connor_mcmonigle wrote: ↑Fri Jul 09, 2021 8:47 pm
I think you must be trolling as no possible formula would satisfy you. Dividing the remaining depth by some N? That won't work as Senpai did something similar first (and probably many others) -> you can't just divide remaining depth. Some type of exponential decay formula? That's already been tried by many historical engines. Some type of sqrt formula? Nope, can't use that either. Would a hardcoded table of values work? No as other engines have already used tables of values for computing a base reduction before (also you'd probably argue that you could fit one of the aforementioned formulas to the values in the table anyways).
and did they test all these other formulas? or they immediately took the easier path copying Stockfish formula changing just a bit the coefficients? you are proving exactly my point.
Lol. Those are all things I tried when I was first starting out... that's where the list came from. I know for a fact that the Koivisto authors have tried many variants even trying to remove the base reduction entirely, solely relying on scaled history values from my communication with them. A log based formula simply ended up working best past a certain point in development for me and many others.
Anyways, the rest of your post is just ridiculously presumptuous, speculative and, therefore, not worth addressing. I don't intend to continue this conversation further as I really don't believe you can possibly be arguing in good faith anymore.
connor_mcmonigle wrote: ↑Fri Jul 09, 2021 8:47 pm
I think you must be trolling as no possible formula would satisfy you. Dividing the remaining depth by some N? That won't work as Senpai did something similar first (and probably many others) -> you can't just divide remaining depth. Some type of exponential decay formula? That's already been tried by many historical engines. Some type of sqrt formula? Nope, can't use that either. Would a hardcoded table of values work? No as other engines have already used tables of values for computing a base reduction before (also you'd probably argue that you could fit one of the aforementioned formulas to the values in the table anyways).
and did they test all these other formulas? or they immediately took the easier path copying Stockfish formula changing just a bit the coefficients? you are proving exactly my point.
Lol. Those are all things I tried when I was first starting out... that's where the list came from. I know for a fact that the Koivisto authors have tried many variants even trying to remove the base reduction entirely, solely relying on scaled history values from my communication with them. A log based formula simply ended up working best past a certain point in development for me and many others.
For years I had my own hand-typed LMR [64] [64] reduction table. Using the SF algorithm resulted in an almost equal table. Using that gave a small improvement after all. So Wilson, put me also on your black list
90% of coding is debugging, the other 10% is writing bugs.
Some people also forget that "stealing" an idea from another program is called "following the best known practices" in other forms of engineering and science. If an improvement is made, -everybody- will implement it, and then people try to improve on that again. People are not going to implement worse solutions with the sole purpose to be different.
Open source is a software development philosophy where you can take idea's and even code from different program, as long as you credit the original authors. Granted, it will lead to a more homogeneous field of chess engines, but there lots of ways to make your engine play differently than other engines.
mvanthoor wrote: ↑Sat Jul 10, 2021 6:13 pm
Some people also forget that "stealing" an idea from another program is called "following the best known practices" ....etc
As they have always said in NASCAR: "If you ain't cheatin', you ain't tryin'"
Cornfed wrote: ↑Sat Jul 10, 2021 6:39 pm
As they have always said in NASCAR: "If you ain't cheatin', you ain't tryin'"
If you take ideas and/or code from other programs without crediting them and their authors, and then go and release your own closed source software so nobody can check this, _then_ you're cheating. If you're taking an idea and/or code from another program and credit them, you're not cheating. The other program could very well have credits pointing to the source THEY got their idea from (hopefully, if it was not their idea).
Rebel wrote: ↑Sat Jul 10, 2021 4:31 pm
For years I had my own hand-typed LMR [64] [64] reduction table. Using the SF algorithm resulted in an almost equal table. Using that gave a small improvement after all. So Wilson, put me also on your black list
Schröder after you called me a stalker I have done some research on you and exactly 10 years ago you accused Doc Hyatt of having copied stuff from Robolito. Hyatt opened a thread on Open-Chess Still waiting on Ed calling you out and asking for the proofs. I suggest everyone to read that funny thread, it's only 8 pages long.
your only proofs were the big jump in strength and the lower branching factor
nothing was copied of course
So now I find quite amusing you are mocking me
have you bought Hyatt a new bullshit detector as he requested or apologized at least ?
mvanthoor wrote: ↑Fri Jul 09, 2021 4:06 am
The one thing that was invaluable to me while getting up to speed with chess programming was all the historical information in either this forum, Google Groups, or on the internet. Public information about chess programming seems to have started at around +/- 1995. Before that, everything was private between the better chess programmers and teams, and the top engines seem to be going back to that state of affairs. That is what I dislike most. Information that is exchanged which then only exists in the head of the participants of the exchange; it is basically lost for any programmer that comes after.
Unfortunately have to agree with this. The cutting edge of using neural networks in search heuristics such as pruning, and in time management is in the Leela discord, while hardly anybody on Talkchess seems to be aware of neural networks as anything other than a supplement or replacement for their hce.
Madeleine Birchfield wrote: ↑Sat Jul 10, 2021 6:54 pm
Unfortunately have to agree with this. The cutting edge of using neural networks in search heuristics such as pruning, and in time management is in the Leela discord, while hardly anybody on Talkchess seems to be aware of neural networks as anything other than a supplement or replacement for their hce.
It's like a group of scientists, exchanging ideas and best practices, eventually doing awesome things which turns the world in their field upside down, and then they all die in a car crash while driving to get a hamburger... and everybody else is asking themselves: "We wonder how they did it." That's just a shame.
Madeleine Birchfield wrote: ↑Sat Jul 10, 2021 6:54 pm
Unfortunately have to agree with this. The cutting edge of using neural networks in search heuristics such as pruning, and in time management is in the Leela discord, while hardly anybody on Talkchess seems to be aware of neural networks as anything other than a supplement or replacement for their hce.
It's like a group of scientists, exchanging ideas and best practices, eventually doing awesome things which turns the world in their field upside down, and then they all die in a car crash while driving to get a hamburger... and everybody else is asking themselves: "We wonder how they did it." That's just a shame.
It's unfortunate that the development of top engines isn't more visible on public forums such as TalkChess. However, if anyone is curious as to why this is, they need look no further than this thread. This wasn't always the case, but TalkChess as of today is proliferated with trolls and low quality posts which just kill the signal to noise ratio.
Rebel wrote: ↑Sat Jul 10, 2021 4:31 pm
For years I had my own hand-typed LMR [64] [64] reduction table. Using the SF algorithm resulted in an almost equal table. Using that gave a small improvement after all. So Wilson, put me also on your black list
Schröder after you called me a stalker I have done some research on you and exactly 10 years ago you accused Doc Hyatt of having copied stuff from Robolito. Hyatt opened a thread on Open-Chess Still waiting on Ed calling you out and asking for the proofs. I suggest everyone to read that funny thread, it's only 8 pages long.
your only proofs were the big jump in strength and the lower branching factor
nothing was copied of course
So now I find quite amusing you are mocking me
have you bought Hyatt a new bullshit detector as he requested or apologized at least ?
Anthony, what happened to your sense of humor? It was a lightly joke, not mocking you.