[Macha]

GPLv3 does, but GPLv2 has basically the same language as the OGL v1.0a in terms of the revocability, share-alike nature and "or later" clause, just with GNU replaced by Wizards and software references like the distinction between source and binary removed (in fact, it seems likely the OGL just copied GPLv2)

Apache prior to 2.0, all versions of MIT or BSD also do not go out of their way to be irrevocable.

[tptacek]

Yep. I'm not super interested at all in tabletop RPGs, but the licensing controversy here is super fascinating. I've read some reasons for OGL 1.0-reliers to be optimistic, but the whole thing seems pretty unsettled. If 1.0's revocability holds up, most HN'ers assumptions about how licensing of WOTC property works are going to be broken, because we've gotten so comfortable with FOSS's comparatively sane norms.

[Macha]

Like, if Wizards does get their way here, it seems like a death knell for the MIT, GPLv2 and BSD licenses in their current form. It would render them unwise to use in commercial code or as referenced dependencies of open source codebase as the author could revoke it any time, as well as invalid to incorporate MIT or BSD code into GPLv3 or Apache 2.0 licensed projects as you do not have the permission to grant an irrevocable license to your sublicensees

[tptacek]

No: the irrevocability of the mainstream FOSS licenses has already been tested.

People are doing a lot of axiomatic reasoning on this thread: "FOSS works this way, so the WOTC license should work similarly". Nope! Your best bet is probably just to go read actual legal analyses (that's all I'm doing, reading and relaying things).

[Macha]

The legal analyses I've found fall on one of two sides:

1. IP licenses, especially one sided take it or leave it licenses, are revocable, sometimes even if the license expressly says otherwise because of restrictions on perpetuities.

2. A license turns into a contract when there is consideration from the other side. Licensing your own work under a license because of a virality/share-alike clause has mixed reactions on if its consideration but more fall on the side that it is than it isn't.

Neither of these positions would seem to be to distinguish the OGL from GPLv2 here, and position 2 would set a position where OGL is not revocable but MIT/BSD are.

[tptacek]

Yep, that's the optimistic argument as I understand it.

[da_chicken]

This isn't really the case. The text of the OGL v1.0a is really very terse, especially compared to the GPL v2.

Further, there's also quite a well established argument that the GPL v2 would withstand an attempt to revoke the license: https://copyleft.org/guide/comprehensive-gpl-guidech8.html#x...

The text that that argument relies on in the GPL v2 is simply not present in the OGL v1.0a.

[BryantD]

Creative Commons also made the same change at about the same time. I wish I knew more about why exactly everyone did this -- I haven't been able to find contemporary discussions that are clear, although I suspect I just haven't looked hard enough.