Checking if Elo system is oppressive [With proofs]

Sort:
basketstorm
magipi wrote:
basketstorm 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

ChatGPT is evolving. Free version can't do much, sure, but most advanced runs analysis, you need to wait for a while and it returns with graphs and tables

What's the chance that those graphs and tables are just completely made up random nonsense?

ChatGPT is a language model, not a statistical tool.

I've already explained that in this topic. Most advanced ChatGPT version writes the code (and you are free to examine it), runs it on actual cloud machine and copies result of real program execution (tables, graphs) into the chat with the user. So that data is not generated by language model. Older versions of ChatGPT used to rely on language model only, true. But it evolved.
P.S. You can also copy the code into your computer, ChatGPT would explain how to run it, what to install, will make adjustments to the code for you if you want to change something etc.

Jindagara

Using statistics and models is unfair in chess. Probability doesn't play out many times.

Jindagara

Big words

Jindagara

Don't make it correct

Jindagara

Chatgpt runs through text... Maybe try Wolfram alpha?

basketstorm

@Jindagara please combine your messages into a single post. Read my posts above, my usage of ChatGPT wasn't about pulling some made up data from the language model. Actual calculations happened in a real program.

xtreme2020
No matter how well you prove it, this guy won’t admit he’s wrong, he just thinks ChatGPT is infallible and perfect
xtreme2020
It is pretty cool
basketstorm

@llama_l

First mistake is to assume that everyone's rating is already somehow magically "established". We can't start from that assumption.

In general it seems like you've tried to solve non-existing problem. Here on chess.com we don't start at 1500. We start at 200 or 400, but in my example (Blitz), I'm testing 200 as starting rating and 100 as floor. Did you have floor by the way?

Also here when paired, rating difference is usually much closer than 200

xtreme2020
#115 *sigh* it doesn’t matter what the actual numbers are, the results are the same. He didn’t assume that either, if I read it right he assumed everyone had a standard rd, which would favor your point even more
basketstorm

That was the whole point of simulation. To start from scratch and see if the rating system works without having to blindly trust that it works or worked at some point. But if you build your simulation on an assumption that system works in order to prove that system works that doesn't make any sense, that's not even a test.

xtreme2020
#117 so go put a bunch of people starting at 200 with a random amount of actual skill in a real simulation, not ChatGPT
basketstorm
xtreme2020 wrote:
#115 *sigh* it doesn’t matter what the actual numbers are, the results are the same. He didn’t assume that either, if I read it right he assumed everyone had a standard rd, which would favor your point even more

It does matter as many players of lower skill range hit the rating floor many times, causing inflation for others and high rating-skill difference for many.

basketstorm
xtreme2020 wrote:
#117 so go put a bunch of people starting at 200 with a random amount of actual skill in a real simulation, not ChatGPT

You keep talking about ChatGPT as if it was just a language model. Update your knowledge about it. Simulation was real (and that " put a bunch of people starting at 200 with a random amount of actual skill in a real simulatio" is exactly what I did)

xtreme2020

edited comment Andrewsmith 

basketstorm
xtreme2020 wrote:
You know what I’m not even gonna try anymore this guy is just an idiot

That's not a proper way to discuss any topics. Reported.

xtreme2020
You do that
IndianCamels
basketstorm wrote:

@llama_l

First mistake is to assume that everyone's rating is already somehow magically "established". We can't start from that assumption.

In general it seems like you've tried to solve non-existing problem. Here on chess.com we don't start at 1500. We start at 200 or 400, but in my example (Blitz), I'm testing 200 as starting rating and 100 as floor. Did you have floor by the way?

Also here when paired, rating difference is usually much closer than 200

Your input info is correct, but you can't use chatGPT to run it. ChatGPT gets better and better. Also, why would you say elo is oppressive. Are you saying that you and a 2000 on chess.com are the same skill, but the elo system is working against you, making sure you don't succeed, and getting good is nothing but a joke and all Grandmasters are lazy?

xtreme2020
#122 If you want to try to prove you have some hidden skill and you’re better than 400 but the elo system is the one to blame, play me any time.
IndianCamels

Also, if you did an actual experiment, you would get a bell curve in your histogram. If you look at the actual placement of chess.com ratings, you would realize they are just a skewed right bell curve. Describing the data accurately doesn't matter if the data is wrong. Compare your rating distribution to an actual rating distribution(go to rapid stats, global), mean rating is 620, and you'll see your model is terribly innaccurate/