Another long one, because I'm trying to get to the bottom of this "scope of license" business.
On 5/20/05, Raul Miller <[EMAIL PROTECTED]> wrote: [snip agreement, about which I am very happy] > I think it's important to note that narrower bounds on the license are > not necessarily less favorable to the offeror. If you're willing to agree > with me on that point, I'm happy. Sure. But I'm not talking about narrower bounds on the set of rights offered with respect to the Program (which is the extant work on which there exist licensable rights). I'm talking about narrower bounds on the definition of "work based on the Program", which leaves room to construe the appropriate permissions for anthologies / buckets-of-works based on the rest of the contract. > > I was attempting to use the phrase "erroneous paraphrase" just as a > > name for candidate C above. As stated more clearly above, the notion > > that it is erroneous doesn't enter into the logic until you try to > > resolve the ambiguity against the offeror. > > And even there that "erroneous" character is contextual. > > I could imagine (for example in a dual-license contract) that the licensee > might prefer the broader interpretation -- for that case, the narrower > interpretation would be "erroneous". As a paraphrase of candidate E, it's erroneous. The grammar, as I read it, doesn't allow it to be anything else. But a licensee is certainly welcome to argue for the presence of an ambiguity there if they have some reason to prefer candidate C. > > The question being asked in "scope of license" analysis is, what > > rights reserved to the copyright holder, as defined in 17 USC, are > > being made available for exercise by the licensee, whatever the return > > consideration may be? In the case of the GPL, the licensed rights > > include copying and distributing the Program itself; modifying, > > adapting, translating or otherwise creating a "work based on the > > Program", and copying and distributing the result; and aggregating a > > "work based on the Program" with other material and copying and > > distributing the result. > > As near as I can tell, those rights are somewhat limited in the context > of modification. > > You seem to be trying to imply that conditions are to be ignored when > construing the scope of the license. But I don't think that's legally > valid -- I've certainly not seen anything that would support that > implication. And, I've seen legal language (for example the concept > of "narrow scope") which implies the opposite. It's possible that you're right; however, the only evidence for this that I have found is internal to the district court's order denying Sun's copyright infringement claims on remand, and the outcome would have been the same either way, so it's not much of a precedent. Long version below. Note that the only consequence would be that some claims might be "upgraded" from breach of contract to copyright infringement (and thus an easier standard for preliminary injunction); all of the rules of construction still apply. My empirical understanding up until now from reading appellate case law (IANAL) is that limitations on how, when, where, and by whom copies (translations, etc.) may be made, and how many and in what form or medium, are all part of the scope of license. Ditto the nature and degree of adaptation, translation, aggregation, etc. Questions of form seem to be particularly subject to judicial construction as to the parties' intent: see Boosey & Hawkes v. Walt Disney ( http://laws.findlaw.com/2nd/969205v2.html ) and op. cit. But the appellate record suggests that nothing other than the exercise of rights reserved to the copyright holder under 17 USC is relevant to this analysis, and obligations of return performance are to be ignored -- and the whole "you must offer source code on demand" bit is indisputably an obligation of return performance. Fail to satisfy it, and you may be in breach of contract, but you can't be successfully sued for copyright infringement unless the contract is first ruled to have been properly terminated. I'm now questioning part of this understanding, based on the Sun v. Microsoft district court's ruling on remand ( http://java.sun.com/lawsuit/012400motionfeds.html ) with regard to the "scope of license" contained in the TLDA. That opinion uses California law to justify reviewing the entire TLDA for evidence of "scope of license". Its ruling against Sun relies on the absence of language in the TLDA "about the license grants being subject to, conditional on, or limited by compliance with the compatibility obligations" in the disputed section. The district court's approach to distinguishing between "contractual covenants" and "restrictions on the license grants" does not appear correct to me, given that all of the appellate judgments I have found that reference SOS v. Payday seem to implicitly use logic similar to mine above. It is worth noting that this is the same district court that was previously overruled for failing to understand and apply this standard, and that they ruled against the licensor (Sun) on remand, so there was no reason for Microsoft to contest this reading. But I'm going to have to look closely at the SOS v. Payday opinion itself to see where the whole "scope of license" bit originated before I regain my previous degree of confidence. In any case, it's not legally correct to say, "X is a limitation on the scope of license because you'd have to do something requiring license before you can physically do X." To get the effect that the FSF professes to want, I think they would have to spell out, in the license text, technical limitations on what collections may and may not be distributed, and say that compiled versions of such-and-such collections must be published together with the source or not at all. Then they're limiting the scope of copyright license -- they dictate what is and isn't a permitted form of modified/anthologized work. That hypothetical "bug-fixed GPL" still couldn't touch distribution under separate cover and run-time combination, per 17 USC 117 and equivalent rules in at least some other jurisdictions. It would still be an offer of contract (and construed accordingly). But the offeror would need only to prove that a work was created and/or distributed outside the scope of license in order to proceed to a copyright infringement claim; and that's at least part of what the FSF is shooting for. Perhaps, if the FSF had had a lawyer who diligently conformed to real-world law, that's how the GPL would have been written. (After all, SOS v. Payday seems to have been a big splash when issued in 1989, and is quite widely cited.) Or perhaps not; competently drafted software contracts (licensing and otherwise) are few and far between, in my experience. > > In another license, the scope might be as narrow as "translate > > alternate pages into French and German and publish the result on > > Post-It (TM) Notes"; but "as long as you pet a cat on alternate > > Tuesdays" isn't part of the scope of license even if it's the first > > clause in the agreement text. "As long as you [do anything]" is > > contract law stuff, even if [do anything] logically requires exercise > > of the rights under copyright that are being offered to you. > > Except... I think you've left out a lot of the narrowness of the GPL. If you're talking about the way that Section 2 is worded, and whether that might bring conditions a, b, c into the "scope of license" calculation, I'm trying to get that straight right now (see above). I've lost my copy of SOS v. Payday and may have to go back to the law library if I can't find an online version. Anyone? > > I think that there's really no question, no matter which path you take > > to construe "aggregation", that it includes both the creation and > > distribution of copyrightable collections (such as distro CDs) and > > uncopyrightable collections (such as Quagga+Net-SNMP+libssl). > > There is some question about whether Quagga+Net SNMP+libssl > is uncopyrightable. No, there isn't. There's no "selection and arrangement" creative expression there. It's silly to say that some third party could obtain a copyright on combining those things and enforce it on the Quagga copyright holders themselves. Copyright doesn't protect ideas, it protects expression; and this is a "doctrine of merger" instance if I ever saw one. > > If you read appellate cases like Sun v. Microsoft and SOS v. Payday, > > you will see that the focus of "scope of license" is on what 17 USC > > rights are on offer with regard to what works in what forms, not on > > what return obligations the licensee may have. > > I think that's largely a reflection of the license in question. Possibly, as discussed above. Additional case law welcome, especially appellate decisions that apply "scope of license" in a non-trivial way, or decisions by courts of fact that do so on remand. [snip agreement] > > And the GPL certainly licenses the right to aggregate a "work > > based on the Program" with other stuff, which amounts to the > > 17 USC definition of "collective works" plus other, uncopyrightable, > > collections. > > That only applies when the work is not based on the Program. > > It seems to me that "based on" is a reference to a concept > from copyright law. Not in any way that matters. There is absolutely no question (IANAL) that "work based on the Program" is a term defined within the GPL's text, in that single sentence from Section 0 with all those quote marks and things. > Paraphrased, the "mere aggregation" clause is saying that > proximity, by itself, isn't enough of an issue to matter. I think Humberto has this right -- it's authorization to create and distribute collective works containing a "work based on the Program", period -- and in any case there are no grounds in the text for construing a magical limit somewhere outside the bounds of a real derivative work but smaller than my living room. I have already commented on why an IP law usage of "aggregation" seems more appropriate to me than recourse to an expert witness, but that's an issue for a court of fact. > > > I think you're still overstating that. For example, consider > > > statutory license, such as that granted by 17 USC 117. > > > > That's not a "copyright license", any more than "fair use" is. It's a > > statement that certain kinds of "copying and adaptation" incidental to > > the use of computer programs are not rights reserved to the copyright > > holder and cannot be used to prove infringement under 17 USC 106. > > "Copyright license" applies only to those rights that it is within the > > copyright holder's power to grant or withhold. > > http://www.answers.com/license > > 1. a. Official or legal permission to do or own a specified thing. Feeble. Get a real dictionary. The word "license", when applied to copyright, has a history as old as copyright itself, and you will find few or no examples (I can find none) of the application to a copyright license, in a courtroom, of any legal theory other than contract. (Not counting copyright law rules about what doesn't require license or can't be licensed at all, of course.) You do understand that the reason 17 USC doesn't spell out how non-exclusive licenses work is that the US federal government has no authority to override contract law, which (as a codification of common law) is reserved to the states? It took explicit authorization in the U. S. Constitution to grant the federal government authority to codify copyright and patent at all. > I think it's clear that we are talking about license here. However, > if it's important to you, I'll agree that it's license in the context > of copyright rather than license to use an otherwise reserved > copyright. If you are trying to use "license" to describe "fair use" and the like, you are deliberately conflating the legal-formal and semi-archaic-informal uses of the word. It's pretty wordplay, and the FSF has a lot of fun with that kind of game, but it's useless in a courtroom. > > I think the rest of your last message either expresses agreement or is > > addressed above, except the question about whether MySQL advanced the > > full panoply of theories under which the distribution of mysqld+Gemini > > might not be permitted under the GPL. I'll hold off on that one > > unless and until I can get the full docket. Suffice it to say that > > the judge was not ignorant of Professor Moglen's claims of > > copyright-based license, over-broad definition of "derivative work", > > and all that. OK? > > Ok. > > I'll simply remind you that that case involved MySQL which was available > under dual-license. Thus it's not really direct evidence about the > scope of the GPL -- only on the scope of the GPL when the proprietary > MySQL license is also a legally valid option. The non-GPL license option to MySQL had no relevance to that case whatsoever. It was not claimed by Progress Software, it is not mentioned in the opinion or in Eben Moglen's affidavit, and as far as I can tell the judge may not even have known that existed. Unless you have some piece of the court record that I don't yet -- in which case, pony up -- this is a lame bit of misdirection. Cheers, - Michael
