Checking if Elo system is oppressive [With proofs]

No, simulation wasn't that stupid, I've checked the code. ChatGPT didn't calculate anything, it made a program to run the simulation. Outcome was simulated for each game using probabilities according to hidden strength of players. You trust in rated Elo too much. Just think about the initial situation: all players are rated the same when they sign up. How accurate is that real Elo then? How can you grow or decrease Elo of a player if he competes against players with innacurate Elo? That's the problem.

If only all Elos were correct in the first place. But if you are rated 600 and competing against freshly-joined 200 who is stronger than you actually and you lose to him and you lose a good amount of Elo because of that because the system treats his 200 as "real", that's not even close to a quantification of someone's skill. That's just chaos.

Beating 50% of the people at your skill is ideal situation on paper but it doesn't happen, you rarely encounter such people. To encounter such people you need accurate Elos for: 1) you 2) opponent. But because all fresh accounts get exactly the same artificially low Elo (200 or 400 in Rapid) there's no way to do accurate pairing, pairing by actual skill. Situation will not improve even after 1000 games because real skill of players fluctuates that much in low-Elo zone. You need to be a really strong player to break through.

Also because there's a limit (100 Elo), and someone grows after beating you while you hit the limit, that's pure inflation. Sum of all ratings in the system becoming greater.

That's not exactly how Glicko works. It's not about being new or old player, the system tracks your inactivity and increases deviation. But while inactive player is penalized for the loss more, you don't gain less than from beating an active player.

Elo on average is accurate especially for skilled players. But not in low-Elo zone where actual skill is random and does not correspond to the rating well.

You don't need very very new. People can keep that skill-rating mismatch for years, I've seen it, I've met such players.

First of all, we need to take into account that this doesn't take real life in. Second, you are basically reinventing the skill range by making everyone the same rating. It's not possible for one to become 2800 if the starting rating is 1000, unless you're using crazy scales. Thus, your experiment has a bunch of logical errors. Calling the rating system oppressive would also be false due to large amounts of proof that rating systems heavily over-reward overperformance, and do not heavily hurt people for underperformance. Don't use chatGPT for math unless you want bad answers. Instead, get a simulation of real people.

There are more reasons why this is inaccurate. If we were using elo, the higher the rating of the player, the less players of his rating, the less points he gets per game. This is why it is so hard for the top 10 to hit 2900. Simply put, they would have to win many chess games in a row.

So you finally admit there's crazy rating mismatch? But why do you think that this unfair zone ever ends somewhere except for the very tops? Simulation data suggests that there's no linearity in actual strength vs rating difference even for >2000 Elo.

As for my games, let's not talk about my games. This is a global problem, not mine. All my points stand strong, your denial based on beliefs, my claims are based on math, realistic simulation data and backed up with real experience of players (see the frequent "game is rigged" topics, there's no wave without wind). And it's enough to have just common logic to understand that such rating system where everyone starts at 200, have floor at 100 and rating changes applied like if every rating is real - can't be fair.

@IndianCamels

I didn't make everyone the same rating. In my simulation no one became 2800 from 1000.

I didn't use it for math. I provided the math. I proof-read his simulation project. All is good. You don't even need a simulation, just a paper and pencil, try thinking about Elo rating system, make some calculations and you'll quickly realize that it's not going to work fairly. It has to be oppressive, there's no other way.

Knowing all the formulas it is trivial to simulate rating flow of any player base very realistically with various parameters and see where it fails. That's not rocket science.

That's not how it works. You don't get less points per game just because there are not enough players in your rating range.

Look up the calculation for elo and then tell me.

Elo uses normal probability to determine how much elo you get. Ex: You have 85% chance of beating a player 400 points less than you