Source files of,,,,, and Contribute:

brussels-rms-transcript.en.xhtml 39KB

  1. <?xml version="1.0" encoding="UTF-8" ?>
  2. <html>
  3. <head>
  4. <title>GPLv3 - Transcript of Richard Stallman in Brussels,
  5. Belgium; 2007-04-01</title>
  6. </head>
  7. <body>
  8. <div class="image right">
  9. <a href="gplv3.html"><img src="/graphics/gplv3-logo-red.png" alt="GPLv3 logo" /></a>
  10. </div>
  11. <h1>Transcript of Richard Stallman on GPLv3 in Brussels,
  12. Belgium; 1st of April 2007</h1>
  13. <p>
  14. Transcript released 4th of April 2007. Background and related
  15. documents:
  16. </p>
  17. <ul>
  18. <li><a href="">The listing</a></li>
  19. <li><a href="diff-draft2-draft3.en.html">Side-by-side diff of v2 and v3</a></li>
  20. <li><a href="">Draft 3 (public comment system)</a></li>
  21. <li><a href="">Rationale doc</a></li>
  22. <li><a href="">dd3 FAQ</a></li>
  23. <li><a
  24. href="">The
  25. interview by Sean Daly from that day</a></li>
  26. <li><a href="">The 2nd draft of
  27. LGPLv3 is out</a></li>
  28. <li><a href="">The audio recording</a></li>
  29. </ul>
  30. <p>
  31. Recording thanks to Sean Daly. Transcription
  32. by Ciarán O'Riordan.
  33. Please support work such as this
  34. by <a href="">joining the Fellowship of
  35. FSFE</a>, by
  36. <a href="/help/donate">donating to FSFE</a>, and by encouraging
  37. others to do each. The speech was given in English.</p>
  38. <h2>Table of contents</h2>
  39. <ol>
  40. <li><a href="#free-sw">What is Free Software?</a></li>
  41. <li><a href="#gnu">The GNU project</a></li>
  42. <li><a href="#licences">Free Software licences</a></li>
  43. <li><a href="#internationalisation">Internationalisation</a></li>
  44. <li><a href="#tivoisation">Tivoisation</a></li>
  45. <li><a href="#tivoisation-d3">Tivoisation - the limits in draft 3</a></li>
  46. <li><a href="#eucd">Deflating the EUCD and DMCA</a></li>
  47. <li><a href="#ms-novell">Novell, Microsoft, and patents</a></li>
  48. <li><a href="#termination">Termination</a></li>
  49. <li><a href="#added">Formalising added permissions</a></li>
  50. <li><a href="#bittorrent">BitTorrent</a></li>
  51. <li><a href="#retaliation">Patent retaliation and the Apache licence</a></li>
  52. <li><a href="#bracketed">The bracketed, dated clause</a></li>
  53. </ol>
  54. <h2 id="presentation">The Presentation</h2>
  55. <div class="image">
  56. <img src="rms-20070401.jpeg" alt="Richard Stallman speaking" />
  57. </div>
  58. <p>
  59. <span id="free-sw">(<a href="#menu">go to menu</a>) [Section: What
  60. is Free Software?]</span>
  61. </p>
  62. <p class="indent">
  63. In order to understand the GNU General Public License, first you have
  64. to understand what free software means. Free Software means software
  65. that respects the users' freedom. Software that's not free is
  66. proprietary software. Non-free software, user subjugating software,
  67. is distributed in a social system that keeps the users the divided and
  68. helpless. Divided because every user is forbidden to share the
  69. program with anyone else and helpless because the users don't have the
  70. source code, so they can't change it, they have no control over it,
  71. and they can't even verify independently what it's really doing. It
  72. may have malicious features, quite often it does, and the users may
  73. not even be able to tell what malicious features it has.
  74. </p>
  75. <p class="indent">
  76. Free Software respects the users' freedom. That's why it's important
  77. when we say free software, we're talking about freedom, not price, we
  78. don't mean gratis software. It's not an issue of price at all, price
  79. is merely a detail. A secondary detail, not an ethical issue. To
  80. understand the term Free Software correctly, think of free speech, not
  81. free beer.
  82. </p>
  83. <p class="indent">
  84. Specifically, Free Software means that the user has four essential
  85. freedoms:
  86. </p>
  87. <ul>
  88. <li>
  89. Freedom zero is the freedom to run the program as you wish.
  90. </li>
  91. <li>
  92. Freedom one is the freedom to study the source code and change it so
  93. that the program does what you wish when you run it.
  94. </li>
  95. <li>
  96. Freedom two is the freedom to help your neighbour; that is, the
  97. freedom to distribute exact copies to others, when you wish.
  98. </li>
  99. <li>
  100. Freedom three is the freedom to contribute to your community; that's
  101. the freedom to distribute copies of your modified versions, when you
  102. wish.
  103. </li>
  104. </ul>
  105. <p class="indent">
  106. Each of these freedoms is the freedom to do something if you wish,
  107. when you wish. It's not an obligation, it's not a requirement.
  108. You're not required to run the program. You're not required to study
  109. it or change it. You're not required to distribute exact copies. And
  110. you're not required to distribute modified versions. But, you are
  111. free to do those things, when you wish.
  112. </p>
  113. <p class="indent">
  114. If this were a speech to introduce Free Software I would explain the
  115. reasons why those freedoms are essential, but in this brief
  116. introductions I will have to omit that. Suffice it to say that if a
  117. program respects these four freedoms, then the social system of its
  118. distribution and use is an ethical system. Which means that, in this
  119. regard at least, the software is ethically legitimate. But if one of
  120. these freedoms is substantially missing, then the program is
  121. proprietary software, meaning that the social system of its
  122. distribution and use is unethical and regardless of what the program
  123. does, it shouldn't be distributed in this way.
  124. </p>
  125. <p class="indent">
  126. Developing a non-free program is no contribution to society. It's an
  127. attack on freedom and social solidarity. Such attacks ought to be
  128. discouraged, and I hope to see the day when they no longer occur.
  129. That is the goal of the free software movement; that all computer
  130. users should have these four freedoms, for all the software that they
  131. use. That subjugation of others, through the software that they run,
  132. should no longer occur.
  133. </p>
  134. <p>
  135. <span id="gnu">(<a href="#menu">go to menu</a>) [Section: The GNU
  136. project]</span>
  137. </p>
  138. <p class="indent">
  139. To bring this about, or at least start, in 1984 I began developing an
  140. operating system whose purpose was to be entirely Free Software. It's
  141. name is GNU, which is a recursive acronym for "GNU's Not Unix". This
  142. system is, technically speaking, compatible with Unix, but the most
  143. important thing about it is that it is not Unix. Because Unix is
  144. proprietary software and GNU, because it's not Unix, can be made free
  145. by us, it's developers, and that's the whole point - to give you an
  146. operating system that you can run without ceding your freedom to
  147. anyone.
  148. </p>
  149. <p class="indent">
  150. There are now tens of millions, perhaps a hundred million users of a
  151. variant of the GNU system. Most of them don't know that it's a
  152. variant of the GNU system because they think it's Linux. Linux is
  153. actually one component of the system they use. A component that does
  154. an essential job and that was developed in 1991, and liberated by its
  155. developer in 1992. The GNU system plus Linux, which is the program we
  156. call the kernel, made a complete free operating system in 1992. With
  157. the GNU plus Linux combination it became possible for the first time
  158. to use a PC compatible computer in freedom.
  159. </p>
  160. <p>
  161. <span id="licences">(<a href="#menu">go to menu</a>) [Section: Free
  162. Software licences]</span>
  163. </p>
  164. <p class="indent">
  165. The way that Linux was liberated was by rereleasing it under the GNU
  166. General Public License. The GNU General Public License is the Free
  167. Software licence that I wrote for use in the components of the GNU
  168. system. When we wrote these components, we made them Free Software by
  169. releasing them under the GNU GPL.
  170. </p>
  171. <p class="indent">
  172. But what does that mean? A program is legally considered a literary
  173. work, and it's subject to copyright. Everything that you write is
  174. automatically copyrighted - which is somewhat of an absurd law, but
  175. that's the way it is - and copyright law, by default, forbids copying,
  176. distribution, modification of a work. So, how can any program ever be
  177. free software? The only way is if the copyright holders put on a
  178. notice saying to the users "you have the four freedoms, because we
  179. don't object". That notice is a free software licence. Any notice
  180. which has that effect is a free software licence.
  181. </p>
  182. <p class="indent">
  183. There are actually dozens of different free software licences but the
  184. GNU GPL is the most popular one. It's used for about 70% of all free
  185. software packages. The main difference between the GNU GPL and most
  186. Free Software licences is that the GNU GPL is a copyleft licence.
  187. </p>
  188. <p class="indent">
  189. Copyleft is a technique that I invented, a way of using copyright law
  190. to defend the freedom of other people. In order to be a free software
  191. licence, the licence has to recognise the four freedoms, if you get a
  192. program under any free software licence, you have the four freedoms.
  193. The licence gives you them. And that includes freedom number three,
  194. the freedom to distribute copies of modified versions, and freedom
  195. number two, the freedom to distribute copies without modification, but
  196. when you do that, when you distribute those copies to other people,
  197. will they have the four freedoms?
  198. </p>
  199. <p class="indent">
  200. Maybe yes, maybe no, it depends how you do it. For instance, even in
  201. 1984, I could see there was a danger that you might distribute a
  202. binary of the program without providing the source code, and that
  203. would deny the subsequent recipients freedom number one. The freedom
  204. to study the source code and change it so the program does what they
  205. wish.
  206. </p>
  207. <p class="indent">
  208. So if you could distribute copies without source code, people would
  209. get it from you and they wouldn't have the four freedoms. Not all of
  210. them.
  211. </p>
  212. <p class="indent">
  213. Another thing you might conceivably do is put on additional
  214. restrictions. When you distribute it, perhaps you would put on a
  215. different licence and that licence might be restrictive, it might not
  216. respect the four freedoms for other people. If you could do that,
  217. they who got the program from you would not have Free Software. Many
  218. Free Software licences permit these things, so they respect the four
  219. freedoms but they don't defend the four freedoms. A middle man could
  220. strip off the freedom from the program and then pass on the program
  221. to you and you would get the code but you wouldn't get freedom. You
  222. might just get a binary, or you might get a binary with a licence that
  223. says you're not allowed to copy it. Or you might even get source
  224. code, but it would say you're not allowed to copy it. Various things
  225. might happen which would result in your not getting Free Software.
  226. </p>
  227. <p class="indent">
  228. My goal, in developing the GNU operating system, was specifically to
  229. give freedom to all computer users. If it were possible for middle
  230. men to make the software non-free - if, for instance, by changing it,
  231. that they had an excuse to make the software non-free, the goal would
  232. be defeated. We would fail to provide freedom to the users.
  233. </p>
  234. <p class="indent">
  235. So I designed a Free Software licence specifically to make sure that
  236. everyone who gets any version of the program gets it as free software,
  237. with the four freedoms. In version one of the GNU GPL, which was
  238. published in 1989, it was designed to block the two kinds of attack
  239. that I just described to you.
  240. </p>
  241. <p class="indent">
  242. One being: don't distribute the source code, release only a binary.
  243. Well, GPL version one said a condition of distributing binaries was
  244. that you make source code available also, to the same people. That
  245. blocked that one attack.
  246. </p>
  247. <p class="indent">
  248. The other attack was to put on additional licence terms, to change the
  249. licence. Well, GPL version one said you have permission to distribute
  250. copies but it has to be under this licence, no more and no less. Any
  251. other way, you're not allowed to distribute. So, no adding other
  252. requirements. No removing the requirements that protect peoples'
  253. freedom. Every copy that everyone gets, legally must be distributed
  254. under the same licence: the GNU GPL.
  255. </p>
  256. <p class="indent">
  257. In 1991, I published version two of the GNU GPL, and this blocked
  258. another possible method of attacking users' freedom. A method I
  259. didn't know about in 1989 but which I became aware of afterward. This
  260. would be that a patent holder might sue a redistributor, and in the
  261. settlement the redistributor might agree "when we redistribute this
  262. program, we will put certain restrictions on the users".
  263. </p>
  264. <p class="indent">
  265. What could we do about that? Software patents are weapons that should
  266. not exist. They attack the freedom of computer users, they sabotage
  267. software development, they're only good for the megacorporations.
  268. It's the megacorporations that lobby for them, with the help of their
  269. pet government in Washington.
  270. </p>
  271. <p class="indent">
  272. The fact is, in countries foolish enough to authorise software
  273. patents, a patent holder can stop the distribution of any program that
  274. implements the patented idea, and there's nothing we can do in the
  275. licence of the program to prevent it from being suppressed in this
  276. way. Because the free software licence is just permission to use the
  277. copyright holder's copyrighted work, it has nothing to do with
  278. somebody else's patent.
  279. </p>
  280. <p class="indent">
  281. We can't prevent a patent holder from killing the program, but, we can
  282. hope to save the program from a fate worse than death, which is, to be
  283. made non-free, to be turned into an instrument of subjugation. Better
  284. that our software should cease to exist, at least for 20 years, the
  285. duration of a patent, than it become an instrument for subjugation.
  286. And that, I found a way to do. So in GPL version two, section seven
  287. says that if any conditions are imposed on you or you agree to any
  288. conditions that forbid you to distribute the program with all the
  289. freedoms granted by the GPL, then you can't distribute at all.
  290. </p>
  291. <p class="indent">
  292. So anyone who tries to sue or threaten you, and tries to settle this
  293. in a way that would turn the program into an instrument of
  294. subjugation, discovers that all he has achieved is to kill it.
  295. ...which at least avoids the fate worse than death for the program.
  296. And in fact, this gives our community a certain measure of safety
  297. because very often the patent holder will not find it particularly
  298. advantageous to kill the program. On the other hand, if the patent
  299. could make it effectively non-free and charge people for permission to
  300. run it, that would be advantageous. So if the patent holder could
  301. inflict a fate worse than death, that would be more tempting than
  302. merely to kill it off. So we actually make it less likely that any
  303. bad thing will happen, by standing firm.
  304. </p>
  305. <p class="indent">
  306. GPL version two has been used now for sixteen years. Over the years,
  307. we saw various reasons to change it. Two years ago, we decided it was
  308. time to start seriously working on this and get out a new version. We
  309. realised it would take time. I set aside several months to work on
  310. GPL version three in 2005, so as to prepare the first draft that was
  311. released in January 2006.
  312. </p>
  313. <p class="indent">
  314. We have just published the third discussion draft and we've decided to
  315. wait for sixty days and then prepare what we hope will be the final
  316. draft for the last bit of comments before the final draft.
  317. </p>
  318. <p class="indent">
  319. So what are the changes we've made?
  320. </p>
  321. <p class="indent">
  322. People often ask for a simple summary of these changes, but there
  323. can't be one, and the reason is that the changes are all in specifics
  324. because the basic idea is the same: defend freedom for all users.
  325. That will never change. So all the changes are in specific areas and
  326. details. Let me tell you about the most important ones. One of them
  327. is internationalisation.
  328. </p>
  329. <p>
  330. <span id="internationalisation">(<a href="#menu">go to menu</a>) [Section: Internationalisation]</span>
  331. </p>
  332. <p class="indent">
  333. We have changed the language to avoid certain terms whose meanings
  334. vary more than necessary between countries. Terms like "distribute".
  335. In GPL version three, we will not use the word "distribute". We've
  336. formulated two new terms, and defined them in ways that get results
  337. that are as uniform worldwide as possible. The first term is
  338. "propagate". That means basically copying, or anything else like it,
  339. that requires permission under copyright law.
  340. </p>
  341. <p class="indent">
  342. Anything except just to modify one copy or run it is to propagate the
  343. work. And then the second term is "convey". "Convey" basically
  344. replaces "distribute to others" but it's not defined in terms of the
  345. word "distribute", it's defined in terms of propagate in such a way
  346. that others may get copies. So we avoid the word "distribution", and
  347. then, in the bulk of the licence, we set terms for propagating the
  348. work and terms for conveying the work. And this way the resulting
  349. conditions are as uniform as possible worldwide. Now, they will never
  350. be exactly uniform, as long as some details of copyright law vary.
  351. </p>
  352. <p>
  353. <span id="tivoisation">(<a href="#menu">go to menu</a>) [Section:
  354. Tivoisation]</span>
  355. </p>
  356. <p class="indent">
  357. Another major change consists of a new form attacking a user's freedom
  358. that we've seen in the past few years. It's called tivoisation. This
  359. is the practice of designing a machine so that if the user installs a
  360. modified version of a program, the machine refuses to run it.
  361. </p>
  362. <p class="indent">
  363. It's named after the first product I heard of which did this, which is
  364. called the Tivo. The Tivo contains Free Software released under
  365. version two, and they provide the source code, so the user of the Tivo
  366. can modify the program and compile it, and install the modified
  367. version in his machine, whereupon the machine won't run at all because
  368. it notices that this is a modified version. This means that in some
  369. nominal sense, the user has freedom number one, but really, in
  370. practical terms it has been taken away, it has been turned into a
  371. sham. And this happens systematically, and it makes a systematic
  372. threat to users' freedom. So we've decided to block this, and the way
  373. we block it is as part of the conditions for distributing binaries, we
  374. say that if you distribute in, or for use in, a certain product, then
  375. you must provide whatever the user needs in order to install her own
  376. modified version and have them function the same way, unless her
  377. changes in the code change the function. But the point is that it's
  378. not just the user has to be able to install it and has to be able to
  379. run, but it has to be able to do the same job, despite having been
  380. modified.
  381. </p>
  382. <p class="indent">
  383. If the mere fact that the program is modified is detected and prevents
  384. the program from doing the same job, that's also a violation, that
  385. means that the installation information is insufficient, and that is
  386. what would happen under Microsoft's scheme that it used to call
  387. Palladium. The idea was that files would be encrypted so that only a
  388. particular program could ever read them. If you got the source code
  389. of that program and you modified it and you compiled it, the checksum
  390. of your version would be different. So your version would be unable
  391. to read the files encrypted for the original version. And the
  392. original version would be unable to read any files encrypted for your
  393. version except that there aren't any. They would pretend that this is
  394. just a symmetrical situation and they haven't done anything to shaft
  395. you, but our conditions say they have to give you what it takes so
  396. that you can make your modified version do the same job, and the mere
  397. fact that it's been modified may make it fail to do that job.
  398. </p>
  399. <p>
  400. <span id="tivoisation-d3">(<a href="#menu">go to menu</a>) [Section:
  401. Tivoisation - the limits in draft 3]</span>
  402. </p>
  403. <p class="indent">
  404. In version three we've decided to limit this requirement somewhat.
  405. It's limited to a category that we call "user products", which include
  406. consumer products and anything that's going to be built into a house.
  407. This is to exclude products that are made specifically for businesses
  408. and are not normally used by consumers at all.
  409. </p>
  410. <p class="indent">
  411. The reason we did this is because the big danger is in the area of
  412. user products, and this way we were able to get at least a partial
  413. agreement of businesses that at least we hope will be able to help us
  414. actually put an end to this threat.
  415. </p>
  416. <p class="indent">
  417. The products that do tivoisation typically do it because there is some
  418. other malicious feature in the product. For instance, the Tivo spies
  419. on the user and it implements Digital Restrictions Management. That's
  420. why they want to stop you from changing the program. Digital
  421. Restrictions Management is unethical. It shouldn't ever be
  422. tolerated. And spying on the user is unethical also, but we have not
  423. put anything in the GPL to say that you can't make the software spy on
  424. the user or that it has to be able to copy things. There is no
  425. restriction at all on what functionality your modified version of the
  426. program can have.
  427. </p>
  428. <p class="indent">
  429. The anti-tivoisation requirement doesn't limit you in that regard.
  430. You can implement the nastiest features you can think of and release a
  431. modified version with those nasty features. We're only trying to make
  432. sure that the users who get the program from you will have the freedom
  433. to remove those nasty features and fill in whatever useful features
  434. you took out. They should be free to modify it just as you were free
  435. to modify it. That's what the anti-tivoisation provisions do.
  436. </p>
  437. <p>
  438. <span id="eucd">(<a href="#menu">go to menu</a>) [Section:
  439. Deflating the EUCD and DMCA]</span>
  440. </p>
  441. <p class="indent">
  442. We also have provisions in section three to try to defeat unjust laws
  443. such as the Digital Millennium Copyright Act in the US and the EU
  444. Copyright Directive and any similar laws that implement that WIPO
  445. copyright directive - a treaty which no government representing its
  446. citizens would ever sign.
  447. </p>
  448. <p class="indent">
  449. It actually has two paragraphs. One which is aimed at the DMCA and
  450. similar laws, and another which is aimed at the EU Copyright
  451. Directive, but they both achieve the same goal. The goal they achieve
  452. is, if somebody uses GPL covered software as part of a scheme to
  453. encode or decode works, then he has waived any possibility of claiming
  454. that some other software which decodes those works is a violation of
  455. these laws. This is simply a way of trying to prevent people being
  456. forbidden to distributed software, perhaps modified versions of the
  457. same software or perhaps not but either way, we're making sure that
  458. these laws can't be used to suppress the modified versions that people
  459. may wish to distribute under the GPL.
  460. </p>
  461. <p>
  462. [26:30]
  463. </p>
  464. <p>
  465. <span id="ms-novell">(<a href="#menu">go to menu</a>) [Section:
  466. Novell, Microsoft, and patents]</span>
  467. </p>
  468. <p class="indent">
  469. There's another form of attack on Free Software's freedoms that we
  470. found out about last November with the Novell-Microsoft deal. What
  471. happened was that Novell made a deal with Microsoft where Novell pays
  472. for distributing copies of GPL-covered software and Microsoft gives
  473. the customers of Novell a very limited patent licence which is
  474. conditional on their not exercising many of the rights that the GPL
  475. gives them.
  476. </p>
  477. <p class="indent">
  478. This is a big threat, so we've gone at it from two directions. One is
  479. aimed at Microsoft's role in that deal. We say: if you make a deal to
  480. procure someone else's distribution of a program under GPL version
  481. three and you provide any sort of patent licence to anybody in
  482. connection with that, then it extends to anybody who gets it. So if
  483. Novell were to distribute software under GPL version three under this
  484. deal, then this affects Microsoft because they're procuring
  485. distribution through this deal.
  486. </p>
  487. <p class="indent">
  488. The other paragraph, and these are both in section [11], is aimed at
  489. the Novell side in the deal, which is, it says that if you distribute
  490. the program under an arrangement you made with someone else, to gain
  491. promises of patent safety for your customers in a discriminatory way,
  492. then you're violating the licence and you lose your right to
  493. distribute.
  494. </p>
  495. <p class="indent">
  496. This actually has a few more conditions because we were trying to
  497. avoid covering certain other things, for instance, consider a patent
  498. parasite, one of those companies that has only one business which is
  499. to go around threatening people with patent law suits and making them
  500. pay. When this happens, the businesses that are attacked often have
  501. no choice but to pay them off. We don't want to put them in a
  502. position of being GPL violators as a result. So we put in a
  503. condition: "this paragraph applies only if the patent holder makes a
  504. business of distributing software". Patent parasites don't. As a
  505. result, the victim of the patent parasites is not put in violation by
  506. this paragraph.
  507. </p>
  508. <p class="indent">
  509. We also did something to exclude blanket cross-licences and various
  510. other practices which are not threatening in the same way.
  511. </p>
  512. <p>
  513. [29:38]
  514. </p>
  515. <p class="indent">
  516. This section [11], which these paragraphs are included in, does somewhat
  517. more. We decided in GPL version three that distributors should
  518. explicitly give patent licences when they distribute the software. We
  519. do this in two ways. Those that contribute to the development of the
  520. program, that make any changes at all, give an affirmative patent
  521. licence to all downstream recipients. Those who just pass along
  522. copies, without changing it, they are bound not to sue any of those
  523. downstream recipients or their licence terminates.
  524. </p>
  525. <p class="indent">
  526. This is a sort of compromise that is designed to encourage lots of
  527. companies to participate in the distribution, and thus, we hope, make
  528. our community somewhat safer.
  529. </p>
  530. <p class="indent">
  531. In previous versions of the GPL, we didn't have any kind of explicit
  532. patent licence. We took for granted that if you distribute someone a
  533. program under the GPL and you say that he's allowed to do certain
  534. things, that you can't sue him for doing those things. And in the US,
  535. that's true, within certain bounds, but it's not true worldwide. So
  536. this is another aspect of internationalisation - getting the same
  537. result around the World.
  538. </p>
  539. <p>
  540. <span id="termination">(<a href="#menu">go to menu</a>) [Section:
  541. Termination]</span>
  542. </p>
  543. <p class="indent">
  544. Another change that we have made is in termination. With GPL version
  545. two, if you violate the licence, you have lost your rights to do
  546. anything with the program. And the only way to get them back is to go
  547. to all the copyright holders and beg for forgiveness. Which in most
  548. cases they will give you, if you show that you are going to follow the
  549. licence in the future, but their may be lots and lots of copyright
  550. holders.
  551. </p>
  552. <p class="indent">
  553. This could be a gigantic amount of work. It could be totally
  554. unfeasible. With GPL version three, your licence is not terminated
  555. automatically. Rather, the copyright holder can write to you can say
  556. "I'm putting you on notice", and after that point, the copyright
  557. holder can terminate your licence, if he wants to. But he doesn't
  558. have to. He could just forgive you instead if he sees that you're
  559. going to comply.
  560. </p>
  561. <p class="indent">
  562. The result is, suppose you do something that violates the licence, and
  563. you distribute an entire GNU/Linux distribution with thousands of
  564. programs in it, and you fail to follow the requirements. And suppose
  565. ten copyright holders contact you and say you violated the licence,
  566. and you say "Oops, I better do this right" and then sixty days go by
  567. and maybe, say, five more copyright holders will contact you within
  568. those sixty days then those are the only ones that can terminate your
  569. licence. But the idea is that you show those fifteen developers that
  570. you are complying now and you promise to keep doing so. And they say
  571. "Ok, we forgive you".
  572. </p>
  573. <p class="indent">
  574. You don't have to find all the other thousands of copyright holders
  575. and ask them for forgiveness. So, as long as you keep on with your
  576. violating practice, all the copyright holders still can put you on
  577. notice when the word gets around to them, but when you stop violating
  578. the licence, then they have sixty more days when they can still
  579. complain, and then they can't complain anymore. So only the ones who
  580. complain within that time are the ones you have to deal with.
  581. </p>
  582. <p>
  583. [34:00]
  584. </p>
  585. <p class="indent">
  586. We've also put in an automatic cure period for first time violators.
  587. If you violate the licence, and it's the first time you violated the
  588. licence for a certain copyright holder and you correct it within 30
  589. days, things are automatically restored. You're ok. But that doesn't
  590. apply the second time.
  591. </p>
  592. <p class="indent">
  593. We're trying to make it easy for accidental violators, those who are
  594. willing to clean up their act, while still facilitating enforcement
  595. against those who don't clean up their act.
  596. </p>
  597. <p>
  598. [34:55]
  599. </p>
  600. <p class="indent">
  601. Another change we've made, has to do... ...what we've done that's new
  602. in section 10 is we've dealt with kinds of situations that arise when
  603. a company subdivides or sells a division to another company, what
  604. happens then? We want to make sure this is clear. It shouldn't be
  605. very controversial, but at least now it's well defined.
  606. </p>
  607. <p>
  608. <span id="added">(<a href="#menu">go to menu</a>) [Section:
  609. Formalising added permissions]</span>
  610. </p>
  611. <p class="indent">
  612. Another thing that we've done is that we've formalised what it means
  613. to give added permission. There's a practice used with GPL version
  614. two which is common enough which is that you say "This is released
  615. under version 2 of the GPL or later, and in addition we say you can do
  616. X, Y, or Z". Added special permission as an exception.
  617. </p>
  618. <p class="indent">
  619. And this is perfectly fine since what it means to release a program
  620. under a free software licence is simply you give the permission stated
  621. in the licence and you can also give other permissions.
  622. </p>
  623. <p class="indent">
  624. But in order to make it clearer what this means, we've formalised it.
  625. So we've explained that you can give additional permissions for part
  626. of the code that you introduce into the GPL covered program, where you
  627. have the right to give those permission. Because the program is
  628. available under the GPL terms, when somebody modifies it, she can take
  629. off the added permissions. The GPL gives her permission to release
  630. her version under the GPL. She doesn't have to give any added
  631. permissions for her version. But she could also pass on those added
  632. permissions because she got those added permissions to you and when
  633. she modifies those parts, she can also give those same added
  634. permissions for those changes when she wants.
  635. </p>
  636. <p class="indent">
  637. So when a program is released under GPL plus added permissions,
  638. everyone who passes it along has a choice. Preserve the added
  639. permissions or remove them. Section 7 explains this. It also
  640. explains that certain kinds of explicit requirements found in other
  641. free software licences are, under our understanding, compatible with
  642. the GPL and that's in fact our practice with GPL version two but
  643. again, it's good to make it explicit.
  644. </p>
  645. <p class="indent">
  646. And they are basically trivial ones. Like, this copyright notice
  647. can't be removed. Or different statements of warranty disclaimers.
  648. They're not substantive conditions.
  649. </p>
  650. <p class="indent">
  651. In draft 2, this section, section 7, was more complicated. We set it
  652. up so that there were a few specific kinds of substantive requirements
  653. that you could add. And so we'd need a lot more complicated mechanism
  654. to keep track of them. A lot of people didn't like the complexity of
  655. this. So we got rid of that and we dealt with those issues in other
  656. ways. For instance, there's a section in the latest draft which says
  657. when a program is released under GPL version three, you can link it
  658. with code that is under the Affero GPL. In a previous draft, we said
  659. that the added requirement in the Affero GPL was one of those
  660. requirements that you could add to parts of a program that you
  661. contribute. Now, it isn't. Now, the GPL is purely and simply the GPL
  662. and it only allows linking with other code that was released by its
  663. developers under the Affero GPL.
  664. </p>
  665. <p class="indent">
  666. The Affero GPL, for those who don't know, is basically GNU GPL version
  667. two, plus one additional requirement which says if you use the
  668. software publicly on a website, then you've got to give the users of
  669. the site a way to download the source code you're running. The source
  670. code corresponding to the version they are talking through.
  671. </p>
  672. <p class="indent">
  673. The basic idea is that for those developing software meant for use in
  674. this way, on servers, they want those who publicly deploy modified
  675. versions to contribute their code back to the community.
  676. </p>
  677. <p class="indent">
  678. When we advised Affero on releasing this licence, we had the intention
  679. of making a future GPL version compatible with it and this is how
  680. we're going to achieve it.
  681. </p>
  682. <p>
  683. <span id="bittorrent">(<a href="#menu">go to menu</a>) [Section: BitTorrent]</span>
  684. </p>
  685. <p class="indent">
  686. Another change that we've made is for support of things like
  687. BitTorrent. BitTorrent is peculiar because when you receive copies,
  688. you automatically end up distributing copies to other people and you
  689. don't even know it. This is so bizarre, that I never imagined any
  690. such thing back in 1991. And in fact, distributing a program under
  691. GPL version two using BitTorrent does things that violate the GPL.
  692. </p>
  693. <p class="indent">
  694. All these people are redistributing copies and they don't know it, and
  695. they are failing to carry out their responsibilities under the GPL.
  696. So we changed it so they won't have any problem. Obviously,
  697. distribution using BitTorrent should be OK.
  698. </p>
  699. <p>
  700. [42:00]
  701. </p>
  702. <p class="indent">
  703. The problem has to do with binaries. The GPL's idea is that if you
  704. distribute binaries, you've got to make source available. But if
  705. somebody's participating in a torrent for the binaries, well, he may
  706. not know where the source code is, and he didn't set it up anyway, and
  707. we don't want to make him responsible for dealing with the source
  708. code, as a mere user trying to get a copy through the torrent.
  709. </p>
  710. <p class="indent">
  711. Another change, that will make things simpler for a lot of people, is
  712. that you will now be permitted to distribute physical copies of
  713. binaries and put the source code on a server for people to get. The
  714. reason that wasn't allowed in the past was that for a lot of users in
  715. the past, downloading the corresponding source code might have been
  716. prohibitively slow and expensive.
  717. </p>
  718. <p class="indent">
  719. In a lot of parts of the World, certainly, sixteen years ago, people
  720. wouldn't have had broadband. All they would have had is dial-up
  721. connections. You could give somebody a CD of an entire system, and if
  722. you could say to him "to get the source code, just pull it down
  723. through your phone line", that would have been ridiculous. So we said
  724. you have to offer them to mail them a source CD, or other source
  725. medium, which would have been a lot cheaper and much more feasible
  726. than getting the whole thing through a phone line. But nowadays in
  727. fact there are services which which download anything for you and mail
  728. it to you on a CD anywhere in the World and they don't charge very
  729. much. So the problem we were trying to protect against, which is that
  730. lots of users would find it unfeasible to get the source, isn't a
  731. problem anymore. These services charge less than a mail order service
  732. required by GPL version two is likely to cost. So we can just say, we
  733. presume people will use those services if they need them and we don't
  734. have to put that requirement on every distributor.
  735. </p>
  736. <p>
  737. [44:44]
  738. </p>
  739. <p class="indent">
  740. I think this is basically it. There are many little details, and I
  741. certainly have forgotten about some of them right now.
  742. </p>
  743. <p class="indent">
  744. We took out the paragraph in GPL version two about putting on a
  745. geographical limit saying that you can't distribute the program in a
  746. certain country if something like patents in that country have
  747. effectively made the program non-free. We think we've defended
  748. against that pretty well and nobody ever used that anyway.
  749. </p>
  750. <p class="indent">
  751. So, in about 90 days...
  752. </p>
  753. <p>
  754. [interruption for tape change]
  755. </p>
  756. <p>
  757. <span id="retaliation">(<a href="#menu">go to menu</a>) [Section: Patent retaliation and the Apache licence]</span>
  758. </p>
  759. <p class="indent">
  760. Previous drafts had a couple of provisions regarding retaliation
  761. against aggression using software patents. First of all, there was a
  762. very limited kind of retaliation directly in the GPL itself which said
  763. that if somebody started using a modified version of a program and
  764. then sued somebody else for patent infringement for making similar
  765. improvements in his own version, that this would cause termination of
  766. a specific right, namely the right to continue modifying the program.
  767. </p>
  768. <p class="indent">
  769. We took that out because it didn't seem like it would be terribly
  770. effective. We also, in section 7, where it allowed putting on
  771. additional requirements, one of them was a stronger kind of patent
  772. retaliation and there were two kinds that were allowed. The latest
  773. draft essentially does one of them itself, and so we don't need to
  774. make that another kind of requirement you can add. And the other
  775. kind, it looked like nobody's actually doing it, so just for
  776. simplicity we took out that option.
  777. </p>
  778. <p class="indent">
  779. The reason that we allowed people to add these two kinds of patent
  780. retaliation clauses was for compatibility, and one of them is used by
  781. the Apache licence. We hoped to make GPL version three compatible
  782. with the Apache licence and we thought we had. We were focusing on
  783. the Apache licence's patent retaliation clause, and in previous drafts
  784. we said it was ok to add that requirement to the code you contribute.
  785. Now we just have that requirement, later on in the GPL, so it doesn't
  786. even need to be added.
  787. </p>
  788. <p class="indent">
  789. So we are compatible with that clause in the Apache licence but a few
  790. months ago we noticed that there was another clause in the Apache
  791. licence requiring indemnity in certain cases, and there's no way we
  792. can be compatible with that. So we're not going to achieve that goal
  793. of making GPL version three compatible with the existing Apache
  794. licence. I regret that.
  795. </p>
  796. <p>
  797. <span id="bracketed">(<a href="#menu">go to menu</a>) [Section: The
  798. bracketed, dated clause]</span>
  799. </p>
  800. <p class="indent">
  801. There's one bracketed clause in the current draft and that concerns a
  802. cut-off date in the paragraph that forbids making deals to get patent
  803. safety for your customers alone. We haven't decided whether that will
  804. apply to deals that have already been signed, or only deals that are
  805. signed after the release of this draft. And that mainly depends on
  806. how good a job we find we have done drawing a line between these
  807. pernicious deals and other kinds of deals that we don't want to
  808. cover. We've come up with some criteria that seemed to do the job,
  809. and we hope to find out whether we've really done the job.
  810. </p>
  811. <p class="indent">
  812. We think that we have addressed the specific deal between Novell and
  813. Microsoft sufficiently well with the other requirement, the one that
  814. is aimed at Microsoft's role in that deal. Even if we post-date the
  815. effect of the second paragraph, the one that aims at the Novell role,
  816. such that it doesn't apply to Novell, we've dealt with that particular
  817. deal. But if we have drawn the lines well enough, or if we can, then
  818. we'll make it apply both to existing and to new deals. It's a
  819. decision we haven't made yet.
  820. </p>
  821. <p>
  822. [52:00, end]
  823. </p>
  824. <p>
  825. [Note: for more information on GPLv3, see <a href="gplv3.html">FSFE's
  826. GPLv3 page</a> and the background and related documents linked at the
  827. top of this page]
  828. </p>
  829. </body>
  830. </html>