Skip to content

Instantly share code, notes, and snippets.

@humphd

humphd/LICENSE.md

Last active October 2, 2021 21:23
Show Gist options
  • Star 1 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
  • Save humphd/2832b267ee1dfc0329a58a638bc20d4c to your computer and use it in GitHub Desktop.
Save humphd/2832b267ee1dfc0329a58a638bc20d4c to your computer and use it in GitHub Desktop.
Download ZIP
Archive of Twitter discusion on MIT License birthdate
Raw
LICENSE.md

Introduciton

The following is an archive of a number of parallel discussions that happened on Twitter on 1 April, 2019.

The original quesion was posed by me, and signal boosted by @stshank:

When was the MIT License created? I can't find any source that gives a year.

Replies

Because Twitter doesn't make it easy to provide a useful timeline, I've tried to include everything in one place with links to original tweets. NOTE: the order isn't necessarily identical, and I've not included some replies that don't add any content, likes, retweets, etc.

  • X11 or MIT? https://directory.fsf.org/wiki/License:X11. ask because MIT is, AFAIR, X11 derivative. I confess to being fallible, and may be totally wrong. @deadsquid
  • My understanding of that though is that FSF objects to calling it "MIT license" because MIT has used a variety of different license. I think it was first used for X11, hence the FSF's "renaming" of it. It's surprisingly elusive to find. I think it was right around the same time as BSD (1988). X11 came out in 1987, though X preceded that by a few years. I don't know if X11 was MIT License (or X11 license as the FSF likes to call it) from Day One. So AFAICT, it's sometime around the same time as BSD and the release of X11 (i.e. 1987/88). But unless someone involved with, say, X11 recalls that they specifically wrote the MIT License for X11 from Day One, exact answer seems obscure. I suspect that there's something non-accidental about the differences between BSD and MIT given the closeness in time but, again, unless there were contemporary Usenet postings or personal recollection, hard to know. @ghaff
  • X11 was 1987, though there were proprietary versions before it. Cc @keith_x11 @JimGettys. Overall a lot of software came out of the Athena project at MIT around that time, predominantly MIT licensed http://news.mit.edu/2018/mit-looking-back-project-athena-distributed-computing-for-students-1111 @_msw_
  • It was a bit after my time there but it was an interesting development at a time a lot of other schools were bringing in PCs en masse. @ghaff
  • X10R3 and X10R4 came with https://keithp.com/data/mit-copyright.h … which is BSD-like, but not the same as the the X11R1 classic https://keithp.com/data/main.c @keith_x11
  • Somehow the left turn of X11R6 sample implementation as released by @TheOpenGroup in 1998 slipped out of my mind... In which they changed the license to one that would require licensing when used in products that generate revenue... https://web.archive.org/web/20060714214059/http://cbbrowne.com/info/x11r6.4.html @_msw_
  • Files in X11R1 with the X11 license appear to have had it added in 1987. Many files also have earlier copyright dates, but the X11 license is always preceded by a 1987 Copyright). Earlier X versions used a different license, also not share-alike. @keith_x11
  • I believe the MIT licence was created in 1985, or thereabouts. I'd have to go dig through my mail/files, as it was created for distributing X. X version 6 was probably around the first version freely distributed. No, the license predates X11, though we were much sloppier then about the text than later. The MIT license was developed originally for X, however. We were much more consistent making sure the text was correct by X11; the license itself well predates X11. Some of it was copyright Barbara Liskov, who Bob Scheifler worked for. Distributing X under license became enough of a pain that I argued we should just give it away. @JimGettys
  • @keith_x11 had an earlier license copy from X10R3 and X10R4 which, as you say, was similar but different. Thanks! @ghaff
  • But IBM would not touch public domain code (anything without a specific license). We went to the MIT lawyers to craft text to explicitly make it available for any purpose. I think Jerry Saltzer probably did the text w. them. I remember approving of the result. The FSF likes to bitch about naming. Ergo "Linux". Best to not worry about such things.... As I said, we weren't very consistent in those days. First there were early X versions, distributed under proprietary MIT terms. Then the code we developed was made freely available, under something approximating today's MIT license, probably around X version 6, though I'd have to check. 1985 I think. @JimGettys
  • Thanks, folks. The original question (from @humphd) around when the license originated was really interesting to me. I didn't know it was even earlier than 1996, and I love learning about this. Was it mostly because a license was expected but not formalized (e.g. put by lawyers)? @deadsquid
  • Specifically, the license was crafted as IBM did not like public domain code (probably correctly, due to differences in international copyright law). IBM was one of Project Athena's sponsors. @JimGettys
  • As I understand it, that was the impetus behind CC0 as well. IANAL, but public domain is apparently problematic in some places. @ghaff
  • Yup. We never made the mistake of the 4 clause Berkeley license. I don't care which people use significantly, though due to history, I'll always ask "3 or 4 clause" for Berkeley, a historical complication that doesn't exist for MIT. Someplace in my files I have a (paper) letter with the exact date that I received authorization to start distributing X freely. Someday I expect to scan that sort of stuff and put it up somewhere on line (probably http://freedesktop.org 's gitlab). I've never tried to figure out if the GPL or MIT license is earlier in time. As far as I could tell, they are roughly contemporaneous. Both licenses have their uses in different circumstances. @JimGettys
  • GPLv1 was 1989, so MIT (and BSD) were earlier. Though the Free Software Definition was a couple years earlier (1986). @ghaff
  • Yeah, but just as @keith_x11 illustrates that we weren't very consistent on wording, IIRC, Stallman emacs was being distributed under something like the GPL already, before the GPL V1. It gets complicated, and it's easier to just note that both licenses were already in use. @JimGettys
  • But it wasn't originally under the license we now call MIT @richardfontana
  • In any case, @keith_x11 provided a 1985 sample for the license used for X, which matches my memory (which I think was also refreshed some years back by going through all this for a patent case I worked on (we won! those bad patents are no more). I came across the letter then. @JimGettys
  • More than a mere academic question. Certain companies contend the MIT license cannot be read as granting a patent license "because MIT says they didn't intend that" - but not clear to me the modern MIT license specifically came out of MIT. I guess @keith_x11 would know though @richardfontana
  • I used to worry about patents, until I went and chatted with my local patent lawyer expert in the IP. He reassured me that there is good case law around there having to be a limited patent grant, due to estoppel. And yes, the license came out of MIT's lawyers.... @JimGettys
  • These are both different from the modern MIT license in certain ways which could be material to the patent license question @richardfontana
  • Do you know when the MIT license settled into its current language? @ghaff
  • Pretty sure with X11R1. We did a lot of release engineering for that and went through the code base pretty carefully. @JimGettys
  • I don't know, but I have wondered whether it involved a mislabeling (by OSI?) @richardfontana
  • The OSI was quite a lot later. I'm sure they made their own mistakes as well, just as we weren't careful early on. We became careful, with time. If I were you, I'd go chat with Scott Peterson, who works for you guys the last I knew. He's the lawyer who put my mind at rest.... (while we were working at HP). @JimGettys
  • It is tough to argue that I gave you a license that did not implicitly licence you the rights to operate within the expressly granted parameters of that license though. I would hope a judge would disregard such an argument. @paulwberg
  • The X11 license stayed the same from R1 through R5 at least. Take a look at dix/main.c: https://gitlab.freedesktop.org/xorg/xserver/blob/master/dix/main.c … That's got three licenses, the Open Group license, the original MIT X11R1 license and a Digital license. I'm betting lawyers were paid to "fix things"... https://gitlab.freedesktop.org/xorg/xserver/blob/master/dix/main.c @keith_x11
  • It looks as if the 1998 Expat license at some point became the OSI-approved MIT License (i.e. the "modern" MIT license) which is in the same spirit but has different particulars. No idea of the process whereby that happened. @ghaff
  • The third (Digital) license seems to be identical to what's now called the MIT license except it has an anti-publicity clause at the end @richardfontana
  • Can I ask why the newer Digital license may have a different impact on patents than the original X11 license? I try to use GPLv3 on new code these days in part because it explicitly mentions patents (as well as being share-alike). Well, that's interesting. Only about half of the files in the X server appear to include the "MIT License". The oldest I could find with 'and to permit persons to whom the Software is furnished to do so' appears to be from 1995 https://gitlab.freedesktop.org/xorg/xserver/blob/master/Xext/vidmode.c @keith_x11

Follow-up (Nov 2020)

Jerry Saltzer has just published a paper "The Origin of the “MIT License”: https://ieeexplore.ieee.org/document/9263265 It is great to get a reliable record of this history into the public record. And it can be freely downloaded, of course...Jerry Saltzer checked with the different players of the time and dug through his mail and archives. I believe that his paper is accurate and puts credit where credit is due. Jerry himself of course should take a good chunk of the credit as his roll was central to it's drafting. @JimGettys

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment