Which are the chess engines written from scratch ?

[S.Taylor]

Hello there,

I plan to run a test based only on scratch engines...

So, which are those chess engines written especially from scratch ?
And how we can be sure that they are really started from scratch ?

Btw,is that true that the bellow ones are scratch engines?
Amateur 2.82
Anatoli 0.35k
Awesome 1.73
Murka 3
Smash 1.03
Tornado 4.88

Source:
http://chessprogramming.wikispaces.com/ ... ew/scratch


Thanks in advance,
Sedat

I didn't even know this was possible, unless the program consciously sidesteps every idea from all other programs, especially from recent material used in "clones" and those which use that knowledge.

Isn't it just knowledge, just like there was knowledge in much earlier engines too?

[Evert]

Of course you can consider Junior as the best engine of the world, he has won the world championship of the ICGA.

But let me think to my that call World Championship a tournament where the best engines do not play is a bad name for the tournament.

What's that got to do with anything?
Mind you - if a tournament is called "the World Championship" and a player decides not to compete in it (for whatever reason), he (she) obviously can't call him (or her) self "the World Champion", meaning "the winner of the World Championship tournament." They might still be the best in the world, of course. These are two separate things.

I have no problem with Sedat organize a tournament between programmers whose engines started from scratch, but I think it's a bad idea to make the tournament specifying that are engines whose code started from scratch as it seems implies that other engines did not.

Doesn't seem like it to me, just that the requirement for entering are "started from scratch", which is by itself a vague term that apparently means something different to each person.

There's a difference between "a free engine tournament", being a tournament between free engines (with no claim of completeness) and "a tournament between all free engines", which indeed does imply that all others are not "free". Replace "free" by whatever adjective you want.

Still, if you find it's confusing (there's really no objective statement here either way, so fair enough), by all means argue that case. Just don't use bullshit arguments like "engine authors who didn't invent their algorithms themselves did not write their engine from scratch".

This is my opinion, it seems that some people have a problem that others have a different opinion.

Don't know what you're on about. Just saying that I think your definition of "from scratch" is bogus.

[Sedat Canbaz]

Hi Adam,

Thanks for your interest. Here I replaced the file with GNU GPL ver2 and Smash source code added.

Download Deep Smash x64

Dear Arsha,

Your latest Smash engine did not work on my system (QX9650,Win XP x64 Pro)

[Sedat Canbaz]

Dear Friends,

My new current project has no any goal about proving which engine is original, which is clone and so on...

With the current test,
Just I want to test and see those engines, which have more own playing styles...

One thing more,
Just I'd like to make it more clear about in case of 50 % + similarity
Then I will allow testing that X engine, which is scored with higher winning percentage


Once more I'd like to mention,
PLEASE dont take too seriously the current my new testing
It's a just a fun testing (it's not official...)

I am really sorry if I have offended anyone

Have a nice weekend,
Sedat Canbaz

Note that written from scratch is not the same as having a different playing style.

If you want to test difference then you can simply use the similiarity tool and decide not to accept 2 engines with more than 55% similiarity.

You can consider an engine to be different enough if it has less than 55% similiarity to every previous engine from different authors and get a competition between engines with different playing style(note that I do not think that above 55% means not written from scratch and I also do not think that below 55% means written from scratch).

Just my 2 cents over this issue,
I expect (if they are really written from scratch and are based mainly on own ideas) the scratch engines to have more own playing styles, than the ones which are started written directly from free open source engines

For example, currently I am running another testing (based on top engines, where some of them are derivatives, even some people say for them clones)
http://www.sedatcanbaz.com/chess/?page_id=515
And I noticed that some of them have almost same own playing styles, even the draw percentage is very very high...

About your comment regarding similarity tool,
No, firstly I want to test those scratch engines...
Let's see and compare the Elo performance and then its a good idea to check the similarity of the participated scratch engines

So, I will follow Adam's recommendation about the limit should be 60 % +


Best,
Sedat

[Sedat Canbaz]

If you consider this stupid reasoning then I have a question for you:

Your consider that the list of 20 engines to be used Sedat is a good representation of engine written from scratch?

I could give the name of 200 engines that could probably get on that list to the same conditions that the engines are on that list.

Of course, Sedat is free to organize the tournament he want and choose the desired engines. But giving the name to the tournament as a tournament between engine written from scratch is not a good name because it seems to imply that the 20 engines are the only written from scratch.

Currently the number of the scratch engines is approx 40 (not 20)
And if there are more scratch engines, please let me know, I can promise to include them too...
And after playing some amount of games, I will pick the top 20 strongest ones

Actually I wish to test all available engines, but unfortunately I have no such free time...

No, the name of the current competition will be SCCT Scratch Rating

Best,
Sedat

[Uri Blass]

Dear Friends,

My new current project has no any goal about proving which engine is original, which is clone and so on...

With the current test,
Just I want to test and see those engines, which have more own playing styles...

One thing more,
Just I'd like to make it more clear about in case of 50 % + similarity
Then I will allow testing that X engine, which is scored with higher winning percentage


Once more I'd like to mention,
PLEASE dont take too seriously the current my new testing
It's a just a fun testing (it's not official...)

I am really sorry if I have offended anyone

Have a nice weekend,
Sedat Canbaz

Note that written from scratch is not the same as having a different playing style.

If you want to test difference then you can simply use the similiarity tool and decide not to accept 2 engines with more than 55% similiarity.

You can consider an engine to be different enough if it has less than 55% similiarity to every previous engine from different authors and get a competition between engines with different playing style(note that I do not think that above 55% means not written from scratch and I also do not think that below 55% means written from scratch).

Just my 2 cents over this issue,
I expect (if they are really written from scratch and are based mainly on own ideas) the scratch engines to have more own playing styles, than the ones which are started written directly from free open source engines

For example, currently I am running another testing (based on top engines, where some of them are derivatives, even some people say for them clones)
http://www.sedatcanbaz.com/chess/?page_id=515
And I noticed that some of them have almost same own playing styles, even the draw percentage is very very high...

About your comment regarding similarity tool,
No, firstly I want to test those scratch engines...
Let's see and compare the Elo performance and then its a good idea to check the similarity of the participated scratch engines

So, I will follow Adam's recommendation about the limit should be 60 % +


Best,
Sedat

Note that there are not only 2 options.

In the case of Movei it did not start directly from chess playing code of another engine because names of some variables arrays and functions is not chess playing code.

I did not start with evaluation code from a different engine so I expect no similiarity between playing style of movei and playing style of tscp.

There is some small similiarity of latest Movei to fruit2.1 becaue I added
some ideas based on fruit2.1's code but even in this case it is clearly lower than 55% and you only has 52.88% so I will be surprised if movei has a similiarity of more than 55% to some previous engine.

You can download the similarity matrix from
http://www.top-5000.nl/clone.htm to get this number.

[Sedat Canbaz]

Btw, I included Fruit 2.1 too in the current scratch testing

As far I know Fruit is a scratch engine and many engines are based or learned from Fruit code, let's see Fruit own playing style, the Elo performance and the similarity comparing with the rest scratch engines...

[Arsha Mahdavi]

Hi Sedat,

Unfortunately I cannot target application for Windows XP at the moment but I will try later. Since changes to the source code are very small, I’m pretty sure others (jose velasco?) can make proper compiles for Windows XP 64.

Here are the Smash (latest) source code.

Deep Smash source code (latest)

[Uri Blass]

Btw, I included Fruit 2.1 too in the current scratch testing

As far I know Fruit is a scratch engine and many engines are based or learned from Fruit code, let's see Fruit own playing style, the Elo performance and the similarity comparing with the rest scratch engines...

Note that fruit's influence is only in latest version of Movei.
early version of Movei are earlier to fruit so they have no fruit influence.

[Sedat Canbaz]

Arasan has a dev history going back to 1994 so I think it counts as a "from scratch" engine. There was hardly anything to clone from back then.

The current codebase has some code of external origin, including the Nalimov tablebase interface and magic bitboard code derived from Pradu Kannan's work.

--Jon

Dear Jon,

Arasan is included in the current SCCT Scrach testing

Arasan is one of my favorite chess programs, I started using Arasan since 1997/98 years

Many thanks for your enormous efforts...

Best,
Sedat