[kragen]

Actually, Keith Bostic has gone on record saying that the reason he started trying to strip out the AT&T bits from the BSD kernel was because he was inspired by GNU, and wanted to see if the BSD kernel could work as a kernel for GNU. Maybe in Open Sources, I don't remember.

[maxerickson]

I wrote the comment thinking that the BSD releases in the early 1980s had been released under the 'BSD License', I even did some amount of checking (I can't remember what evidence in particular made me decide to post). I see from looking more carefully this time around that the ~1980 license from Berkeley was not the same thing as the later one.

[kragen]

Well, it's also true that the liberalization of the later versions of the BSD license (removing the advertising clause) was at least in part a result of FSF pressure. But that wasn’t what I meant. I had thought that the early BSD releases were under the four-clause BSD license — but you needed the AT&T Unix source code to compile them.

However, this turns out not to be the whole story at all. I tried rooting around in http://minnie.tuhs.org/cgi-bin/utree.pl?file=4BSD without any success at finding a 4BSD license. http://www.oreilly.com/openbook/opensources/book/kirkmck.htm... has Kirk McKusick’s recollection of the history of the licensing; up to at least 4.3BSD-Tahoe in 1988 it’s talking about “site licenses” rather than free-software licenses, and all the recipients needed AT&T licenses as well. It wasn't until the NET-1 release ("the networking tape") in 1989 that what we know as the BSD license existed, and it wasn’t until the NET-2 tapes in 1991 — largely impelled by Bostic — that there was anything like a complete free BSD OS.

http://www.oreilly.com/openbook/freedom/ch09.html talks about how Bostic credits Stallman with inspiring him to care about software freedom.

[maxerickson]

I opened the .tap here in a text editor yesterday, it mentions executing a license and returning it to Berkeley:

http://sourceforge.net/projects/bsd42/files/Install%20tapes/...

(I wasn't super careful to determine that the language there applied to the whole distribution, but it seems fairly likely)

The argument I was trying to make would have survived a 4 clause BSD license in 1980; I wasn't saying that BSD and derivatives as they exist owe nothing to GNU, I (thought I) was pointing out that people were sharing a sophisticated base system under a liberal license prior to GNU. Starting from there and proceeding without GNU it's of course hard to say where things would have ended up, would they have further liberalized the license, would they have filled in the rest of the system, who knows? They certainly might have.

I do think there are economic forces that encourage some sort of open model for software that is necessary and reasonable well understood/explored, but it's hard to examine a notion like that without a history machine.

[kragen]

Right, and it turns out that they were sharing a sophisticated base system; it was just that the licensing, once people started applying copyright to software at all, didn't protect the users very well.

And, yes, clearly information sharing is very good for software — as I pointed out elsewhere in this thread, that’s what SHARE was founded for in 1955 — but it’s also good for chemistry, and yet it took many centuries before we got Priestley and the Invisible College instead of alchemists writing notes in code so their apprentices couldn’t steal them.