(Before going further, you should (at the minimum) read the disclaimer on Monday’s post and (ideally) make sure you’ve read all of the past two posts. (first, second.))
Q: Companies hate this! The sky is falling if the GPL is adopted!
A: Find your calm place. There is no doubt that the new license will create new uncertainty, which companies hate, and that the license seeks to place rights more firmly in the hands of users, which many companies (though not all) hate. So companies have reason to be nervous.
But the license was created with input from an extensive range of companies who both produce and use free software-based products, so I think you’ll find that it really isn’t that bad for companies who are playing by the rules- that is, for 95% of companies, 95% of the time. Remember, this license is (for most purposes) much more evolutionary than revolutionary, so it gives the same rights, sometimes clothed in different languages, in most circumstances. Most companies will not be affected by the tivo-ization and DRM clauses, and will actually appreciate the additional protections of the new license. Especially once companies have had time to digest the license, and to realize that none of the early adopters have turned into a newt, most companies should be fine with the license most of the time.
Q: OK. A little calmer now. My company makes products based on open source components. What are my new risks?
A: Have you burned the boats? If you’ve truly burned the boats, you’re fine- the patent provisions and installation provisions of the license only help you get closer to your customers and differentiate you from competitors and free-riders.
On the other hand, if you or your customers are still relying on some form of proprietary lockin, the picture is much less clear. You have new responsibilities that you may not have previously had. In particular, if you contribute to GPL v3 projects, you have to open up your patent portfolio to those projects and their users, and if you distribute GPL v3 code, you have to give your users the ability to reinstall code that they modify with very few restrictions. These may not be dealbreakers, but if your business model depends on users who are somehow kept under lock and key, you will have to deeply examine the new license, and may have to reconsider basing your business on GPL’d code.
Q: Do I get any benefits to go with those risks?
A: Like everyone else, you should get greater certainty about patents, and greater legal certainty in general as a result of the improved language. The same clauses that protect your customers from you will also protect you from competitors who seek to use some sort of patent or device lockin to compete unfairly against you, so even they aren’t all bad.
You’ll also get more control over trademark as part of the compatibility with the APL- you can explicitly ban use of the trademark by derivatives, among other things. Most community projects are not overly concerned by trademark, but this is clearly important to companies who are trying to build a brand, and more explicitly spelled out now as an optional addition to the license.
You also get the ability to transmit code to contractors without being considered to have ‘conveyed’ the code, allowing you to use consultants and outsourcers to modify your private code forks without accidentally freeing your work. If you sell non-consumer products (IBM S390 servers, for example, or medical devices which might be under government regulation) you are now even allowed to prevent modification/reinstallation by those who you sold servers to.
If you’re using patents or hardware to lock in your customers, none of these things are going to be much solace. But for everyone else, they should be useful compensations which make up for the short-term uncertainty.
Q: Why is Sun seemingly so excited about this? Should I share their enthusiasm?
A: Maybe. Sun is in a fairly unique situation. They badly want to win developer mindshare away from Linux, so becoming More FSF-Free than Linux is a big PR win for them- more so than it might be for most companies who aren’t competing with one of the poster children of open source and with the ghosts of their own proprietary past. In addition, some of the license features which are problematic for many people may be a benefit to Sun. For example, most people will look at v3’s incompatibility with v2 and think ‘what a pain’. Sun, in contrast, might think of that as a benefit, since that would continue to prevent Sun code from being used in the Linux kernel- who have to be seen as a significant competitor for them. Similarly, the new clauses which make embedding and DRM difficult may be seen as a detriment by many developers, but they play very well into Sun’s business model for Java, which is based on licensing into the embedded market. Anything that encumbers Free Java helps drive embedded market revenue for them, so GPL v3 may be preferable to v2 from that perspective.
All that said, we should probably give Sun some of the benefit of the doubt at this point- they’ve freed a lot of code, and by all accounts are trying hard to become a more open company. If they really wanted incompatibility and embedded revenue, they could just stick to the CDDL or a proprietary license. GPL would be a huge step for them, and is a step that the Free Software community should welcome with open arms, even if their motives are not 100% pure.
Q: Should Novell be quaking in their boots?
A: Hard to say. With their recent joint announcement with the EFF they appear to have started the process of really burning the boats. A company fully committed to competing on the basis of quality and service, and willing to say no to customers who ask for bad things, should have very little to fear from GPL v3, and it seems like (belatedly) Novell may finally be headed in that direction.
Q: But Novell are on the hook for the MS deal, right?
A: Yes and no. While a deal like the Microsoft-Novell deal, which induces a third party to grant a non-assert agreement, older deals got grandfathered in- the FSF appears to believe that it is better to have Novell tentatively on our side than actively using their large warchest to fork everything that goes v3. But it seems unlikely that MS can continue to uphold their part of the deal without licensing their entire patent portfolio to some key technologies like gcc and libc. So MS has a strong incentive to get the deal voided, and even if not, it will expire in the not too far future. (Remember, they negotiated a license which only covers five years of twenty year patents.) If they do renegotiate, the new deal will almost certainly look like the Xandros and Linspire deal, where MS offers indemnification to individual users instead of to all users. This will be acceptable under the new GPL language.
Q: So can we break out the champagne? Is the MS patent threat over?
A: Definitely not. MS can, of course, cease giving out coupons, and try to induce Novell to adjust the deal. Moreover, because of the complexity of the patent language, there is almost certainly another loophole for them to exploit, though it is hard to see what it migh be at this time. They could, for example, offer indemnification directly to customers- which appears to be the route announced in the Xandros and Linspire deals. Even if they do continue the relationship with Novell, lots of very big chunks of code that aren’t GPL v3, like the kernel and Open Office, will not be protected. So we’re still dead in their sights, no matter what.
Q: What does MS think, anyway?
A: They are gearing up their FUD machine, of course. Expect the old reliables: GPL is anti-American, GPL will steal your rights, GPL will eat your babies. Expect some new favorites too- lots of talk about how MS ‘builds bridges’, and about how the GPL hates interoperability. Of course, no mention will be made of the tolls that MS wants to charge on those bridges, or of the tons of GPL licensed code which builds bridges both to open standards and (despite the difficulties) to MS ‘standards.’
Q: What about the little guys in the embedded/consulting space? Opened Hand, Fluendo, etc.?
A: Each will have their own sets of problems and evaluations of the license. The new LGPL should allow Fluendo to continue down the mixed proprietary/open path they have chosen. Perhaps a more difficult question is what this will mean for those who are thriving in the embedded phone/handheld device space, like Opened Hand, OpenMoko, etc. The new license would appear to require that they allow modification, even for things like phones- which may be very problematic for networks (despite the out for network service providers included in the license, which allows you to withdraw services from modified devices if the modifications cause problems.) It will be interesting to see what openmoko’s lawyers say- they appear committed to resolving the problems between a locked down network and an open device, but this may be too much even for them to overcome.
Q: What about Google and other SaaS companies?
A: Nothing here that impacts them; they can continue business as usual. There is a new Affero GPL, which closes the SaaS loophole, but the FSF has made it clear that for now they will use moral persuasion rather than licensing to encourage code to migrate out from behind the closed Google doors. You might call this ‘picking your battles.’
Q: what is the bottom line for companies and the GPL?
A: For most? Simple. If you just use GPL-licensed code (for example, as a server or desktop), you have no worries. If you contribute code to GPL projects, you’re OK, as long as you feel comfortable granting your patents to the version of the project you’re contributing to. If you distribute GPL code as part of a device, you’re also OK, as long as you allow users to modify their devices. These categories cover the vast majority of corporate participants in the existing GPL community.
Despite that, this won’t be a license for everyone; some companies may prefer to stick with v2 or perhaps even use other alternatives instead. Like the choice for developers, the choice will require a case-by-case analysis of the costs and benefits. I do look forward to the day when we have a new license, representing a new and appealing business model which makes these kinds of choices easy, but for right now we’re stuck doing the hard and sometimes unpleasant work of protecting more user rights, and that means compromise and muddiness, not clear wins. Sorry about that…
[see all parts: part 1, part 2, part 3, part 4.]