Reckless 0.10.0 dev ...

[Frank Quisinsky]

Hi there,

the engine is really fantastic.
A lot of short wins, perfect move-average ... have many fun here with Rexchess 0.9.0!

Now a new dev is available:
https://github.com/codedeliveryservice/Reckless/tags

Possible that this version is stronger as Stockfish 18 or the last dev of Stockfish?
As always, I have no idea.

Best
Frank

[Frank Quisinsky]

RexChess?
by Don Daily ... no, no
37 Years ago, with a super fast 286er UNI system.
OK, I had also many fun with RexChess and his timeref

No, no ... Reckless is the topic not RexChess.
Sorry!

[Jouni]

No one can beat Stockfish with 155 contributors and occasional supercomputer access. Interesting change MAX_PLY increased to 240 from 128.

[Frank Quisinsky]

Hi Jouni,

I am not sure!

You know, in times today are not the late mid-game and transposition into endgame most interesting. Often I am thinking Stockfish is playing near the perfection here. If we can speculate that an engine indeed are able to beat Stockfish it must be a very strong attacker in the first playing phase after opening book moves. When I look at the statistics, I don't see much in the way of real improvements in Stockfish in this area (here is a comparison of the developments ... but of course not sure at all). Furthermore, it should hold the strong late mid-games and transposition into endgames. Reckless have after statistics more the aggression in the early stages of the game and can hold the playing strenght in late mid-games / transposition into endgames.

Perhaps its much better if now a smaller group of persons try to optimate Reckless.
Many persons can be counterproductive if ... if the path to further improvements may be blocked by the many people involved.

Thinking on Arena times.
We had 250 beta testers and the forum full of people with a lot of very nice ideas. I tried to optimize everything using lists of suggestions, which ended up costing me a lot of time during my own testing. With the final results: We were all getting in our own way when it came to making Arena even better.

And to the topic Super-Computer:
Strong computers are strong enough I think.

Stockfish can be beat in my opinion only by a very small group of very strong developers.

On the other hand:
What be found for Reckless can be use for Stockfish ...
A lively exchange of ideas.

Perhaps in 3 years we have 100, better 200 or more available engines, all are in the near (+-10) to the number 1.
In this case the question for humans is ...
Which of all these programs play the style the most of humans like.
And most important here: The earlier mid-games, many pieces on board, then a fireworks display!!

My personal opinion about it.
And this since more as 30 years.
I am searching allways the attackers in my rating systems I created and wrote a lot to playing styles for chess newspapers, my own site and so on. My opinion about it never changed, all the years!

The problem is:
We can see the details from all the TOP-Engines only with very strong statistic programs.
Sure, a human, GM or not can't see that in detail. All the TOP-Engines are clearly to strong.

Best
Frank

[Rebel]

Reckless is closing in, Reckless 0.90 scored 48% against SF 18, current run : 49.4%

But more than 2000 games need to be played.

[Frank Quisinsky]

yes, in this case absolutely.
And very important ...
vs. a lot of opponents ...
if the different is for the moment +1-3 Elo more or less.

With a lot of games much more strong stats are possible!

I think in this case 25 opponents min., around 100-150 Elo weaker as both engines, with the quantity of lesser draws the result will
be much better.

[Ipmanchess]

Hi Frank..maybe you have seen this list: https://ipmanchess.yolasite.com/r9-7945hx.php

And Reckless 0.10.0 avx512 is in testing..

[Frank Quisinsky]

Hi Ipmann,

your site is so huge.

I can't say this is the site of our number one tester. The others are doing a good job too. But in reality, it is the site of our best computer chess tester for years. There's no doubt about that.

You know, the testers don’t really get along anymore. All that hidden jealousy—it’s always been that way. I’m making myself unpopular again; I really should just shut up.

Most important is, that not all do the same and own ideas will go in such a work. The ideas with playing styles is really good. I like the work Stefan Pohl create also, what Ed do is very very nice.

But after all ...
For me your produced results are most interesting.

Best
Frank

[Frank Quisinsky]

But what I've written here is just between us—the others aren't supposed to read it.

Do you know what is my personal favorite for a number 1 place in the near feature?
It is pawnocchio for different reasons, most of time I am looking on pawnocchio games on my ultrabook.
I can't back that up with statistics; it's just a feeling.

Reckless is great, it's my second favorite from the group of TOP Generals.
In the feature I see pawnocchio on a very high place.

But from all the available engines my opinion is quit clear:
Wasp, Patricia, Velvet and Texel are my personal favorits.
Tcheran and Zangdar ... will go its own way very fast.

Not all I know, to much new engines for me are available in the last years.

[Frank Quisinsky]

49,4% ... and now?

Ed, I know our bad discussion here ...
Forget if for a moment, how much is the different now?