Checking if Elo system is oppressive [With proofs]

Sort:
RandomChessPlayer62
jankogajdoskoLEM wrote:

Low elo is full of underground masters because they are perpetualy stuck and gatekept, i was 2311 fide 13years ago before they wiped my records clean and threw me into jails and mental institutiin, i was and still i am persecuted and watched.

How come when I play these "masters" in blitz I win?

Alexeivich94
RandomChessPlayer62 wrote:
jankogajdoskoLEM wrote:

Low elo is full of underground masters because they are perpetualy stuck and gatekept, i was 2311 fide 13years ago before they wiped my records clean and threw me into jails and mental institutiin, i was and still i am persecuted and watched.

How come when I play these "masters" in blitz I win?

You really bother to respond to that bait

MasterJyanM

JUST DO A BUNCH OF TOURNEYS AND SEE WHO IS IN AVAERGE, THE BEST PLAYER!

basketstorm
RandomChessPlayer62 wrote:

Low Elo is chaotic not because of the system but because of the players at low Elo, low Elo is a trap because players play randomly before eventually learning strategy, this means that low Elo players may randomly make an absolutely brilliant move that wins the game by accident, trapping low-skill players in the chaos until they become strong enough to withstand the occasional brilliant and exploit the frequent blunders. It has nothing to do with the rating system itself. In your simulation, the low 1000s replace the chaotic 100-300s, but then due to your simulation using probability to represent strategy they become chaotic, requiring extreme skill to escape the new low Elo zone, trapping the 1000s away from their real rating. It's not a product of the system but a quality of the players whose real rating is low Elo causing chaos that can't be easily escaped by low Elo players.

There’s no such thing as a "real rating" as some absolute value - everything is relative. We must seek for fair difference in ratings. You’re trying to dismiss the actual skill of low-Elo players by labeling their play as chaotic or random. That's not nice. Their skill levels vary greatly. There are strong and weak players even at low Elo. We can all observe this: some low-Elo players consistently perform better or worse than others, yet they struggle to break through the low-Elo gates (of Oppession) because, occasionally, they meet even stronger low-Elo players. This prevents their rating from reflecting their true difference from other, weaker players.

The problem isn't randomness or luck or cursed chess pieces or "equal" lack of skill. The issue is the fluctuations in ratings caused by the system failing to distinguish players of varying skill levels. With a starting rating of 200 and a floor of 100, there’s simply not enough room for meaningful differentiation, and rating adjustments are disproportionately large for each game. For example, if a 150-rated player loses a single game, they can lose 5% of their rating when playing against another 150-rated opponent of UNKNOWN and incorrectly estimated (relatively to others) skill. This creates an unfair system where players are constantly being misjudged.

One possible fix is shifting to a higher starting rating to help reduce this problem. However, this would inflate the ratings of higher-rated players, which isn’t ideal. Simply adding 1500 to everyone’s rating would result in bizarre numbers - turning 2000-rated players into 3500s, for instance.

A more balanced solution would be a sloped, linear adjustment where lower-rated players get a more significant bump, while higher-rated players receive only moderate inflation. This could help uncover a broader range of values for players who are weaker than most low-Elo players but still deserve differentiation.

For example, we could adjust ratings like this (if 3500 is max on the platform):

NewRating = OldRating + (3500 - OldRating)/4

2800 becomes 2975
1800 becomes 2225
800 becomes 1475
400 becomes 1175
100 becomes 950

With this adjustment, we create more room below 950, allowing a starting rating of 1000 or 1500. Now weaker/stronger players can move down/up more naturally and consistently with less frustration and more equal pairings.
Many chess orgs conducted rating adjustments several times, it's not a crime.

Does this solve all the issues? No, unfortunately not. Problems like pool localization and other issues I’ve discussed still remain. However, this adjustment would make a significant improvement to how ratings are distributed and how low-Elo players are evaluated.

RandomChessPlayer62

I've never played any of your "strong" low Elo players, if they're strong they can get past the chaos. You may say that they will meet a stronger low Elo player, but then what keeps them from advancing? Another stronger low Elo player? It's low Elo players all the way down! There's a finite number of players in low Elo, they have to advance eventually.

basketstorm
RandomChessPlayer62 wrote:

I've never played any of your "strong" low Elo players, if they're strong they can get past the chaos. You may say that they will meet a stronger low Elo player, but then what keeps them from advancing? Another stronger low Elo player? It's low Elo players all the way down! There's a finite number of players in low Elo, they have to advance eventually.

The idea that a player is either "strong enough to break through" or "not strong enough" isn't far from reality but it oversimplifies how skill works - skill is relative, not binary. And it ignores various skill levels below that break-through margin.

It’s not that these "strong" low-Elo players can’t get past the chaos because they’re not good enough. The issue is that the system itself creates barriers. Even if a player is stronger than most of their low-Elo peers, they can get stuck because the rating adjustments are too volatile at the lower end.

For example, a strong low-Elo player may lose a game against someone rated slightly lower (but with greater actual strength) and have their rating drop significantly, even though they’re improving overall. Fluctuations are more pronounced at lower Elo because the ratings are compressed into a narrow range.

RandomChessPlayer62

Alright then, I'll agree with that, just that though, one small mistake that makes low Elo unplayable, nothing else in your argument unless I am shown irrefutable proof

RandomChessPlayer62

And I'm more focused on overthrowing a dictatorship than out-mathing you.

basketstorm
RandomChessPlayer62 wrote:

Alright then, I'll agree with that, just that though, one small mistake that makes low Elo unplayable, nothing else in your argument unless I am shown irrefutable proof

I appreciate that you're open to some of the points, but I’ve already provided mathematical and logical proofs throughout this discussion that show why the rating system struggles to accurately reflect skill at low Elo. The problems I’ve mentioned, like rating volatility and the lack of differentiation among players in this range, aren't just based on opinion - they’re mathematically verifiable with basic calculations.

IndianCamels
basketstorm wrote:
RandomChessPlayer62 wrote:

Alright then, I'll agree with that, just that though, one small mistake that makes low Elo unplayable, nothing else in your argument unless I am shown irrefutable proof

I appreciate that you're open to some of the points, but I’ve already provided mathematical and logical proofs throughout this discussion that show why the rating system struggles to accurately reflect skill at low Elo. The problems I’ve mentioned, like rating volatility and the lack of differentiation among players in this range, aren't just based on opinion - they’re mathematically verifiable with basic calculations.

Graphs can lie.

TitanMaster101

Ur method is better than @JankogajdoskoLEM’s. You two should get together. As for elo gatekeepers, just get a coach. I’m sure that will help

TitanMaster101

And if you want to play against higher rated people, just go to open challenges

TitanMaster101

This is just a theory, but @basketstorm did join 24 days ago, so is it possible that it is @JankogajdoskoLEM’s second account? And bro just joined for barely a month and is already complaining

IndianCamels
TitanMaster101 wrote:

This is just a theory, but @basketstorm did join 24 days ago, so is it possible that it is @JankogajdoskoLEM’s second account? And bro just joined for barely a month and is already complaining

No, because this guy has some proof behind is arguments and isn't saying that he is an FM thats stuck at 500. Also this guy has proof, however correct it is. You have to give him that.

TitanMaster101

True, but it still could be Janko just using another method to convince people

TitanMaster101
xtreme2020 wrote:
There’s no way chatGPT can simulate something like this, it can’t even answer the simplest problems I’ve ever seen

I ask chatGPT a math question and get an incorrect answer. Chances are this graph is wrong too

LeeEuler
basketstorm wrote:

Using ChatGPT powers I simulated 1000000 chess games in a pool of 1000 players. Pairing was rating based with small diffusion to emulate online presence factor. Win/loss factor - just like prescribed by Elo. All players had hidden strength in Elo: 90% of players - from 1000 to 1400, 10% players - from 1400 to 2800. Initial rating was 200, rating floor - 100.

Graphs:

Blue: initial strength distribution.

Green: rating after simulation show that the largest group is minimal-Elo players. Mid-Elo group received artificial bump despite the fact that strength of players was constant during simulation!

Full table with data for each player (names are all fake based on names of real great players and names repeat but that doesn't matter because each player has unique id):
https://pastebin.com/raw/JqGKun3K

Conclusion:

best of the best climbed to the top easily.
Low elo players unfairly end up in a various rating ranges, apparently because of luck, not because of lack of skill. And now you can't blame virtual players for lack of skill. Because game result was dictated by their actual hidden strength.

So in the end we have cases like:

id player_name hidden_strength_Elo final_rating_Elo 176 Magnus Portisch 1097 509 468 Vladimir Svidler 1263 497 571 Sergey Short 1239 1042

That means actual strength could be 1200, but rating could be 500 OR 1000.

Or look at this oppressed guy:

id player_name hidden_strength_Elo final_rating_Elo 467 Boris Nepomniachtchi 1203 355




With strength 1203, his rating is 355.
Each player here played 1000 games!

Some more oppression:

id player_name hidden_strength_Elo final_rating_Elo 266 Hikaru Gajdosko 1112 100 322 Magnus Capablanca 1003 132

Magnus is weaker than Gajdosko but Gajdosko is stuck at 100. Is this fair?

This all aligns with my observations and experience here on chess.com and explains why many people astonished by randomness in apparent strength of their opponents that have same rating.

Thoughts?

Really nice analysis. Could you post the script or link the git so we can dig in further?

LeeEuler
TitanMaster101 wrote:
xtreme2020 wrote:
There’s no way chatGPT can simulate something like this, it can’t even answer the simplest problems I’ve ever seen

I ask chatGPT a math question and get an incorrect answer. Chances are this graph is wrong too

Terry Tao seems quite happy with the progress of AI in the math realm...
https://unlocked.microsoft.com/ai-anthology/terence-tao/

Forsaken

Eh

basketstorm
IndianCamels wrote:
basketstorm wrote:
RandomChessPlayer62 wrote:

Alright then, I'll agree with that, just that though, one small mistake that makes low Elo unplayable, nothing else in your argument unless I am shown irrefutable proof

I appreciate that you're open to some of the points, but I’ve already provided mathematical and logical proofs throughout this discussion that show why the rating system struggles to accurately reflect skill at low Elo. The problems I’ve mentioned, like rating volatility and the lack of differentiation among players in this range, aren't just based on opinion - they’re mathematically verifiable with basic calculations.

Graphs can lie.

For sure graphs can lie. Like some graphs on chess.com, see here: https://www.chess.com/forum/view/community/chess-com-rating-distribution#comment-106230355

But in the post you quoted I was talking about basic logic and arithmetic that prove my point.