Chess programs :: Source available, but not today?

[Guenther]

(Question for my XB/UCI chronology and CPW and CCW)

What to do with programs, which were announced and released as open source, e.g. at github, but later their repos vanished for unknown reasons?

This is the case for all repos currently from Folkert (Embla, Micah,feeks) and for Dorpsgek from Matthew, his github was deleted already
a few years ago IIRC.
For now I have removed all obsolete github links and instead of describing them as 'source available = yes' I changed it to '%',
which usually means just 'partially - or for some version(s) only' - actually it's just a reminder for me to perhaps change it to not available.

Should those sources be collected and made available again, or would this be a bad idea ethically (we don't know, if the authors want this)?

Guenther

[jorose]

This is an interesting problem. My intuition would be that if they are open source it should be fine to rehost them. That being said if the source is deliberately being taken down by the program author (eg because they don't want to have a github account), the ethical situation is less clear.

For this reason and for the record let me here state: My intent when releasing open source chess programs is for them to be for the community to enjoy for eternity. I don't want my code to be a burden for anyone, but if anyone sees value in it they should feel free to host and distribute it freely under the license terms.

[hgm]

Releasing something as open source with a license that permits rehosting (such as GPL, MIT) is an irreversible act. There is no way the copyright holder can rescind this. It is a binding contract with the rest of the world.

[syzygy]

Releasing something as open source with a license that permits rehosting (such as GPL, MIT) is an irreversible act. There is no way the copyright holder can rescind this. It is a binding contract with the rest of the world.

That may be the idea behind such licences, but it isn't necessarily true.

At least in Dutch law, the concept of a binding contract with the rest of the world doesn't exist. If the GPL is a contract, it is between the copyright holder and individuals who have somehow performed an act that expressed their will to accept the copyright holder's GPL offer. Once the offer is withdrawn, individuals who have not yet accepted the offer can no longer accept it. But even those that have accepted it are not safe, since contracts agreed to for an indefinite period of time can normally be cancelled by either side provided sufficient notice is given.

In my view, the GPL isn't a contract but a one-sided act of giving recipients of a copy permission to do the acts for which the copyright holder has the exclusive rights, like putting up a sign allowing everyone to picknick in your garden. Obviously putting up such a sign does not prevent you from taking the sign down again. (I guess you may have to accept that some people finish their ongoing picknick.)

In any event, releasing version 1.0 under license X doesn't oblige you to release version 1.0.1 under the same license.

[Ras]

Releasing something as open source with a license that permits rehosting (such as GPL, MIT) is an irreversible act.

Actually, it isn't - because licence agreements don't make laws, and laws are always higher in rank so that they can invalidate any part of a licence text.

For example, the US have an interesting copyright law that was designed in 1976 to address the situation that a company pressures an author into a poor deal and then makes millions while the author gets next to nothing. 35 years after the licence transfer, the author has a 5 years time window to retro-actively terminate the copyright transfer, with 2 years of written notice.

Source: https://en.wikisource.org/wiki/Shrinkin ... the_Public

[hgm]

In any event, releasing version 1.0 under license X doesn't oblige you to release version 1.0.1 under the same license.

But this was not the issue. The issue is whether I would be allowed to host version 1.0 (source-wise), or whether the copyright holder can forbid me to do so.

As Rasmus pointed out, copyright law might provide an opportunity to revoke any licence agreement for copying(including the GPL) between 35 and 40 years after the first release. But it would require legal action by the copyright holder. Merely removing the source from their own website cannot be considered a termination request for the license, even if it would occur in the mentioned time interval. I am not sure anyway whether this holds globally, or is just an American thing.

[Ras]

But it would require legal action by the copyright holder.

Only written notice. It's completely unclear how that would even work out in practice with multiple forks all over the internet, or with only code fragments being used.

I am not sure anyway whether this holds globally, or is just an American thing.

That's US copyright. Another interesting thing about the US law is that it cannot be licenced away via the licence text. That makes sense because otherwise, the same companies that enforced bad deals would also pressure authors to waive this right - which would make the law pointless.

Other countries have different laws, and it may not even be obvious which national law is applicable.

[hgm]

So the theoretical possibility exists that MicroSoft in 10 years will approach the heirs of Linux Torvalds, and offer them a couple of million bucks to revoke the permission to use Linux.

[Ras]

So the theoretical possibility exists that MicroSoft in 10 years will approach the heirs of Linux Torvalds, and offer them a couple of million bucks to revoke the permission to use Linux.

At least in the US, but not for Linux as a whole, only for the code parts Torvalds himself wrote between 35 and 40 years before the reclaiming (if they are still in Linux by then). It's unclear how that would play out because the letters of the law are one thing, but categorising a real world case and deciding under which laws it falls, that's where it becomes difficult. It's also unclear who exactly is the licencee, and whether/how that applies to sub-licencees, and who they are. Or the law could be updated until then to address this constellation in whatever manner. Very difficult even for expert laywers, I think.

[syzygy]

As Rasmus pointed out, copyright law might provide an opportunity to revoke any licence agreement for copying(including the GPL) between 35 and 40 years after the first release. But it would require legal action by the copyright holder. Merely removing the source from their own website cannot be considered a termination request for the license, even if it would occur in the mentioned time interval. I am not sure anyway whether this holds globally, or is just an American thing.

The possibility to revoke a license after 35 to 40 years is an American thing. In the Netherlands, you can almost certainly unilaterally revoke the GPL at any time (provided you give some consideration to persons who have come to rely on the license).

The fact that the US has a specific regulation suggests that in the US the GPL cannot be revoked earlier than after 35 years.