BANNED For Cheating.
28.04.2024
➡️ READ THE ARTICLE:
➡️ Get My Chess Courses:
➡️ Get my best-selling chess book:
➡️ My book in the UK and Europe:
➡️ Mein Buch auf Deutsch:
➡️ Mi libro en Español:
➡️ Start Playing Chess FOR FREE:
➡️ Enjoy my videos? Donate Here :
Email me your games: [email protected]
Sponsors, Business, Media: [email protected] – [DO NOT SEND GAMES HERE]
⭐️ Follow Me If You Are Amazing:
➡️ CAMEO:
➡️ FACEBOOK:
➡️ SNAP:
➡️ INSTAGRAM:
➡️ TWITCH:
➡️ TIKTOK:
➡️ TWITTER:
➡️ GOTHAM DISCORD:
➡️ THUMBNAILS BY:
Kramnik's position is perfectly sound. Why read a report if the methodology can't yield meaningful findings?
lol what an insanely wrong thought: The rates of upsets live and online are different, which is evidence of an ABSENCE of cheating.
First, if cheating online is easier than live, one would predict to find different rates of upsets as a result. Second, this proves nothing, because, if one would predict a difference in rates of upsets live vs online for ANY REASON AT ALL OTHER THAN CHEATING, then finding such a difference isn't evidence of cheating (or an absence of cheating). Here's a plausible mechanism: better rated players are better at controlling their nerves and are more experienced at over-the-board chess, so they get upset LESS live.
if it happens online, compare perfornance measure between online and OTB, and use the Chi-test, very easy and established
So a problem with this report is that you would expect a bigger variance in the OTB data, given that way less blitz games are played OTB. This could be easily debunked if the report presented that the OTB games are fitting with the expected results distruition from the Elo curve. They are picking 36k OTB games vs 167k title tuesday games, you have to show that this is not impacting your results.
Another point is that I would eliminate games that ended in flagging, as its easier to flag online due to pre moves, that might skew the data set one way or the other.
Levy is A Stone Cold G in my book.❤❤
Why is levy all serious and riled up
Banned for clickbait.
These numbers are so satisfying! I never thought someone would be able to have some proper research, but the numbers don't lie. Very cool stuff!
levy blink twice if you’re still being held captive
P-values can be deceptive – if the study was underpowered or the distribution doesn't match their test assumptions then p-value will likely be bogus. I'd like the know what their confidence intervals (and the shape of underlying distributions) are like. I suspect your (Gotham) hypothesis that each individual's performance level is so highly variable that this methodology simply isn't going to discover anything. I'd extend this to say that over the board performances are even more highly variable because human factors are often our of your control at these events, but the number of games you get to play are far fewer than online – which means the noise of OTB will cause biases in online play to be masked and appear insignificant.
I wonder too if looking at the ratio of win vs draw might be more useful than directly comparing win rate alone, because engines will drive for a win, when humans will often be prepared to accept a draw vs a stronger opponent, or even just to secure their position in a tournament.
come dublin levy
Nice clickbait by the way!
I havnt played much chess, but it sounds so easy to cheat.
Cant you just grab a tablet, load a second chess account (Or stockfish, or any other service) and just load the moves as they go then just do what the computer says (85% of the time) ? If you're already a top-top player, use it 10%~ of the time to edge a win?
As someone who is 1200 blitz ive beat 2000s a couple of time in online tournament because they dont take it seriously then make one little mistake. Lock the position, hold the pressure.
Shit even my mom beat me one time because of a simple blunder.
I would agree cheating online is way too easy. I’m not so sure about at your level that it happens often, but I’m positive it happens. At my low level it happens about every 1 out of 4 games I’d say. This is just off of my experience. It’s wild when you watch a 455 ranked player uncork a 96% accuracy in a game with no mistakes and no blinders. And for what? We arnt playing for money.
Levi never fails to 240p
Broooo. If you went to Amsterdam and DIDNT record running the board on a whole coffee shop I’m going to be VERY disappointed.
But for real though, Levy is right. Who the hell makes a statistical analysis and doesn't back it up with hypothesis tests and statistically significant p values?? Do they not know how to?
imagine being that low iq to cheat with low blitz ratings. I am not saying it doesn't happen, but there are certain external factors that data cannot show with slight enhancement. Take in for example professional athlete, sometimes PED players would perform better than their opponent. If applies to here, its obvious that sometimes you can use the engine to get an slight advantage, how big of difference is subjective, but nevertheless exists.
I'm not a statistics major but I do work in data science. Reflecting on Levy's "p-values" comment, isn't 5 percentage points (in either direction) a gigantic difference for this sample size??
Like I'm almost tempted to draw the exact opposite conclusion from this data – i.e. we have statistically significant signal that there is a difference in how elo predicts results of Titled Tuesday games and OTB blitz games (regardless of whether it errs on the side of the higher or the lower rated player). Whether that's because of cheating or for any other reason, I don't know, but "empirically" it looks like something is going on.
The problems with the data collected almost make it seem like they don’t want there to be cheating
Thank God there's a new video out. Skipping a day after a long overnight shift is not good for my mental health.
Can u gen martin vs stockfish
As a 1800-ish player that has played for years I swear if I had the time, inclination, and effort I could develop a system to cheat and not get caught. But why? Who cares? But, I bet I could.
Happy with my 1500 rate.
I mean this is the correct year for the rookies to prove themselves you know and maybe you will find these people across the board for sure though.
Obviously spatial awareness and the amount of time and effort it takes to actually move and strike the clock as compared to a mouse click/premove heavily influence the outcome of OTB and online play.
Lemi, speaking of Kramnik, you always should say a line for the Toiletgate Scandal in his match vs Topalov.
Your barber is cheating on you
You should have explained exactly what Tilled Tuesday is instead of assuming we know.
Levi come to play the Sants Open in Barcelona pleaseeeee ❤️
Each time i hear the word "interesting" i can only think of Kramnik now lol
Statistician/Data scientist here. Obviously these stats neither prove nor disprove cheating, however, they paint a picture of not wide spread cheating. I imagine if 5% or less cheated this wouldn’t reliably show up in these stats since the effect on the aggregate might be minimal. What would be interesting is actually looking at “proven” cheaters and comparing there results if found on title Tuesday events. (If enough data).
Or does „scoring“ means in the report, that they scored also .5 points for a draw?
Extending an invite to India from all the Indian fans amidst all the Invites of Europe! 🙂
In other news, Philip Morris put out a report showing that doctors endorse their product for a healthy lifestyle, despite fearmongering by others in the medical field.
A problem with Kramnik's suggestion to use cc ratings is that a reliable strength estimate is needed to correctly register underdog victories. When someone cheats regularly, their cc rating quickly adjusts to the cheating strength. For instance, if a 2000 player cheats their way up to 2300, they become undistinguishable from a a player who's actual strength is 2300. For the cheater, every win against a 2200 is an underdog win, but it will not be registered for the analysis because it looks like a regular win. For this reason probably most underdog wins would go unnoticed, undermining the point of the analysis. FIDE ratings don't have that drawback because you can't inflate them by cheating.
It's so annoying to see Kramnik boss everyone around as if he was a pro at detecting cheaters, while apparently missing such a basic detail.
Levy's on a boat!
Come to Paris so as you also meet with hate and not only love in your European tour
Will you be visiting Ireland 🇮🇪 ??
Would be a funny video idea to pick a super GM upset, and then go down the chain of that person’s upsets and so forth to see how low of a rating you can get to. Clickbait title being Hikaru isn’t as good as a 800 or something to that effect.
Gothan you gotta stop with these ridiculous thumbnails
I know I’m getting clickbaited, and yet, I click.
This might be controversial, but I think blitz is the wrong time control to look at while trying to calculate how many people are cheating. When playing 3+1, you don’t have time to look at the computer every move and spend 5 seconds on an obvious move, especially when you also want to stay hidden by “thinking” or playing worse moves. I believe 5 or 10 minute time controls are better for this, since it is still impossible to cheat in such events OTB, but it is easier to cheat (and thus would be expected more) online. I realize that this report is about title Tuesday, but for an actual estimation of cheating it is way less accurate. It’s good to see Kramnik getting his arguments fact checked tho
Your one sentence summary ignores a possibility. It is possible that rated players are cheating more, perhaps due to ego, or for some other reason.
I'm not saying that is what's occurring, just that it is an outcome that cannot be discounted. Maybe you address this point later in your video.
I’m pretty sure the Confidence Interval won’t show significance
The major issue with their analysis is the lack of like for like data sets. It means their supposed variable is not controlled.
This would be genuinely interesting if they analysed Titled Tuesday results vs a contolled similar sized data set of players playing normally (not in titled tuesday). That way only one variable is being altered.
FIDE results are simply not stable enough or large data enough for this kind of analysis. Chess website that published this have an extremely stable data set because of the 1000000s of games played, whereas FIDE played too little in this format and new players affect the results too a far greater degree within that data set.
doesnt that defeat the purpose of ELO rating? isnt there a relation between the two ratings and the probability of victory/defeat between two players?
BUT, Does the underdog who played and won offline also play online and vive versa ?. How come the report fails to talk about this ?
Hikaru Ratings is inversely proptional to Kramniks Reputation now a days 😂