Does Lichess mark accuracy on a curve. (How am I playing at a world chess champian level in bullet?)

@nadjarostowa said in #10:

Not sure if you are trolling or if it is a language thing. But as my example shows, accuracy very obviously does not show the quality of the play (moves).

So what do you mean with "quality of the play", then? Or maybe it was just a random "look, I got a high accuracy game" post that we better had ignored?

When you say "Or maybe it was just a random "look, I got a high accuracy game" post that we better had ignored?" It shows maybe you were motivated because you didn't like what I said. I was asking on honest question that had nothing to do with me showing off, but maybe the very fact that I showed a high rated game rubbed you the wrong way. Very immature. Who cares about some random person on the Internets high rated game? Not me; Not other smart people.

Think about it. In this case, the question was supported by showing the game, because it showed that I had reason to ask the question, and if someone thought by default "no because games are always rated low" then this example would help remove that criticism. So even though showing the game was helpful to the context of the question, you feel that one should not sure useful information because it might look like they are showing off? Ridiculous. Its not me that is acting out of ego, but you!

A really high accuracy means very few mistakes were made... 100% accuracy means no mistakes were made... that says a lot about the quality of play...Saying that going for a mate in two instead of a mate in when is less accurate is just subjective. Unless there is no time left on the clock it doesn't make a difference either way.

@PowerEnt said in #1:

I noticed that the world chess champagne games tend to be around 97-98% accurate according to lichess game review. I had a bullet game (and have had others) that have the exact same states. lichess.org/PldRW0xBXGgc 0 inaccuracies, 0 mistakes, 0 blunders, 7 average centipede loss, 98% accuracy. This is just slightly better then the last world chess championship game. (97% accuracy, and 1-2 inaccuracies) I understanding that its way different because my opponent had 85% accuracy, but it still feels crazy that I would have world chess champion (long time control) states in a bullet game.

Does lichess save computation power by analyzing lower rated faster games with less precision?
The only difference is that Gukesh or Ding did not blunder the way your opponent did. Its very easy to get a high accuracy if your opponent blunders early on and you play the first to fifth or sixth best computer move.

@PowerEnt said in #11:

[...]

What are you smoking, dude? You asked a question, and people answered. Namely that accuracy alone does not say a lot about the quality of your play. Yes, you may not blunder into a loss, but you can do extremely stupid moves and still get high accuracy.

A really high accuracy means very few mistakes were made...

As I have proven, it absolutely does not. And yes, I actually did try that within a (private) study.

100% accuracy means no mistakes were made...

This only shows that you do not understand how accuracy works.

Although you might say from a mathematical / game theoretical point of view you are right. When you're up 20 points in material, blundering a piece away is not a blunder. But for any evaluation that a human would respect, you'd say it's a mistake.

that says a lot about the quality of play...

Again, no. It only defines a lower bound of the kind of mistakes you made. Nothing more.

I really suggest to look up how accuracy is calculated to correct your misleading assumptions.

Saying that going for a mate in two instead of a mate in when is less accurate is just subjective. Unless there is no time left on the clock it doesn't make a difference either way.

Mathematically correct. Although I didn't go for a mate in two... more like a random mate in 20. But for a human, this is a mistake and not a sign of good moves, which you claim.

When people look at the moves and ask rightfully "are you stupid", I wouldn't say the quality of my moves compare to players of the World Championship. I would rather say "my understanding of what is a good move" and "my understanding of the accuracy number" needs to improve tremendously.

@PowerEnt said in #3:

It doesn't say much about the quality of the player, it says a lot about the quality of the play.
@PowerEnt said in #11:
A really high accuracy means very few mistakes were made... 100% accuracy means no mistakes were made... that says a lot about the quality of play...
I'm not really in a mood to write a long answer explaining why this belief is wrong so just some links to earlier discussions where I and other users did:

https://lichess.org/forum/community-blog-discussions/ublog-pCxNOHqU#5 (comments 5 and 6)
https://lichess.org/forum/lichess-feedback/remove-accuracy-2

Lets bring this back down to earth. My game shows 98% accuracy. You say that doesn't mean much, yet, it turns out, none of my moves gave away a big advantage, so the accuracy does seem to correlate with quality of play. If you check other games of very high accuracy against the computer you will find the same thing; all the moves give away very little. I feel like I'm explaining this to little kids for the first time...

This is not how logic works.

A => B doesn't imply B => A. It's only -B => -A.

If you flag, you lose. If you haven't lost, you didn't flag. But if you don't flag, you can still lose!
All cars have four wheels. If it doesn't have four wheels, it's not a car. But not everything that has four wheels is a car.

Or applied to accuracy and "good moves":

Good moves gives you a high accuracy.
If you don't have high accuracy, at least one of your moves was bad.
It does not mean: if you have high accuracy, you had only good moves.

The fact, that in the game shown you moves are indeed rather good just cannot be implied from the accuracy score.

But as mentioned, the main problem is that the accuracy score gives not indication whether those moves are easy to play (like in your case, where most moves simply win because the opponent blundered - it would be hard to find bad moves!), or if you need to find difficult only-moves.

Often, the accuracy score says just as much about your opponent's play as yours...

Your suggesting that lichess is lying to all of us? When it says a percent for "accuracy", it doesn't actually mean a percent for "accuracy" ? You're going to have to take this up with lichess, not me.

Did you even try to understand what accuracy means? Here you can find some details:
https://lichess.org/page/accuracy

Lichess doesn't lie about accuracy. In fact, it states quite clear that high accuracy does neither imply grandmaster play, nor cheating.

It's just that your interpretation of it is wrong. It simply doesn't imply what you think it does, and nobody here except you make those claims.

This is probably be my last response in this thread. So far you have ignored all explanations. You have been given the links to educate yourself. If you chose to stick with your wrong view, at this point trying to convince you seems a waste of time. But I'd kindly ask you to stop spreading that misinformation and confuse others.

I never said it did imply grandmaster play or cheating. You crazy

Consider an analogy...

A race car driver can go around a track pushing himself and the machine to the limit. The smallest error could send him careening off the track. If that happened, we could call his driving innacurate.

An average person drives a race car around the track at 1/10th the speed of the race car driver. His driving could be completely erratic, but he is in no danger of leaving the track at those speeds, so technically it could be classed as accurate if he has no mud on the tires.

However, is it in any way realistic to say that the second "accurate" drive was in any way better than the first "innacurate" one?