Fruit vs. Toga poll

[bob]

My comments weren't based on "license" issues. They were based on the usual "one entry per author" rule every tournament I have played in has used.

Sure, and I agree about that, as does almost everybody else, I guess. The question is: If two people want to enter the same tournament with programs which share code, which of the two programs should be allowed to participate? This question is easy to answer in the case of two Craftys and in the case of Stockfish/Glaurung, but not so easy in the case of Fruit/Toga.

Tord

I think it is obvious there as well. Fruit. That is the _original_ work. If Fabien doesn't want to enter, he could certainly say "OK, Toga can enter, I am not playing." In 1997 at the Jakarta WMCCC event, originally the university there said they wanted to enter a modified crafty since they were hosting. And they asked for my permission. I talked with David Levy and group and told them "sure". But nothing happened, they fooled around and a month prior to the event they had still not decided to enter. I contacted David and said "OK, they have not entered and it is not clear that they are, so I am entering, and I did. And then just before the tournament they finally decided they were going to enter as well, after we had already made arrangements for the machine (GM Roman Dzhindi and a friend of his bought a P6-200 for the event, and they bought a ticket to Jakarta for someone that was interested in operating Crafty.) There was no way to get a refund and I told David "We are coming." After lots of discussion they asked me "Do you mind of Gunda-1 (a crafty derivative) enters also?" I replied that I thought it was a bad idea, but I would not object if no one participating objected. Crafty and Gunda finished in either 3rd and 4th or 4th and 5th, I am not sure. I thought it unfair to the new participants and decided that would never happen again with Crafty. And I think that same reasoning applies to fruit/toga. Either one, or the other, but not both. I'm willing to play either one. And personally, I am willing to play both. I'd prefer to have the strongest field possible myself. But I don't think it reasonable for newcomers to have to face two really strong programs that are almost identical in source. where they have a very small chance of beating either one. It can be a bit depressing for them to get drubbed by everyone, when they may well become the next bright spot in computer chess, given enough incentive.

As an example, ICC is out of control with computers. I enjoy playing most everyone. But today, most everyone == Rybka. Those of us working on engines can't get many games against others that are actually authors, because of all the Rybka copies running there. I don't see the reason for running something you didn't write. But it is happening, in huge numbers.

[bob]

Tournament rules may make this simpler, and even apply to the glaurung/stokfish case. If we follow similar rules that the ones followed to publish a scientific paper:

* Every author should authorize the submission/registration.

Then, Fabien decides. And in your case, you can block stockfish registration if you want.

I don't think I could, and even if it is possible from a strictly legal point of view (which I doubt, but I am not a lawyer), it would be a violation of the central philosophy of a free program like Glaurung or Stockfish. When my program ships with a license that says that you are free to do whatever you like with it (apart from making it proprietary), what right do I have to tell you later that you are not allowed to use it in some tournament?

The whole point of free software like Glaurung is that it belongs to the entire community, and not to a single author. It isn't really my program any more than it is yours. I just happened to write most of the code. It is true that there is a copyright notice that says "Copyright (C) 2004-2009 Tord Romstad", but if you read the GNU GPL carefully, you will see that this copyright notice is there only to make it possible to use standard copyright laws to enforce the license and make sure the program remains free.

Therefore, I don't have any more right than anyone else to enter my program or a program based on it in the CCT. The guy who will be playing with Glaurung in the coming CCT actually did ask for my permission, which is a polite and friendly thing to do, but not technically required.

Tord

I think this is black and white with no grey at all. Most tournaments say one program per author, and that no program can be a derivative of another program.

So at first cut, Stockfish is not allowed. But since it is derived from your code, you _could_ (if you choose to do so) say "I will allow Stockfish to play as the 'glaurung entry' to eliminate the no derivative issue since now Glaurung is not playing and therefore two modified versions of the same program are not entering, and as the original author you can "bless" either one you want IMHO.

Your license agreement can't override the tournament rules being used, in any circumstance, so that can't be a problem.

[CRoberson]

As far as Fruit is concerned, I request that the specific version
of Fruit be supplied. That is to ensure that Toga isn't used since
some claim Fruit and Toga are the same.


Regardless of Fruit's entry status, Toga should not be allowed.

I've reviewed petitions from several Toga clones that want to enter
the ACCA tournaments. Each one claims their work is significantly
different than the rest and significantly stronger. I investigate the
claims and none have been valid. The amount of code changed is
minimal. They typically change only 5% of the code.

The other problem with Toga is which do you allow to enter? If one
then all or first come only in?

As far as tournaments go, all of the rules prohibited clones prior to
the release of Fruit much less Toga. The only reason I could see the
ICGA allowing cluster Toga in was to allow a scientific experiment on
the clustering of a chess program.

All the tournaments are author tournaments and disallow clones.
The only people pushing for clone entries are the cloners and
some enthusiasts. I don't see how spectators should be allowed a
vote.

An anecdote is the experience I had with Ted Summers at the last
ACCA Pan Am Champs. Ted spent the weekend with us and conversing
with us on all subjects. At the end, he told me that his opinion of
the clone issues had changed. He had learned much more about
what it takes to write a chess program. It seems most spectators
have little idea what it takes and thus can't tell the difference
between a clone and an original effort.


Amongst the authors, we have modified our definition of clone over
the last 20 years. Now, it comes down to you can reuse ideas but
not code. Also, you can not reuse all ideas. Specifically, you can
only reuse anything that is black box oriented - it produces the
same output given the same input no matter who writes the code.
This means reusing the eval is off limits as well as move ordering
and several other things.

The Toga group reused everything! They didn't just reuse ALL the ideas,
they reused ALL the code as well. Also, they have not contributed
a single new idea.

Remember we are talking about entering tournaments which
means tournament rules are the decision maker not license rules.
Let us not confuse the two.

To me the issue is like this:
Someone writes a book and called it "War and Peace". Change
about 5% of it, claim your version is significantly better and sale
it gaining all proceeds for yourself. Didn't somebody do exactly that
in 2008? The enthusiasts should not condone this behavior.

[Christopher Conkie]

What if Fabien Letouzey personally enters Toga and list all the people that worked on the project as co-authors with the authorization of all of them? Would you still say no?

Why enter Toga? His program is called Fruit.

You call changing less than 100 lines work? I bet those who changed them cannot even write a "Hello World" program. If they can (doubtful), they should be able to create their own entry for CCT. Ask yourself why they don't. Ask yourself more importantly why they choose one of the strongest open source programs to amend (or even break).

Could it be that they want to win the tournament?

Could it be that they want to revel in the glory of someone else's hard work?

if the chat performance is an issue, then no operators should be allowed (it may not be a bad idea for some of the tournaments, but that is another issue).

Miguel

By operators do you mean those who changed less than 100 lines in Fruit and then renamed the engine?



If so, yes....they should be barred...from the whole thing, not just the chat.

Christopher

[Eelco de Groot]

As far as Fruit is concerned, I request that the specific version
of Fruit be supplied. That is to ensure that Toga isn't used since
some claim Fruit and Toga are the same.


Regardless of Fruit's entry status, Toga should not be allowed.

I've reviewed petitions from several Toga clones that want to enter
the ACCA tournaments. Each one claims their work is significantly
different than the rest and significantly stronger. I investigate the
claims and none have been valid. The amount of code changed is
minimal. They typically change only 5% of the code.

The other problem with Toga is which do you allow to enter? If one
then all or first come only in?

As far as tournaments go, all of the rules prohibited clones prior to
the release of Fruit much less Toga. The only reason I could see the
ICGA allowing cluster Toga in was to allow a scientific experiment on
the clustering of a chess program.

All the tournaments are author tournaments and disallow clones.
The only people pushing for clone entries are the cloners and
some enthusiasts. I don't see how spectators should be allowed a
vote.

An anecdote is the experience I had with Ted Summers at the last
ACCA Pan Am Champs. Ted spent the weekend with us and conversing
with us on all subjects. At the end, he told me that his opinion of
the clone issues had changed. He had learned much more about
what it takes to write a chess program. It seems most spectators
have little idea what it takes and thus can't tell the difference
between a clone and an original effort.


Amongst the authors, we have modified our definition of clone over
the last 20 years. Now, it comes down to you can reuse ideas but
not code. Also, you can not reuse all ideas. Specifically, you can
only reuse anything that is black box oriented - it produces the
same output given the same input no matter who writes the code.
This means reusing the eval is off limits as well as move ordering
and several other things.

The Toga group reused everything! They didn't just reuse ALL the ideas,
they reused ALL the code as well.

I don't dispute this

Also, they have not contributed
a single new idea.

I dispute this, but would agree most of Toga and Fruit is about getting rid of bad ideas. Most ideas have been tried already in the past, but Fabien managed to find engine parts that actually fit together to make a high performance engine. But that was Fabien's opinion about his own program also.

I agree that for me personally at least, most of this process involves a lot of trial and error. Every change tried however can also be seen as a new approach. Most ideas do not really work in practice. Some of the TDDB programmers have I am sure better ideas than me and certainly much better qualifications as programmer than I do. But how many Formula one engine builders can say they have revolutionized the internal combustion engine?

Remember we are talking about entering tournaments which
means tournament rules are the decision maker not license rules.
Let us not confuse the two.

Your use of the word clone in the earlier part of your post is not very helpful regarding the confusion Charles. Seriously.

To me the issue is like this:
Someone writes a book and called it "War and Peace". Change
about 5% of it, claim your version is significantly better and sale
it gaining all proceeds for yourself. Didn't somebody do exactly that
in 2008? The enthusiasts should not condone this behavior.

The enthusiasts I am sure do not condone this. But basically I think it was a misunderstanding and underestimation of the clone issue in general by the author of said program.

Regards,
Eelco

[Cubeman]

Only one entry, either Fruit or Toga.Leave it to Fabien to decide.The rules should be quite clear about this.If you top programmers have a beef with the Toga people then don't chat to them.They might be able to help you improve your own engine if only you would listen.

[Graham Banks]

Amongst the authors, we have modified our definition of clone over
the last 20 years. Now, it comes down to you can reuse ideas but
not code. Also, you can not reuse all ideas. Specifically, you can
only reuse anything that is black box oriented - it produces the
same output given the same input no matter who writes the code.
This means reusing the eval is off limits as well as move ordering
and several other things.

The Toga group reused everything! They didn't just reuse ALL the ideas, they reused ALL the code as well. Also, they have not contributed a single new idea.

Then why is Toga so much stronger?

[Graham Banks]

Amongst the authors, we have modified our definition of clone over
the last 20 years. Now, it comes down to you can reuse ideas but
not code. Also, you can not reuse all ideas. Specifically, you can
only reuse anything that is black box oriented - it produces the
same output given the same input no matter who writes the code.
This means reusing the eval is off limits as well as move ordering
and several other things.

The Toga group reused everything! They didn't just reuse ALL the ideas, they reused ALL the code as well. Also, they have not contributed a single new idea.

Then why is Toga so much stronger?

I'll add that Toga is based on Fruit 2.2.1, which is vastly different from Fruit 2.3.1 and subsequent versions (according to Ryan).
If the Toga developers haven't contributed a single idea, then why is the latest Toga almost 100 elo stronger than Fruit 2.2.1? That seems to paint Fabien as inept (which he is far from being) because some amateurs could rewrite a few lines of code and get 100 elo improvement.
If a commercial programmer could get that much improvement, they'd be selling us a new engine. In fact, some do so with far less a gain.

[michiguel]

What if Fabien Letouzey personally enters Toga and list all the people that worked on the project as co-authors with the authorization of all of them? Would you still say no?

Why enter Toga? His program is called Fruit.

Of course this is a hypothetical question, but I guess that would be up to him to decide,
For instance, Toga is stronger.

You call changing less than 100 lines work? I bet those who changed them cannot even

It is irrelevant how so few lines are different. Fabien is the main author of Toga and you are just making my point. If he decides to enter Toga, why do you think you have the right to complain about it?

Miguel

write a "Hello World" program. If they can (doubtful), they should be able to create their own entry for CCT. Ask yourself why they don't. Ask yourself more importantly why they choose one of the strongest open source programs to amend (or even break).

Could it be that they want to win the tournament?

Could it be that they want to revel in the glory of someone else's hard work?

if the chat performance is an issue, then no operators should be allowed (it may not be a bad idea for some of the tournaments, but that is another issue).

Miguel

By operators do you mean those who changed less than 100 lines in Fruit and then renamed the engine?



If so, yes....they should be barred...from the whole thing, not just the chat.

Christopher

[swami]

Fabien is the main author of both Toga and Fruit. If he wants to enter Toga under the co-authorship in any tournament instead of Fruit, then that'd be arguably ok. It's no different to other co-authorship. Munjong and Munstin co-authored Ikarus. Ralf and Volker co-authored Spike.

However, Fabien _can't_ enter both Fruit and Toga in the same tournament.
Now that Fruit has joined the tournament, It's not possible for Toga to join anyway.