The open source scam

[Michael Sherwin]

First, let me say that this is not directed towards anyone nor do I have anyone in mind. This is just a hypothetical possibility. I know that this will not win me any friends--oh well.

The more that I think about having open source for strong chess engines the more that I am not in favor of it. This is just the latest notion to have crossed my mind about the matter--that it could be a very clever scam.

This is how the scam would work:

1.) write a strong open source engine, but leave out some really good secrets

2.) get lots of help improving it (it is strong, so people will help)

3.) close the source and add the secrets

4.) win the World Championship and/or go commercial making lots of money

5.) pitty all the poor suckers that helped you do it

I disagree.
If you can do 1 then you probably do not need other people to do 2 for you.

I think that you overestimate 2.
What Thomas did for fruit is things that Fabien also could do for himself
but he could prevent Vasik to learn from fruit source by not publishing it.

It means that Fabien could probably get the same level without releasing his source but the opponents could not get the same level so fruit could be relatively better in comparison to other programs assuming that it is a closed source program.

Uri

You are making this specificly about Fruit, when it originally was not.

Even if you can do 1.) it may help to do 2.) in the very competitive situation at the top. Your logic is flawed on this point.

I do agree with you that Fabien should not have released his source code. And I agree that Fruit would be the strongest program if he had not done so. I hope that Fabien writes a completely new program that does win the world championship and dominates world chess for the next decade. I hope also that he does not release the source code untill way after.

[Alessandro Scotti]

Fabien did not continue from code of other but from code of himself.
I believe that Fabien did a big mistake by releasing his code and the result is that fruit lost the first place for rybka.

Uri

Hi Uri,

One hundred percent agreed on second point.

But, how do you know for sure that the newer Fruit versions did not benifit from contributions gained from the open source version?

I agree with Uri on both points (and also with you)!

We can never know for sure what is inside a closed source engine of course, but in the particular case of Fruit I think Fabien has convincingly shown that he can do it better... no need to copy anything!

[Michael Sherwin]

Fabien did not continue from code of other but from code of himself.
I believe that Fabien did a big mistake by releasing his code and the result is that fruit lost the first place for rybka.

Uri

Hi Uri,

One hundred percent agreed on second point.

But, how do you know for sure that the newer Fruit versions did not benifit from contributions gained from the open source version?

I agree with Uri on both points (and also with you)!

We can never know for sure what is inside a closed source engine of course, but in the particular case of Fruit I think Fabien has convincingly shown that he can do it better... no need to copy anything!

And I agree with you to a point. However, there is still the possibility of 'blind luck' involved with the strength of Fruit. I gather from Fabien's writings that he was quite surprised at how strong Fruit was becomming. Many of his words indicate that he thought that he was not producing anything super special. He even went so far as to say that he did not invent anything new. I am willing to give Fabien the benifit of the doubt and 'believe' that any information that he had gathered from making Fruit open source is not being used in the now closed source version. Innocent untill proven..., however, that is a stretch of credulity.

But once again, my original post has nothing to do with a specific program or its author. I only brought up the case of Fruit, because, of your statement that Fabien can do anything that he wants as he is the original author. Is this really true or not? Another poster in this thread, Jon, seems to indicate that it is not. And that my concerns are baseless as long as the engine is placed under a GPL that prevents a person perpetrating such a scam.

Personally, I think that GPL's have very little teeth and unless someone is openly and blatantly violating the GPL not much can be done against them.

[Alessandro Scotti]

But once again, my original post has nothing to do with a specific program or its author. I only brought up the case of Fruit, because, of your statement that Fabien can do anything that he wants as he is the original author. Is this really true or not? Another poster in this thread, Jon, seems to indicate that it is not. And that my concerns are baseless as long as the engine is placed under a GPL that prevents a person perpetrating such a scam.

OK, here's another view on this subject.

When you write an original piece of code you, as the only author and owner, can do whatever you like with that code.

In particular, you can give someone else permission to use that code, which is called licensing. Licensing does not mean that you part with your authorship or ownership of the code, only that others can use it, usually subject to some licensing terms.

The GPL is one set of licensing terms. If I want to use some code that has been released under the GPL, I can only do it _legally_ if I accept to follow the GPL terms. Note that the original author is not subject to the GPL, because he does not have to license his own code to use it! The license only applies to "others" that want to use the code.

The GPL states that modifications to the original source code must be made public and also licensed under the GPL (unless I keep everything private and never distribute the derived software).

Still, every contributor retains ownership of his changes/additions (they should be noted in the source code, which is almost never done though) and those cannot be claimed back by anyone else.

So if the original author of some GPL code wants to "close" the source and keep future code changes for himself he is certainly entitled to do that. However, he would not be able to include any code snippet from derived version because in that case he would need a license to use that code (he is not the author of those parts) and because that license is GPL he would then be forced to release as GPL all the rest.

In our case, this only applies to plain source code. If you take some ideas from GPL code and implement those ideas in some original way then you don't have to release your original implementation under the GPL.

[Michael Sherwin]

But once again, my original post has nothing to do with a specific program or its author. I only brought up the case of Fruit, because, of your statement that Fabien can do anything that he wants as he is the original author. Is this really true or not? Another poster in this thread, Jon, seems to indicate that it is not. And that my concerns are baseless as long as the engine is placed under a GPL that prevents a person perpetrating such a scam.

OK, here's another view on this subject.

When you write an original piece of code you, as the only author and owner, can do whatever you like with that code.

In particular, you can give someone else permission to use that code, which is called licensing. Licensing does not mean that you part with your authorship or ownership of the code, only that others can use it, usually subject to some licensing terms.

The GPL is one set of licensing terms. If I want to use some code that has been released under the GPL, I can only do it _legally_ if I accept to follow the GPL terms. Note that the original author is not subject to the GPL, because he does not have to license his own code to use it! The license only applies to "others" that want to use the code.

The GPL states that modifications to the original source code must be made public and also licensed under the GPL (unless I keep everything private and never distribute the derived software).

Still, every contributor retains ownership of his changes/additions (they should be noted in the source code, which is almost never done though) and those cannot be claimed back by anyone else.

So if the original author of some GPL code wants to "close" the source and keep future code changes for himself he is certainly entitled to do that. However, he would not be able to include any code snippet from derived version because in that case he would need a license to use that code (he is not the author of those parts) and because that license is GPL he would then be forced to release as GPL all the rest.

In our case, this only applies to plain source code. If you take some ideas from GPL code and implement those ideas in some original way then you don't have to release your original implementation under the GPL.

Everyone interested in this subject should copy this explaination. It is 'spot on'.