, maybe. If we assume a game is 40 moves, then...

Each image is 50 games which is about 2000 moves.

All together I did 1000 games for you, so 40,000 moves(!)

Having a few over 1 minute doesn't really mean much IMO. Maybe someone interrupted you, or it took extra long to change to a song you like you youtube or something heh.

True, but I estimated about 20 minute-plus moves. I can't be sure about that number, but if it is accurate, that means I thought for a minute once every other game in a 100-game span!

*1000-game span*

1 out of almost 5000 of your moves took more than 1 minute.

All together your shape is this:

-

-

Which I agree is just a tiny bit slow. Maybe because I mostly looked at higher rated players (remember I was trying to get cheaters) but usually 1 second was the most common, followed by 0 seconds (premoves). You have 2 seconds as your most frequent and premoves as your 5th most common.

For comparison B1Z's blitz games are this:

-

-

... see what I mean when I say all the shapes are the same? lol

Except of course for [cough] certain types of players

ah ok could you do it tomorrow?

Some of your rapid games are 10|0 and some are longer, so it's more bumpy and of course you have more moves on the right side.

-

---

All together:

-