Source files of,,,,, and Contribute:
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

fisl-rms-transcript.en.xhtml 45KB

  1. <?xml version="1.0" encoding="utf-8" ?>
  2. <html>
  3. <head>
  4. <title>GPLv3 - Transcript of Richard Stallman from the second
  5. international GPLv3 conference, Porto Alegre, Brazil;
  6. 2006-04-21</title>
  7. </head>
  8. <body>
  9. <h1>Transcript of Richard Stallman at the 2nd international GPLv3
  10. conference ; 21st April 2006</h1>
  11. <p>
  12. See our <a href="gplv3.html">GPLv3 project</a> page for information
  13. on <a href="gplv3#participate">how to participate</a>. And you may
  14. be interested in
  15. our <a href="">list of
  16. transcripts on GPLv3 and free software licences</a>.
  17. </p>
  18. <p>The following is a transcript of Richard Stallman's presentation
  19. made at the second international GPLv3 conference, held in Porto
  20. Alegre in conjunction with the International Free Software Forum
  21. (FISL). Good quality videos of the entire conference should become
  22. available online in the future. This transcript was made from a low
  23. quality recording made with a digital camera. There are some gaps
  24. due to hardware limitations.
  25. </p>
  26. <p>Transcription of this event was undertaken
  27. by Ciaran O'Riordan.
  28. Please support work such as this by
  29. joining <a href="">the Fellowship of FSFE</a>,
  30. and by encouraging others to do so.</p>
  31. <p>Richard Stallman launched
  32. the <a href="/documents/gnuproject.html">GNU project</a> in 1983,
  33. and with it the <a href="/documents/freesoftware.html">Free
  34. Software</a> movement. Stallman is the president of FSF - a sister
  35. organisation of FSFE.</p>
  36. <p>The speech was made in English.</p>
  37. <h2>Presentation sections</h2>
  38. <ol>
  39. <li><a href="#begin">The presentation</a></li>
  40. <li><a href="#what-is-gpl">What is the GNU GPL and Free Software?</a></li>
  41. <li><a href="#what-is-copyleft">What is copyleft?</a></li>
  42. <li><a href="#before-gnu-gpl">Pre-history: before the GNU GPL</a></li>
  43. <li><a href="#liberty-or-death">The "Liberty or Death" clause, the main change from v1 to v2</a></li>
  44. <li><a href="#why-change-it">Why change it? It was always in the plan, hence "or any later version"</a></li>
  45. <li><a href="#tivoisation">First main change: malicious DRM part 1, preventing tivoisation</a></li>
  46. <li><a href="#circumvention-not-prohibited">Malicious DRM part 2, grant of permission to access generated data</a></li>
  47. <li><a href="#software-patents-grant">Second main change: software patents part 1, explicit patent grant</a></li>
  48. <li><a href="#software-patents-retaliation">Software patents part 2, narrow retaliation</a></li>
  49. <li><a href="#licence-compatibility">Third main change: licence compatibility</a></li>
  50. <li><a href="#limits-of-requirements">How the addable requirements are limited</a></li>
  51. </ol>
  52. <h2>Questions and Answers session sections</h2>
  53. <ul>
  54. <li><a href="#q10-gfdl">Is GPLv3 good for documentation (about GFDL)?</a></li>
  55. <li><a href="#q20-affero">Why not put the Affero clause for web services into the licence?</a></li>
  56. <li><a href="#q30-cooperating-with-osi">About cooperating with OSI / ESR</a></li>
  57. <li><a href="#q40-creative-commons">Creative Commons is not one licence</a></li>
  58. <li><a href="#q50-osi-responds">Michael Tiemann of OSI comments on RMS's work</a></li>
  59. <li><a href="#q60-translations">Will there be official translations of GPLv3</a></li>
  60. <li><a href="#q70-fragments">Fragments of a tivoisation question before the camera dies</a></li>
  61. </ul>
  62. <ul>
  63. <li><a href="#links">Links to documents for more information on the
  64. topics covered</a></li>
  65. </ul>
  66. <h2 id="begin">The presentation</h2>
  67. <p>
  68. [Richard starts by noting bad news that he has heard the Brazillian
  69. patent office is now accepting software patents and says that this is
  70. the most important thing for Latin American Free Software supporters
  71. to work on in the coming months.]
  72. </p>
  73. <p class="indent">
  74. Richard Stallman: So now I will turn to the subject I was supposed to
  75. speak about, which is somewhat related: the GNU General Public
  76. License, or the GNU GPL for short.
  77. </p>
  78. <p id="what-is-gpl">[section: What is the GNU GPL and Free Software?]</p>
  79. <p class="indent">
  80. First of all: what is the GNU General Public License?
  81. </p>
  82. <p class="indent">
  83. The GNU General Public License is a Free Software licence, that means
  84. a licence which is designed to give you four essential freedoms - the
  85. four freedoms that define Free Software.
  86. </p>
  87. <p class="indent">
  88. A program is Free Software if you, the user, have these four freedoms:
  89. </p>
  90. <ul>
  91. <li class="indent">
  92. Freedom zero is the freedom to run the program as you wish, for any
  93. purpose.
  94. </li>
  95. <li class="indent">
  96. Freedom one is the freedom to study the source code and change it to
  97. make it do what you wish.
  98. </li>
  99. <li class="indent">
  100. Freedom two is the freedom to help your neighbour, that's the freedom
  101. to make copies and distribute them to others when you wish.
  102. </li>
  103. <li class="indent">
  104. Freedom three is the freedom to help your community: the freedom to
  105. publish or distribute modified versions when you wish.
  106. </li>
  107. </ul>
  108. <p class="indent">
  109. With these four freedoms, the program is Free Software because it is
  110. distributed in an ethical fashion and respects the user's individual
  111. freedom and respects the social solidarity of their community.
  112. </p>
  113. <p class="indent">
  114. If you one of these freedoms is missing, that means the program is
  115. distributed in an anti-social way that tramples the user's freedom,
  116. that attacks social solidarity, in a way that keeps the users divided
  117. and helpless, in a way that should never be done. So, any licence
  118. that grants the user these four freedoms is a Free Software licence.
  119. </p>
  120. <p class="indent">
  121. There are many Free Software licences - too many perhaps - it's
  122. somewhat inconvenient when there are so many but the fact is that
  123. anyone can write a license that grants these four freedoms, and that
  124. is a Free Software licence. If it does grant these four freedoms, it
  125. is a Free Software licence.
  126. </p>
  127. <p class="indent">
  128. The GNU General Public License was among the first Free Software
  129. licences. I do know of one that existed before and that is the
  130. licence of TeX, but the GNU General Public License was designed
  131. specifically to be a Free Software licence.
  132. </p>
  133. <p id="what-is-copyleft">[section: What is copyleft?]</p>
  134. <p class="indent">
  135. There are two broad categories of Free Software licences. Those that
  136. are copyleft and those that are not.
  137. </p>
  138. <p class="indent">
  139. Copyleft means that the licence actively defends the freedom of all
  140. users and what this means is: we don't just stop with giving you these
  141. four freedoms but we ensure that every user that gets a copy of the
  142. program gets the same four freedoms in the same way.
  143. </p>
  144. <p class="indent">
  145. This is the idea of copyleft. A non-copyleft licence grants the four
  146. freedoms but it allows creating modified versions which are not free
  147. at all which means that by the time the program gets to you, it might
  148. have passed through some intermediary who has taken away the freedom,
  149. so you might get the code without the freedom. Whereas a copyleft
  150. licence says that when intermediaries pass the code along to you, they
  151. must give you the same freedom they got from us. So you must get the
  152. freedom. All users must get the freedom. That is the basic idea of
  153. copyleft.
  154. </p>
  155. <p class="indent">
  156. So the GNU General Public License says: you are permitted to modify
  157. this program, you are permitted to copy this program, you are
  158. permitted to distribute those copies, you are permitted to distribute
  159. modified versions. But there's a condition: whatever version you
  160. distribute, must be under this licence, and it must be available as
  161. source code. That means that people who get it from you will get the
  162. same freedom that you got, so all the users get those freedoms.
  163. </p>
  164. <p id="before-gnu-gpl">[section: Pre-history: before the GNU GPL]</p>
  165. <p class="indent">
  166. The first General Public Licenses that I worked on in the 1980s were
  167. for individual GNU programs. For instances, GNU Emacs had the GNU
  168. Emacs General Public License, but then I realised that there was a
  169. problem. Different GNU packages had licences which were just slightly
  170. different because each licence had to state the name of the program
  171. that covers it. So GNU Emacs had the GNU Emacs General Public License
  172. and GCC had the GCC General Public License and they were the same
  173. except that one says that it is the GNU Emacs General Public License
  174. and the other said it was the GCC General Public License, which meant
  175. that they were not actually the same, they were just isomorphic.
  176. </p>
  177. <p>
  178. [laughter]
  179. </p>
  180. <p class="indent">
  181. And this was a practical problem. I realised that it would be much
  182. more convenient if the licences were identical in all programs, so I
  183. worked on this some more and I arranged that the licence did not need
  184. to have the name of the program in it anymore. So the licence for
  185. Emacs could be word for word identical to the licence in GCC.
  186. </p>
  187. <p class="indent">
  188. So the GNU General Public License version 1 was published in 1989 as a
  189. Free Software copyleft licence that you could drop into your program
  190. in order to release your program under that licence.
  191. </p>
  192. <p id="liberty-or-death">[section: The "Liberty or Death" clause, the main change from v1 to v2]</p>
  193. <p class="indent">
  194. Some fairly small changes were made to produce version 2 in 1991. I
  195. think the biggest change in version 2 was to introduce a "Liberty or
  196. Death" clause - the clause that says if somebody uses a patent or
  197. something else to effectively make a program non-free then it cannot
  198. be distributed at all. And the reason is this clause is extremely
  199. important. You see, there is nothing we can do to take away the legal
  200. problem of software patents. If somebody has a patent that covers a
  201. free program, he can stop that program from being distributed. He can
  202. sue everybody who distributes it if he wants to actually pay lawyers to
  203. do that.
  204. </p>
  205. <p class="indent">
  206. We can't prevent that by what we put in the licence for the program
  207. itself, but we can prevent something that's even worse. A fate that's
  208. worse than death for the program and that is suppose the patent holder
  209. says to every distributor "I'll sue you unless you take out a licence
  210. and pay me" and thus in effect, people will have to get permission and
  211. pay for permission to distribute the program, it would not be a free
  212. program anymore.
  213. </p>
  214. <p class="indent">
  215. We can't write our licence to make software patents go away, but we
  216. can write our licence so that they can kill the program but they can't
  217. enslave it.
  218. </p>
  219. <p class="indent">
  220. This makes a difference because if they thought that by suing, they
  221. could make everybody pay them, it would be extremely attractive,
  222. extremely tempting to do that. And so we would be seeing lots of free
  223. programs that had been made effectively be proprietary, whose licences
  224. have become fiction because you wouldn't really have the freedoms that
  225. they say you have.
  226. </p>
  227. <p class="indent">
  228. To kill the program is less attractive because that doesn't bring you
  229. money - at least not directly. So by saying "You can kill this
  230. program but you can't enslave it's users", to some extent we have
  231. protected ourselves from either one because only if somebody is acting
  232. from malice, from an attempt to just cause trouble for us, does he
  233. gain.
  234. </p>
  235. <p class="indent">
  236. Of course, there are some companies that do have patents that might
  237. want to attack us just for the sake of making us fail and one of them
  238. has obtained substantial presence here and asks for our cooperation,
  239. but at least this clause - the "Liberty of Death" clause - does some
  240. good. That was the main change in version 2 in 1991.
  241. </p>
  242. <p class="indent">
  243. Back then, 1991, it was mostly parts of the GNU system, GNU packages,
  244. that used the GNU General Public License, but during the 1990s many
  245. other people started writing Free Software and releasing their
  246. programs under the GNU GPL. It was designed so that this would be
  247. easy to do, I hoped people would do this, and in the 1990s they
  248. started. So, the result is that the GNU GPL is now the principal free
  249. software licence, used for about 70% of Free Software packages.
  250. </p>
  251. <p id="why-change-it">[section: Why change it? It was always in the
  252. plan, hence "or any later version"]</p>
  253. <p class="indent">
  254. So why are we changing it? Well, from the very beginning, I figured
  255. that we would be changing it some day. Things change, new problems
  256. always develop, laws are changing, and so I realised that we would
  257. have to change this licence some day. And, I also figured that this
  258. would apply to many programs written by lots of different people, so I
  259. suggested to people: please release your programs saying "GNU GPL
  260. version 1 or greater", saying this program may be used under version 1
  261. of the GNU GPL or any later versions, and this way, when version 2 came
  262. out, those programs would automatically be usable under version 2 and
  263. then since version two, we've been releasing programs saying "this can
  264. be used under version 2 or any later version" so that in the future,
  265. when GPL version 3 is finalised and is ready for use and officially
  266. published, all those programs would be available under version 3 as
  267. well, and this way, all these programs will make a smooth transition
  268. to GPL version 3.
  269. </p>
  270. <p class="indent">
  271. Ok, it's not surprising that there would be a change someday, but why
  272. specifically are we changing it now? And what are the changes?
  273. </p>
  274. <p class="indent">
  275. Well, first of all, there is no one big change. There is no one
  276. reason for these changes, each change has it's own little reason. The
  277. basic idea is the same: make sure all the users get these four
  278. freedoms. All the changes are in details. Many of the changes are
  279. clarifications - they're not intended to change the meaning at all,
  280. just make it clearer to everybody that the meaning is what it is.
  281. </p>
  282. <p class="indent">
  283. Some changes are meant to make the GPL more independent of variations
  284. between countries and their laws for software. Although copyright is
  285. pretty much the same thing in all countries, some details vary, so now
  286. we know more, and we're in a position to ask more people for advice
  287. about what the differences are between particular countries' laws.
  288. </p>
  289. <p class="indent">
  290. We've designed the new language to insulate itself from the changes so
  291. that regardless of these changes, the meaning of the GPL is the same.
  292. </p>
  293. <p class="indent">
  294. Well, that's going to be very non-controversial, I'm sure. But there
  295. are some changes that really are meant to be substantive changes.
  296. There are three main areas for this. One is to do with Tivoisation,
  297. another has to do with software patents, and another has to do with
  298. compatibility with other Free Software licences.
  299. </p>
  300. <p id="tivoisation">[section: First main change: malicious DRM part 1, preventing tivoisation]</p>
  301. <p class="indent">
  302. First: Tivoisation. That's a word I made up. It's named after a
  303. product called the TiVo, which is the first example - or at least the
  304. first I remember - of a certain specific threat to our freedom. The
  305. Tivo contains a GNU+Linux operating system - a fairly small one, but
  306. that's what it is. And this means it contains programs released under
  307. the GNU GPL, and as far as I know they're not violating the GNU GPL,
  308. so I'm presuming that the users can get the source code, that they
  309. make it available in one of the appropriate ways. So the users can
  310. get the source code, they can edit it, they can compile their modified
  311. versoins, they can install them in the Tivo, and then they won't run.
  312. </p>
  313. <p class="indent">
  314. You see, the Tivo is designed so that it won't run your changed
  315. versions, it refuses to even start and thus, nominally, Freedom number
  316. 1, the freedom to change the source code so that it will do what you
  317. want, nominally the users have that freedom but really the user does
  318. not have that freedom. You can change the source code so that it
  319. would do what you want, but then it won't do anything.
  320. </p>
  321. <p>
  322. [laughter]
  323. </p>
  324. <p class="indent">
  325. So this is called Tivoisation and this is what we're trying to
  326. prevent. But why are we so concerned with this? Well, why would
  327. anybody do Tivoisation? What is the motive? The reason is because
  328. they're putting malicious features into the code and they don't want
  329. the users to fix these deliberate problems. The Tivo has two kinds of
  330. nasty features.
  331. </p>
  332. <p class="indent">
  333. One is that it spies on the user - it reports what the user watches.
  334. And second, it doesn't allow the user to copy any of the TV programs
  335. out of the Tivo. So this is a form of Digital Restrictions
  336. Management. Now, they're goal is to restrict the users and deny the
  337. users freedom. So, when the users are getting Free Software, the
  338. users can really exercise Freedom 1 and change the software so as to
  339. get around the restrictions, but that would defeat the whole scheme to
  340. restrict you. So, they tivoise the software to stop you effectively
  341. changing the software and that way you can't escape from the
  342. restrictions. In general, tivoisation is done by someone who wants to
  343. impose nasty features, malicious features, on another.
  344. </p>
  345. <p class="indent">
  346. Now, I disagree with those features, I think they're unethical. Well,
  347. as long as people continue to have the full, the real enjoyment of
  348. Freedom number 1, I don't need to worry about those malicious features
  349. because people will take them out, they will fix them, but once the
  350. software is tivoised, the users can't fix it.
  351. </p>
  352. <p class="indent">
  353. So GPL version 3 was designed to prohibit tivoisation. It does not
  354. prohibit the malicious features of the code. It does not prohibit
  355. changing the software to report what the user does. It does not
  356. prohibit changing the software so that it refuses to copy things or so
  357. that it doesn't have a feature to copy things at all, but it does
  358. prohibit stopping the user from making other modifications.
  359. </p>
  360. <p class="indent">
  361. The way it does this is something we had to work on, we had to really
  362. think about how to achieve this. It actually doesn't do this by
  363. prohibiting any mode of distribution. They can design the hardware to
  364. that it requires the binaries to be signed by a certain signature key
  365. in order to run, but they must give you the signature key so that you
  366. can sign your modified binaries. They must give you whatever it takes
  367. to authorise your version so that it will run. So, that is the main
  368. anti-Digital-Restrictions-Management provision of GPL version three
  369. and it actually doesn't talk about Digital Restrictions Management, it
  370. just prohibits tivoisation, so yes, you I can put Digital Restrictions
  371. Management in the software and provide it to you, but you can take out
  372. the DRM.
  373. </p>
  374. <p class="indent">
  375. Well, that's not the only thing we do to prevent DRM, there are a few
  376. other things. We're actually preventing DRM from being imposed on you
  377. such that you can't actually change the software to take it out.
  378. </p>
  379. <p id="circumvention-not-prohibited">[section: Malicious DRM part 2,
  380. grant of permission to access generated data]</p>
  381. <p class="indent">
  382. One is, when a GPL covered program is used to produce encrypted data,
  383. then, the GPL says that they have given you permission to write your
  384. own software to read that data because some countries have a very
  385. nasty law - of course, you can guess which country was the first - a
  386. law that actually prohibits software that enables people to bypass the
  387. restrictions of DRM. If somebody had released a work that is designed
  388. for digital restrictions, and somebody else releases a program that
  389. can access the work without restrictions, that second program is
  390. illegal - in my country, and in Europe, and about fifty countries
  391. who've signed a treaty whose sole purpose is to restrict their own
  392. citizens.
  393. </p>
  394. <p class="indent">
  395. So, GPL says if the first program, the one that makes these works
  396. restricted with DRM is GPL-covered, then that gives everybody
  397. permission to release other GPL-covered programs to read the same
  398. works. In other words, they cannot take advantage of those laws to
  399. forbid the other programs.
  400. </p>
  401. <p class="indent">
  402. Once again, we're not prohibiting people from changing the software to
  403. do DRM, it's only stopping them from prohibiting other from releasing
  404. software they want to use and making the software they want to make.
  405. </p>
  406. <p class="indent">
  407. Also, if the GPL'd program produces output for DRM or communicates
  408. through a protocol which imposes Digital Restrictions Management, they
  409. must give you the necessary codes to write the other side, an
  410. authorised server. So, essentially we've tried to cover all the
  411. bases. Any way that they can try to use a GPL covered program in such
  412. that way that block you or forbid you to change it or write the other
  413. software that works with it, that's not allowed. They have to respect
  414. your right to change the software and write other interoperable
  415. software.
  416. </p>
  417. <p class="indent">
  418. So, that's how we deal with DRM. Mostly by not dealing with it at
  419. all, we just guarantee your freedom so that the people whose goal it
  420. is to take away your freedom cannot succeed.
  421. </p>
  422. <p id="software-patents-grant">[section: Second main change: software
  423. patents part 1, explicit patent grant]</p>
  424. <p class="indent">
  425. The next major area is that of software patents. Now, we took care of
  426. the most direct issue in version two with the "Liberty or Death"
  427. clause. What we've done with version three is a few secondary things.
  428. </p>
  429. <p class="indent">
  430. First of all, in GPL version two we rely on an implicit patent
  431. licence. The issue is: what if the company who distributed the GPL
  432. covered program has a patent that applies to something in the program?
  433. and what if they say "We'll distribute this to you under the GPL but
  434. if you try to use the rights that the GPL gives you, we'll sue you for
  435. patent infringement"?
  436. </p>
  437. <p class="indent">
  438. In the US they can't. In the US, by distributing the program to you
  439. under the GPL they're saying they have no objectins if you carry out
  440. your rights under the GPL, so if they then try to sue you for doing
  441. so, they will lose. However, this isn't necessarily true in other
  442. countries. Some other Free Software licences in the past several
  443. years have had explicit patent licence grants and we decided to follow
  444. them. We're putting in an explicit patent license grant which says,
  445. more or less, what US law would do implicitly.
  446. </p>
  447. <p class="indent">
  448. But then there's another peculiar case. What if someone distributes
  449. the software to you under the GPL and he has a patent licence saying
  450. that nobody is going to sue him, that the patent holder will not sue
  451. him when he distributes the software, but you, when you distribute the
  452. software might be sued by the same patent holder. In fact we could
  453. imagine that the patent holder is going around threatening everybody
  454. all the time.
  455. </p>
  456. <p class="indent">
  457. That would be a bad thing because it would mean that the freedoms
  458. might not be really there. This one distributor gets to enjoy the
  459. freedoms and he's supposed to respect your freedom but he could be
  460. taking advantage of the fact that the patent holder will sue you if
  461. you try. So it's sort of a collusion between these two companies to
  462. make your freedom a sham. So GPL version three says that if he is
  463. knowingly relying on a patent licence for his own right to distribute
  464. this, without getting sued for patent infringement, he must make sure
  465. that you are protected too. He has to negotiate his license so that
  466. it includes everything that's implied by his releasing of his version
  467. which includes your changing the program and your releasing and so on.
  468. </p>
  469. <p class="indent">
  470. However, there's a subtlety here. The big companies have made blanket
  471. cross-licences. IBM and Microsoft surely have a blanket patent cross
  472. licence for all the patents they have and will ever have, and IBM
  473. doesn't even know what the Microsoft patents cover and IBM has one of
  474. these with every other big company that's involved closely with
  475. computers or software.
  476. </p>
  477. <p class="indent">
  478. So IBM has patent licences for loads of things that they don't know.
  479. So the result is that they could have a patent license that makes them
  480. safe and they don't know it. So, we said that it's not fair to put
  481. them in a worse position than you would be in just because they have a
  482. blanket cross-licence and somebody else is explicitly negotiating a
  483. licence, so we said, alright, it will only apply if you knowingly rely
  484. on a patent licence. So if IBM has a patent licence as part of a
  485. blanket cross-licence and doesn't know, then this doesn't apply to
  486. them, but if they find out that this problem is happening and they
  487. have a patent licence, then they have to do something. IBM doesn't
  488. seem to like this very much.
  489. </p>
  490. <p id="software-patents-retaliation">[section: Software patents part
  491. 2, narrow retaliation]</p>
  492. <p class="indent">
  493. Anyway. There is one little bit more about dealing with software
  494. patents. We're concerned about the case where somebody modifies a GPL
  495. covered program and has a patent covering the modifications and never
  496. releases [the program]. If he releases it, he will be giving the
  497. patent licence grant to all the users. So there'll be no problem, so
  498. what if he never releases the program, he just runs it on his
  499. webserver?
  500. </p>
  501. <p class="indent">
  502. Then he's not giving the patent licence grant to anybody. So he might
  503. say "I modified this GPL covered program to do this job, and if you
  504. write your own program to do this job, you write your own code to do
  505. this job, I'll sue you for patent infringement". So, you may say that
  506. at that point, he loses his rights to make more copies and to modify
  507. the program and so on. So basically, for business purposes, his uses
  508. will be untenable. And this way, we protect against that threat.
  509. </p>
  510. <p class="indent">
  511. Now, there are many other Free Software licences that have some other
  512. software patent retaliation clause. Some of them are designed to
  513. retaliate only against aggression. Some of them, in licences that
  514. were written for one company, sometimes retaliated for any suits
  515. against that company. That's not a good thing. Suppose, for
  516. instance, that Company A releases a program that says the licence
  517. terminates if you sue Company A for patent infringement. Well that
  518. might seem reasonable since suing anyone for software patent
  519. infringement is a bad thing, but suppose Company A has a lot of
  520. patents and it goes around suing others, so suppose Company A sues
  521. Company B for patent infringement. What Company B wants to do then is
  522. to sue Company A is the aggressor and Company B is retaliating, so
  523. when B retaliates, B loses her right to use that program because the
  524. program licence says if you sue Company A you lose the right to use
  525. this program. So what this means is that it is retaliating for
  526. retaliation, so it's actually supporting aggression by Company A.
  527. </p>
  528. <p class="indent">
  529. Anyway, we looked at these various kinds of software patent
  530. retaliation clauses and we decided that we didn't want to put any of
  531. them in the GNU GPL, except the one that I described for you five
  532. minutes ago. The very limited one in the case where somebody makes
  533. patented changes and doesn't release them. But, there is a way that
  534. other patent retaliation clauses can get into programs covered by GNU
  535. GPL version three and this brings us to the third main area of changes
  536. which is increased compatibility with other licences.
  537. </p>
  538. <p id="licence-compatibility">[section: Third main change: licence
  539. compatibility]</p>
  540. <p class="indent">
  541. First of all: what does compatibility mean?
  542. </p>
  543. <p class="indent">
  544. Suppose we have two or three programs and each one has its own licence
  545. and you would like to merge them or link them together. Can you do
  546. that? Well, each of these licences might make a condition about the
  547. licence of the combination. So, the question is: is there anyway you
  548. can satisfy both of these licences at once? If there is, then the
  549. licences are compatible, you can make a combination because there is a
  550. way to licence the combination that satisfies both of these licences.
  551. </p>
  552. <p class="indent">
  553. But if there's no way to licence the combination so as to satisfy both
  554. of these requirements together, the combination can't be released.
  555. That means the licences are incompatible.
  556. </p>
  557. <p class="indent">
  558. Now, licence incompatibility is a fact of life. There have always
  559. been incompatible Free Software licences. I'd say the most glaring
  560. kind of incompatibility concerns the licence of TeX, which is
  561. incompatible with itself.
  562. </p>
  563. <p>
  564. [laughter]
  565. </p>
  566. <p class="indent">
  567. If you have two programs that are released under the licence of Tex,
  568. you can't merge them at all. The reason is that the licence of Tex
  569. says that the only way you can distribute a changed version is by
  570. distributing the original version plus a change file - a patch file.
  571. </p>
  572. <p class="indent">
  573. So think about it, you have these two programs both released that way
  574. and you want to merge them. How will you distribute the merged
  575. version. How will you distribute the merged version? Well, this
  576. program says you have to distribute this program plus the change file,
  577. but this [other] program says it has to be this particular program
  578. plus a change file. It can't be both at once, so the licence of TeX
  579. is incompatible with itself.
  580. </p>
  581. <p class="indent">
  582. The GNU GPL is a copyleft licence. The idea of copyleft is, the
  583. licence says when you release a modified version it has to be under
  584. this exact same licence. The result is that, generally, two different
  585. copyleft licences are incompatible. Each one says the modified
  586. version has to be under this license, the other one says, no, the
  587. modified version has to be under this licence. If the two licences
  588. are different, the modified version can't be under both at once. If
  589. it's under this one, it's not under [the other] one.
  590. </p>
  591. <p class="indent">
  592. So, two different copyleft licences, generally, are incompatible. But
  593. in GPL version three we tried a technique that we thought of to make
  594. GPL version three compatible with more licences. Now, GPL version
  595. three is compatible with some licences. Namely, simple permissive
  596. licences. Things like the X11 licence, the revised BSD licence, and
  597. several other important Free Software licences are simple, permissive
  598. Free Software licences.
  599. </p>
  600. <p class="indent">
  601. Now, the reason they're compatible with GPL is because they don't make
  602. any requirement which is inconsistent with the GPL. You can take the
  603. GPL and put it on top of those licences and those licences don't
  604. object, because they're very permissive.
  605. </p>
  606. <p class="indent">
  607. And there's another class of licences which are compatible with GPL
  608. version two, these are the disjunctive dual licences. The typical
  609. example is the licence of PERL. It says "you can use this under the
  610. GNU GPL or the Artistic licence". This is a disjunction between two
  611. licences. A or B. And because the GNU GPL is one option, the licence
  612. of PERL is compatible with the GNU GPL.
  613. </p>
  614. <p class="indent">
  615. So, some other Free Software licences are compatible with the GPL
  616. version two. There are many that are not because they have other
  617. requirements that are not in the GPL and since the GPL version two
  618. does not allow adding any requirements whatsoever, any licence that
  619. contains any other requirement that is not in GNU GPL version two is
  620. incompatible with GNU GPL version two.
  621. </p>
  622. <p class="indent">
  623. So, what we have done in the draft of GPL version three is we have
  624. said there is specific list of certain kinds of requirements that you
  625. can add. And these include some rather trivial requirements that we
  626. think you could add anyway, like saying "my licence must be
  627. preserved", "my copyright notice must be preserved", that does no
  628. harm. And terms like, "you have indicate somehow that it's a modified
  629. version". GPL has always required you indicated modified versions but
  630. it does so in one particular way. But you could add code with a
  631. different requirement for noting modifications and GPL version three
  632. says it's compatible. It accepts that.
  633. </p>
  634. <p class="indent">
  635. And, another particular kind of requirement is patent retaliation
  636. clauses. Within a certain range, those that are not subject to abuse,
  637. we had to work very carefully to design a description of those patent
  638. retaliation requirements that do not support aggression, and those are
  639. the ones we permit.
  640. </p>
  641. <p class="indent">
  642. And finally, we permit the "Affero" requirement. The Affero
  643. requirement is found in the Affero General Public Licence, the Affero
  644. GPL. The Affero GPL is like GNU GPL version two but it has one other
  645. requirement, and that is, if you put the program on a publicly
  646. accessible website, you have to release the source code of your
  647. version.
  648. </p>
  649. <p class="indent">
  650. This is a requirement about public use of a program. So, that's one
  651. kind of requirement that can be added. It's not in GPL version three
  652. itself but it can be added.
  653. </p>
  654. <p id="limits-of-requirements">[section: How the addable requirements
  655. are limited]</p>
  656. <p class="indent">
  657. Let me make something clear. You can't
  658. just take existing GPL version three code and add additional
  659. requirements. What you can do is write your own code with these
  660. requirements placed by you on your own code, or merge in somebody
  661. else's code which has these requirements placed by him on his code,
  662. and when it's merged in, these requirements remain only on the code
  663. that they were applied to. So they don't actually spread to the rest
  664. of the program. If you're using the whole program, of course it
  665. includes that part, so you have to follow the requirements for that
  666. part. But they stay on that part, so if you delete that part
  667. </p>
  668. <p>
  669. [From here onward, battery and memory limitations prevent the
  670. recording from being complete]
  671. </p>
  672. <h2>Questions and Answers session</h2>
  673. <p id="q10-gfdl">
  674. Q1: Is GPLv3 useful for documentation and/or compatible
  675. with the GNU Free Documentation Licence?
  676. </p>
  677. <p class="indent">
  678. Richard Stallman: The two licences are incompatible, but they're just
  679. wholly incompatible. They're two very different copyleft licences and
  680. two different copyleft licences are always going to be incompatible.
  681. The GNU GPL has requirements that are extremely inconvenient for
  682. anyone who wants to publish printed books such as manuals. So, it's
  683. not a good idea to use the GNU GPL for a manual. So, I developed a
  684. different licence for manuals, a licence which I developed
  685. specifically with the idea of encouraging commercial companies to use
  686. it for their manuals. I put in certain features that I hoped would
  687. make it easier to make a profit selling these copylefted manuals and
  688. another special feature which is supposed to make it more comfortable
  689. to release a standards document under this licence.
  690. </p>
  691. <p class="indent">
  692. There's a special kind of section called the "endorsements" section
  693. which can say "This is the official Foobar standard" and anyone who
  694. makes a modified version of the document has to delete that section,
  695. so it will not say that this is an official version of the standard.
  696. </p>
  697. <p class="indent">
  698. And you can put that together with an invariant section saying "if you
  699. don't see a section saying this is the official version of the
  700. standard, then it is not". And so, if your goal is to make sure
  701. everyone knows whether your book contains the official version of the
  702. standard, this will do it. Now it's sad that most standards
  703. organisations real goal in publishing their standards is to make money
  704. from them, so they refuse to cooperate at all, but we want it to be
  705. possible for them to cooperate so that when we pressure them to
  706. cooperate, we need to make it practical for them to cooperate, and the
  707. GNU Free Documentation Licence does.
  708. </p>
  709. <p class="indent">
  710. So these things are incompatible with the GPL.
  711. </p>
  712. <p id="q20-affero">
  713. Q2: It seems the Affero clause is clearly designed to protect people
  714. who use software through their browser...
  715. </p>
  716. <p class="indent">
  717. Richard Stallman: No that's not true, but continue. That was not our
  718. intention. I'll explain but we'll take a look at your question and
  719. then I'll explain.
  720. </p>
  721. <p>
  722. Q2b: Ok, my question which may be based on false premises, is: why was
  723. the decision made to keep the Affero clause as an option?
  724. </p>
  725. <p class="indent">
  726. Richard Stallman: We don't want make that kind of change in the
  727. licence of all the existing GPL covered programs. Some people don't
  728. want that. We were considering putting this feature in the GPL but we
  729. decided it would only be active if the program did something to
  730. activate it. And then I realised that if you want to consider ways to
  731. activate it, this way is just as good as any.
  732. </p>
  733. <p class="indent">
  734. Anyway, this is not really a matter of protecting a user who does
  735. things through a browser. It's not about that actually, it's about
  736. Person A writes a GPL covered program for use on websites, and
  737. releases it, and Someone B makes improvements and runs them on his
  738. website and doesn't release them and sees no reason to ever release
  739. them because he instead he just sells the right to use it through his
  740. website. Then, developer A finds that he's working at a disadvantage.
  741. Every time he makes changes, B can get them, but every time B makes
  742. changes, A cannot get them. So, this way, if A releases under the
  743. Affero GPL then B has to publish his changes because he's providing
  744. public access use.
  745. </p>
  746. <p class="indent">
  747. So that gets rid of the problem.
  748. </p>
  749. <p class="indent">
  750. Now, the issue you brought up is a real issue, but it's a different
  751. kind of issue. You see, we try to make sure that you have freedom to
  752. change and redistribute and run the program as you wish, but that has
  753. to be your copy. Obviously we don't want to give you the freedom to
  754. change someone else's copy because he's the one that's supposed to
  755. have control over his copy, so to have these freedoms you've got to
  756. have your own copy. In fact, this is not just a matter of what the
  757. GPL says, it's a fact of life, it can't be avoided. If you don't have
  758. your own copy of the program, you don't have control. Someone else
  759. can hold your copy for you, you might use a timesharing service, and
  760. </p>
  761. <p>
  762. [The recording is interrupted again]
  763. </p>
  764. <p id="q30-cooperating-with-osi">
  765. Q3: [Unrecorded, but the question was about cooperation
  766. with Open Source Initiative]
  767. </p>
  768. <p class="indent">
  769. The Free Software movement, which I founded in 1983 focuses on
  770. freedom and community, on human rights for software users. "Open
  771. source" was founded in 1998 as a way to stop talking about those
  772. things. To hush them up, to bury them, put them out of people's
  773. sight. So they talk about practical advantages that come from using
  774. the software. Well, I also talk about practical advantages in my
  775. speeches. So here's what I say [Stallman outlines a large circle with
  776. his hands], and here's what they say [Stallman outlines a smaller
  777. circle within the first circle] - except that they go into more depth
  778. on it, and that is useful, y'know. Making the case to businesses that
  779. they will get some practical advantage out of releasing their software
  780. under, usually, a Free Software licence, that's useful, but the point
  781. is it's still a more superficial part of the issue.
  782. </p>
  783. <p class="indent">
  784. So they still say they want to cooperate, and they wish we would
  785. cooperate by forgetting what we consider the most important thing and
  786. joining them in saying only the superficial part. This is the way
  787. Eric Raymond puts it, he's very clever at asking us to abandon the
  788. most important thing and making it sound like he's only being
  789. reasonable.
  790. </p>
  791. <p id="q40-creative-commons">
  792. Q4: [inaudible but he mentions "the creative common licence"]
  793. </p>
  794. <p class="indent">
  795. Richard Stallman: There is none.
  796. </p>
  797. <p>
  798. Q4b: There isn't?
  799. </p>
  800. <p class="indent">
  801. Richard Stallman: There is no such thing as "the Creative Commons
  802. licence".
  803. </p>
  804. <p>
  805. Q4c: Not for software
  806. </p>
  807. <p class="indent">
  808. Richard Stallman: There is no such thing.
  809. </p>
  810. <p>
  811. [laughter]
  812. </p>
  813. <p class="indent">
  814. And the reason I'm responding in this way is that this error is
  815. extremely common
  816. </p>
  817. <p>
  818. [The recording is interrupted again, Stallman goes on to explain that
  819. there are many Creative Commons licences, and some are free and some
  820. are not free, so the "Creative Commons" label does not indicate
  821. anything meaningful in relation to freedom.]
  822. </p>
  823. <p id="q50-osi-responds">
  824. Q5, by Michael Tiemann, President of Open Source Initiative: About a
  825. question you were asked earlier about Eric Raymond, I want to point
  826. out a fact which is that while Eric Raymond was formerly the president
  827. of OSI, he no longer is.
  828. </p>
  829. <p>
  830. Eric, does speak for himself, but less and less for the OSI. I would
  831. also like to clarify that as president of OSI, I have always supported
  832. the GPL as the model licence for developers. The licence is the only
  833. licence I have released work under aside from the LGPL for my own
  834. programming.
  835. </p>
  836. <p>
  837. I recognise your position, which is to say that if I am not talking
  838. about freedom, first and foremost, then I am burying it, but I think
  839. of it myself differently...
  840. </p>
  841. <p class="indent">
  842. Richard Stallman: Well, you might be doing something in between.
  843. There are things in between. When Eric Raymond was president of OSI,
  844. I could perceive his intention to bury talk of freedom very clearly.
  845. And there are others who talk about "open source" who clearly are
  846. trying to bury software freedom. That doesn't imply that everyone who
  847. uses the term... what is true about their use of the term is that it
  848. generally doesn't call attention to freedom very much.
  849. </p>
  850. <p>
  851. Q5b, Michael Tiemann: Well, at this conference I do want to support
  852. that what you are doing is incredibly valuable and I respect that, and
  853. thanks.
  854. </p>
  855. <p class="indent">
  856. Richard Stallman: Thank you.
  857. </p>
  858. <p id="q60-translations">
  859. Q6: [barely audible, the question is about making official
  860. translations of the GPL, and the audience member advocates FSF making
  861. official translations because otherwise the courts will have to make
  862. their own translations which may be done with less care.]
  863. </p>
  864. <p class="indent">
  865. Richard Stallman: The danger with an official translation is, if we
  866. make a mistake, it could be an absolute disaster.
  867. </p>
  868. <p class="indent">
  869. ... and we can't read Portuguese. I can't and Eben Moglen can't. And
  870. even if we could, that doesn't mean that either of us would understand
  871. it as a lawyer. You see, if I work with him when writing GPL version
  872. three in English and I can say "well, here's what I think I want" ask
  873. him to tell me if it really does what I think it does.
  874. </p>
  875. <p class="indent">
  876. Working like that in foreign languages would be much harder, so
  877. producing a translation of the GPL in any other language would be much
  878. harder than what we've already done and I can't trust this. I don't
  879. know who I can trust this to do. I know lawyers in various country
  880. who strongly support the Free Software movement, but to entrust this
  881. to them is a so much... especially when a mistake can destroy things
  882. world wide.
  883. </p>
  884. <p class="indent">
  885. Now, I considered the idea that maybe we'd have translations that
  886. would be limited to one country. At least that puts certain amount of
  887. bounds on the amount of disaster that a mistake would cause.
  888. </p>
  889. <p class="indent">
  890. Another thing I imagined - I just had this idea today - I don't know
  891. whether it's a good idea, we could make translations and say "this
  892. translation is only valid for one year unless it's renewed or replaced
  893. by another more accurate translation". We can't do that with the GPL
  894. itself. If we could change the GPL and force people to switch to a
  895. new version of the GPL, we could take away their freedoms, but the
  896. English version of the GPL does not work that way. Once you get a
  897. program under version X of the GPL, you can always use it under
  898. version X of the GPL forever. So, as long as that remains true, maybe
  899. it's ok if the translations can be revised, so that if there is a bad
  900. mistake in a translation, maybe we can fix it. I don't know if this
  901. is really acceptable. I have this argument which I can argue that
  902. it's acceptable because the real English version will not be
  903. revokable, but I'm not convinced this argument is strong enough. I'll
  904. have to talk to people about it.
  905. </p>
  906. <p>
  907. Q6b: That's easy for you because you live in an English
  908. speaking country, but in Brazil where there is a different language
  909. </p>
  910. <p class="indent">
  911. Richard Stallman: I agree with you but that's not the point. The
  912. reasons why it would be good to have translations are obvious. The
  913. point is: can we get rid of the danger?
  914. </p>
  915. <p class="indent">
  916. So this idea that I had today might be an acceptable way to get rid of
  917. the danger, and if so, we'll have translations.
  918. </p>
  919. <p id="q70-fragments">
  920. Q7: [inaudible]
  921. </p>
  922. <p class="indent">
  923. Richard Stallman: ...and if some program in the set allows
  924. Tivoisation, then they can Tivoise, so if enough important programs
  925. move to GPL version three, Tivoisation will be practically speaking
  926. not feasible.
  927. </p>
  928. <p class="indent">
  929. Of course, this depends on these programs moving to GPL version
  930. three.
  931. </p>
  932. <p>
  933. Q7b: [inaudible]
  934. </p>
  935. <p class="indent">
  936. Richard Stallman: Ah, no, but the point is that those things will be
  937. rather old, and at some point it's just not useful to use them
  938. anymore. They're not a reasonable option because the code has been
  939. released so long
  940. </p>
  941. <p>
  942. [The recording gives up]
  943. </p>
  944. <h2 id="links">Getting more information</h2>
  945. <ul>
  946. <li>FSFE: <a href="gplv3.html">FSF Europe's GPLv3 project page</a></li>
  947. <li>FSF: <a href="">The GPLv3 website</a></li>
  948. <li>FSF: <a href="">The current draft of GPLv3</a></li>
  949. <li><a
  950. href="">A
  951. transcript of the opening session of the first international GPLv3
  952. conference</a> (Eben Moglen discusses the changes made to the
  953. license)</li>
  954. <li><a href="">A transcript of
  955. Richard Stallman's GPLv3 talk from FOSDEM 2006</a></li>
  956. <li><a href="torino-rms-transcript.html">A transcript of
  957. Richard Stallman's GPLv3 talk from Torino</a></li>
  958. <li>FSF: <a href="">The page
  959. on the official GPLv3 wiki listing transcripts and similar texts</a></li>
  960. </ul>
  961. </body>
  962. <timestamp>$Date$ $Author$</timestamp>
  963. </html>
  964. <!--
  965. Local Variables: ***
  966. mode: xml ***
  967. End: ***
  968. -->