Impact ELO developer

[Henk]

I also see a Vincent Diepeveen with ELO 2327.

https://ratings.fide.com/card.phtml?event=1003054

[mar]

Yes of course, how could I forget Vincent! There can be more.

[Joost Buijs]

In the past (35 years ago) when I was still young I had something like 2050 Dutch Elo which was something like 2150 Fide.
Dutch Elo was always lower than Fide Elo, I don't know how it compares nowadays.
At the moment I guess my rating is about 1850 Dutch Elo, it is about 10 years ago I played in the Dutch league and they removed me from the Elo list.
Maybe time to start playing again. LOL

[Henk]

Perhaps related: My chess.com blitz rating is 100 points higher than my Dutch KNSB rating for normal games. If they would be comparable it would be best not to think too long in normal games I guess.

[Uri Blass]

Well, speaking of chess knowlege, search requires exactly zero chess knowledge.
Good eval can be say several hundred elo (guesstimate), the rest is search.
You only need to know how pieces move to write a move generator.
So assuming someone who is above -inf elo should be able to write a relatively strong engine.
Add very basic chess knowledge + tuning (eval) and you can have a strong engine (relatively strong of course) and you really don't have to be a GM at all.

Conclusion: I see no strong correlation. For example, rook on semi-open file didn't add any elo in my tests, but it's considered good.
Of course there are strong players who are also engine authors, I think Miguel is pretty strong (don't remember his elo but you can ask him

I disagree that search require exactly zero chess knowledge.

Good chess knowledge can be productive for better search rules.

I believe that search that is not based on special chess knowledge+ evaluation that is based on very basic knowledge can be enough to get computer rating above 2800
but I believe that it is possible to get significantly better after it both by better evaluation and better search that is based on chess knowledge.

My fide rating is 2019

[Henk]

Well, speaking of chess knowlege, search requires exactly zero chess knowledge.
Good eval can be say several hundred elo (guesstimate), the rest is search.
You only need to know how pieces move to write a move generator.
So assuming someone who is above -inf elo should be able to write a relatively strong engine.
Add very basic chess knowledge + tuning (eval) and you can have a strong engine (relatively strong of course) and you really don't have to be a GM at all.

Conclusion: I see no strong correlation. For example, rook on semi-open file didn't add any elo in my tests, but it's considered good.
Of course there are strong players who are also engine authors, I think Miguel is pretty strong (don't remember his elo but you can ask him

I disagree that search require exactly zero chess knowledge.

Good chess knowledge can be productive for better search rules.

I believe that search that is not based on special chess knowledge+ evaluation that is based on very basic knowledge can be enough to get computer rating above 2800
but I believe that it is possible to get significantly better after it both by better evaluation and better search that is based on chess knowledge.

My fide rating is 2019

Name and ELO rating of your chess engine ?

[Uri Blass]

Well, speaking of chess knowlege, search requires exactly zero chess knowledge.
Good eval can be say several hundred elo (guesstimate), the rest is search.
You only need to know how pieces move to write a move generator.
So assuming someone who is above -inf elo should be able to write a relatively strong engine.
Add very basic chess knowledge + tuning (eval) and you can have a strong engine (relatively strong of course) and you really don't have to be a GM at all.

Conclusion: I see no strong correlation. For example, rook on semi-open file didn't add any elo in my tests, but it's considered good.
Of course there are strong players who are also engine authors, I think Miguel is pretty strong (don't remember his elo but you can ask him

I disagree that search require exactly zero chess knowledge.

Good chess knowledge can be productive for better search rules.

I believe that search that is not based on special chess knowledge+ evaluation that is based on very basic knowledge can be enough to get computer rating above 2800
but I believe that it is possible to get significantly better after it both by better evaluation and better search that is based on chess knowledge.

My fide rating is 2019

Name and ELO rating of your chess engine ?

I quit chess programming some years ago.

Name of my engine is movei.

CCRl rating of it at 40/40 is 2667

[elcabesa]

I'm one example of mediocre chess player with a good chess engine, my Elo rating is 1500.

[Henk]

Dirty evolution is sugestive:

Pradu started alone (Good programmer , bad player ) -> 2200
Andres joined the team (Bad programmer, medium player) -> 2400
Fonzy joined the team (Good programmer, good player) -> 2800

Nice to hear from you
That sort of undermines my theory.
You should include Lars (good tester, grandmaster) -> 3400

Maybe ELO tester is more important.

[Rochester]

The good developer write the engine. The bad developer write the talkchess.