ENGINE SPEED AND STRONG

[Joost Buijs]

The program temporarily uses your 4x4 evaluation data and not your evaluation function as a whole, it is my intention to have this data replaced pretty soon because I'm going to switch to larger patterns anyway. Maybe this was not clear to the average reader, well now it is.

Now I have to rethink whether I will continue with the project or not, for me the fun is programming and I'm not looking after recognition of any kind, which seems very important for you. I already had my doubts about putting so much effort into a game that appears to be a 'black hole'. Maybe all that effort is a waste and I rather could have spend it on my chess program that I didn't touch for four years.

I think the draughts community is so small and everyone is in it for the fun.

But just consider Fabien's position: he already had one piece of labor stolen from him (Fruit). And now you carelessly use and distribute his essential draughts contribution (Scan eval) without proper attribution. Can you imagine how that might spoil his fun?

Suppose you enter a draughts tournament, what would you propose as a procedure to convince others of the fact that you are entering with an original program?

Well, I think the simplest way is to have the source code examined and to verify that the executable is build with that source, not something stupid like the 'similarity comparison' that CSVN does.

I admit that it was a mistake to give away an unfinished program which still contains data that is the property of Fabien, it was never my intention to release it but you know how it goes, people are asking, and I never thought about the fact that Fabien could have a problem with this.

I don't have any problem with giving proper attribution, Fabien could have contacted me and I would have added it. Every programmer understands that I don't have ML running yet and that the weights are based on scans weights, maybe I should have stated this more clearly.

What should I do? Adding a note that the evaluation weights are based on scan weights? I never dealt with GPL before so I don't have a clue how that works.

Pretty soon the whole problem will be over, I already generated 2 million games from random starting positions and I hope to have the weights recalculated within a few weeks, problem gone.

Joost

[Ed Gilbert]

We can easily check that it's no secret. Let's ask Krzysztof, Catherine, and Sidiki (not to mention others) whether it was 100% clear that Argus was actually using Scan's evaluation function (and not just the idea of 4x4 patterns which would have been OK).

I had assumed that Joost generated these values because I did not read anywhere here on the forum that they were taken from Scan.

-- Ed

[Joost Buijs]

We can easily check that it's no secret. Let's ask Krzysztof, Catherine, and Sidiki (not to mention others) whether it was 100% clear that Argus was actually using Scan's evaluation function (and not just the idea of 4x4 patterns which would have been OK).

I had assumed that Joost generated these values because I did not read anywhere here on the forum that they were taken from Scan.

-- Ed

It was my wrong assumption that every programmer understood that these values were derived from the scan values since I don't have ML up and running yet, how could I have calculated them otherwise? I told Bert, but maybe I should have been more clear about this. I didn't think this to be very important because it was my intention to replace the values within a few weeks anyway.

Now that I see how much hassle this gives I'm thinking about abandoning the whole project and start working on something else. Time wasted and an experience richer, not to say that I learned several things during the course of this project about search enhancements that I can probably use in my chess program.

Joost

[BertTuyt]

So herewith my 5 cents.

For me it was 100% clear that Joost used the data from Fabien.
And I also think that he implicitly mentioned that in several posts.
Maybe the phrasing was not fully transparent, but I'm sure his intentions are pure.
On Monday 31-October-2016 he wrote, "out of curiosity I've added the Scan20 evaluation-function".
And in a later post Tuesday 1-November-2016 "at first I will use the same patterns as Scan".

So maybe the use of the Fabien data-file should be made more explicit, so I would not consider this a major crime.
Joost shares a lot of ideas, especially related to search speed improvement opportunities.
He also mentioned several times, that he didn't start with the Machine Learning part.
If he would have done this, I'm also 100% sure he would have shared all details.
And only while some people insisted he shared his engine (containing the evaluation data from Fabien).

Our community is not that big, only a few members are really active, and certainly there is no commercial gain possible (which is different for Chess).
Next to that I feel that there is great respect between programmers, and we don't have an attitude to steal and rob, on the contrary, the community is more about sharing.
Fabien learned us new ways to evaluate.
Joost teaches us a lot, that there is far deeper search possible, when we embrace new Intel Instructions-sets like BMI.
Ed shared his 8P DB, including the source of the driver.
And myself Im working on a Tournament Manager inside the Damage GUI based on rubbish Hub-ish..

So lets not blow this outside normal proportions.

I live in a small village (600 people) in Switzerland.
On a mountain at 1000m height.
The road stops a few kms from our apartment.

People outside our village call it a dead end, we call it the beginning of the world.

As long as we all believe there are 200-300 ELO points to win, I think our journey is far from over.
And when you want to climb the Mount Everest, you can only do that when you team up together, and built upon each team member strengths ........

Bert

[Krzysztof Grzelak]

Please do not swing in the clouds gentlemen programmers. A good programmer is one that can write engine under each processor and other features are important contemporary processors.

[BertTuyt]

Krzysztof,

within the context of the discussion, wise words are welcomed.

Bert

[Krzysztof Grzelak]

Krzysztof,

within the context of the discussion, wise words are welcomed.

Bert

Too many attribute to the strength and power of the processor and it is not so. Let me give you an example. It is a tournament game Damage - Scan. Damage game on hardware:

processor - PHI - 72 cores
ram - 256 GB

Scan game hardware:

processor - Intel Core I7 - 16 cores
ram - 16 Gb.

Understanding some of programmers, the Damage should always win the match with the program Scan because it has a better processor and the majority of the number of ram. And unfortunately it is not.

[BertTuyt]

Krzysztof,

you are right, unfortunately it is not.

Bert

[Krzysztof Grzelak]

Krzysztof,

you are right, unfortunately it is not.

Bert

As you can see processor (even this modern) and a lot of ram is not enough.

[BertTuyt]

Krzysztof,

also here you are right, and your remarks are spot on.
I also can see that processor (even this modern) and a lot of ram is not enough.

Bert

[Joost Buijs]

Krzysztof,

within the context of the discussion, wise words are welcomed.

Bert

Too many attribute to the strength and power of the processor and it is not so. Let me give you an example. It is a tournament game Damage - Scan. Damage game on hardware:

processor - PHI - 72 cores
ram - 256 GB

Scan game hardware:

processor - Intel Core I7 - 16 cores
ram - 16 Gb.

Understanding some of programmers, the Damage should always win the match with the program Scan because it has a better processor and the majority of the number of ram. And unfortunately it is not.

Do you really think that the PHI is better suited to run a draughts program on?

The PHI consists of many very slow Pentium like cores and is suitable for scientific calculations. but it is not so great to run a draughts program which is a serial algorithm in nature.

And draughts seems to be very insensitive to calculation depth (unfortunately), yesterday I ran a match between two versions of my program, one being on hardware 20 times faster as the other, the fast version won the match but not by a very great margin, not what you would normally expect with such a great hardware difference when using other games.

[Fabien Letouzey]

And draughts seems to be very insensitive to calculation depth (unfortunately), yesterday I ran a match between two versions of my program, one being on hardware 20 times faster as the other, the fast version won the match but not by a very great margin, not what you would normally expect with such a great hardware difference when using other games.

Stop calling that ... thing ... "my program".

[Joost Buijs]

And draughts seems to be very insensitive to calculation depth (unfortunately), yesterday I ran a match between two versions of my program, one being on hardware 20 times faster as the other, the fast version won the match but not by a very great margin, not what you would normally expect with such a great hardware difference when using other games.

Stop calling that ... thing ... "my program".

Really I don't know what you mean, every line of code is original and written by me. I realize now that I made a mistake to use the 'precious' data from your ... thing ... in the test version because I haven't generated the weights myself yet.

I want to apologize for the fact I gave the program to a few others without mentioning explicitly that it currently uses evaluation weights derived from Scan, and if you don't want to accept my apologies so be it, no hard feelings.

Joost

[Fabien Letouzey]

Really I don't know what you mean, every line of code is original and written by me.

Code is slowly becoming obsolete. This is the ML era; data is everything. Replace my data with zeroes (AAAA with your encoding), and you can claim authorship. Not before.

[Joost Buijs]

And draughts seems to be very insensitive to calculation depth (unfortunately), yesterday I ran a match between two versions of my program, one being on hardware 20 times faster as the other, the fast version won the match but not by a very great margin, not what you would normally expect with such a great hardware difference when using other games.

Stop calling that ... thing ... "my program".

Really, I don't know what you mean, every line of code is original and written by me. I realize now that I made a mistake to use the 'precious' data from your ... thing ... in the test version because I haven't generated the weights myself yet.

I want to apologize for the fact I gave the program to a few others without mentioning explicitly that it currently uses evaluation weights derived from Scan, and if you don't want to accept my apologies so be it, no hard feelings.

Joost