Why is Stockfish removing killer moves in move ordering ?

[mar]

I wonder since by removing check extensions the engine will be severely crippled tactically

source: I made it up

experience - FTFY
(in fact I've made a change recently that goes against elo but fixes some tactical issues)

of course you need to pick the right measure in this case, which is not elo but a set of sufficiently difficult tactical test positions (hence tactics)

then pick 2 engines, say A and B where both are the same, except that A has check extensions removed and B enabled where the measure is lower time to solution
I put my money on B

of course, if you live in a world where the only measure of happiness you know is wealth, then all you'll do is accumulate wealth

[Viz]

Pick test suites - show improvement on them - develop engine like this - show that it's better at literally anything but solving test suites you picked.
Last of this for whatever reason never happens with this type of "development".
And the opposite happens for non-idiotic type of development, aka stronger engine on average starts getting better and better with testsuites on average (even if one exact version may have temporarely drop).

[Viz]

https://github.com/vondele/matetrack?ta ... ectiveness
if you want the "best analysis engine" or w/e garbage might as well monitor matetrack results, why the heck no.

[mosfel24]

I wonder since by removing check extensions the engine will be severely crippled tactically

source: I made it up

experience - FTFY
(in fact I've made a change recently that goes against elo but fixes some tactical issues)

of course you need to pick the right measure in this case, which is not elo but a set of sufficiently difficult tactical test positions (hence tactics)

then pick 2 engines, say A and B where both are the same, except that A has check extensions removed and B enabled where the measure is lower time to solution
I put my money on B

of course, if you live in a world where the only measure of happiness you know is wealth, then all you'll do is accumulate wealth

I see you have a similar testing methodology as me. Like you, I do not understand this army of blind elo slaves, and I agree with you that tactical ability is as important if not much more so. I just want to ask how many test positions do you use, is it more than 10 or less than 10?

[mar]

nothing you wrote is based in any fact or science, where as the removal of the concept is. So I defer to science.

removing random parts of the engine to see what happens is hardly science (yet it may work)

[mar]

I see you have a similar testing methodology as me. Like you, I do not understand this army of blind elo slaves, and I agree with you that tactical ability is as important if not much more so. I just want to ask how many test positions do you use, is it more than 10 or less than 10?

I never claimed that. I test patches based on self-play games, not testpositions. I wouldn't get very far otherwise.
except that I'm willing to sacrifice some elo for accuracy now and then - after all, I do this for fun

A0 has shown us that tactics in chess is much less important than we originally thought, I still hate if my engine is completely clueless in some positions or loses a game due to a tactical blunder

I have a set of positions (12) that I want my engine to solve in reasonable time (let's say seconds rather than minutes, for example)

[Viz]

If you start talking about "A0 have shown"

Moderation

Insult removed

[mar]

If you start talking about "A0 have shown" you can't call yourself any sort of a serious chess engine dev, at best a clown, although it may be offensive... towards clowns.

too much vodka?

[emadsen]

If you start talking about "A0 have shown" you can't call yourself any sort of a serious chess engine dev, at best a clown, although it may be offensive... towards clowns.

Chill out. "You can't call... at best a clown." People pursue this hobby for their own reasons, not to placate you. Who are you to dictate to someone else their motivations and goals? Good God, the level of animosity and megalomania on this forum at times. Try instead to be helpful.

[shawn]

removing random parts of the engine to see what happens is hardly science (yet it may work)

I would not say this is the result of "removing random parts of the engine". The change is rather justified...