Originally Published: Monday, 10 July 2000 Author: Bruce Perens
Published to: columnists/Bruce Perens Page: 1/1 - [Printable]

Gated Communities

"Gated communities" are a non-Open Source form of shared software development. The idea is that developers all enter into a license, and are able to exchange source code modifications with other licensed developers, but not with the general public.

   Page 1 of 1  

"Gated communities" are a non-Open Source form of shared software development. The idea is that developers all enter into a license, and are able to exchange source code modifications with other licensed developers, but not with the general public. Gated communities are in the news of late because one is being used for the "Inferno" operating system, and because the concept is being promoted by Tim O'Reilly of O'Reilly Press.

A "gated community" license generally allows some form of binary distribution, sometimes with a royalty payment to the Initial Contributor required. Gated communities are un-equal: the Initial Contributor of the program has more rights than anyone else in the community. In general the Initial Contributor has the right to distribute your modifications under any license it pleases, the Initial Contributor may demand royalties for sale of the software, and there may be other special terms to the license that go outside of what you'd expect in an Open Source license. O'Reilly actually calls them "Gated Open Source Communities," but this is incorrect: none of the gated community licenses are compliant with the Open Source Definition. Open Source allows anyone to participate.

Gated communities are not new, though that term surfaced only recently. AT&T operated a gated community for Unix in the 70s and 80s, and it was continued by the subsequent owners of Unix. I participated in that community starting in 1981 or so, when I first started with BSD on the VAX 780. At that time, BSD was derived from ATT Unix 32V, and an AT&T license was mandatory for access to the source code. AT&T gave these licenses to schools at a low price, but for-profit corporations had to pay more. Once you had a license, you could exchange modifications with any other licensee, and you could produce binary copies of Unix and sell them as long as you paid a royalty to AT&T. But outsiders weren't allowed -- in fact, Kevin Mitnick was prosecuted for downloading the Unix source code without a license, with the value of the stolen code in the case declared at over $200 million.

It's interesting to note that AT&T's gated community strategy wasn't very successful in promoting Unix. Although the system gained a loyal following in academia and made a fortune for workstation manufacturers and SCO, just a few years ago Unix programmers were resigning themselves to the fate of having to work on Microsoft Windows at their future job, as it seemed that Unix was about to die out. Then Linux came along and everything changed.

Although Linux is now becoming superior to older Unix systems, it started out as a Unix clone with only one crucial difference: its licensing. Linux discarded the "Legally-Restricted Elite Group" method of development, replacing the gated community with a seeming free-for-all in which anyone is allowed to participate but the quality of your code ultimately determines whether or not it will be accepted. This opening of the gates propelled Linux into the mainstream just when its work-alike Unix had been given up for dead.

One problem with AT&T's Unix gated community was that the prerequisites for admission were simply too high. One either had to be a school or non-profit research organization, or had to be willing to pay tens of thousands of dollars for corporate source licensing. At one time, I paid $50k for the right to sell binary Unix. In contrast, an AT&T licensee now operates a gated community for AT&T's "Inferno" operating system with an admission cost of only $300, and offers those licensees the right to distribute binaries without a royalty.

Another problem with the Unix gated community was that AT&T wasn't really interested in modifications from outside. They chose to ignore the fine work being done under Army sponsorship by the BSD group at Berkeley and developed Unix System III and then System V on their own, never really reaching the quality of Berkeley's system.

O'Reilly contends, erroneously in my opinion, that GPL licensing also creates a gated community, by denying entrance to anyone who doesn't want to release their modifications under the GPL. On the contrary, the GPL actually thwarts the creation of a gated community by requiring that source code be distributed to everyone without charge, and that everyone be granted the same rights on the code as anyone else. You could even call the GPL an "open-gates" license.

Even with the new low price of admission, it's not at all clear that gated communities offer a good deal to Secondary Contributors like you and me. The gated community paradigm assumes that one party, the Initial Contributor of the program, deserves more rights than any other because that contributor did the bulk of the work. But what happens when contributors add important work? The Initial Contributor still has more rights.

One crucial thing present in Open Source licensing and absent from the gated community paradigm is the concept of circumvention. If I don't like what Red Hat is doing with their Open Source system, I am free to circumvent them and do what I wish with the code, distributing the source code to anyone I wish and creating my own community. In contrast, with a gated community, it's not possible to circumvent the party operating the gate, and it may even be possible for the gate-operator to terminate your licensing, rendering your entire investment in the modifications you've developed unsalable, and thus worthless. This isn't good for customers or Secondary Contributors.

A common criticism of Open Source is that there's no direct-revenue-capture for sales of software. If you want to make money, you have to sell something else, like integration, a trusted brand, or service and support. The gated community paradigm makes direct revenue capture possible, but generally only for one party: the Initial Contributor. A proposal by Matthew Parry about a year ago was a gated community in which a jury would award part of the money from the sale of software to Secondary Contributors based on the value of their modifications. I objected to Parry's proposal because I felt the process was cumbersome, with too many legal and tax complications, and because there was too much room for favoritism and gang rule to corrupt the jury. Parry's suggestion, however, still addresses the main failing of the gated community, bias toward the Initial Contributor, better than current proposals.

Open Source licensing works because it's simple to enter and use (at least compared to other licensing), and because it's fair to all parties. Gated community licensing, so far, is either less fair or much more complicated. Is gated community licensing worth it for the licensor? Perhaps, if it offers them an advantage they can't get from Open Source.

I don't think the restriction to only those who can pay improves the quality of development. Being "serious" enough to spend some money to join doesn't make you a good coder, and it reduces the number of coders available to the community.

Gated communities may be an advantage to the Initial Contributor if they help that contributor maintain a "lock" on their software, so that they have a commercial advantage over any other vendor and the participation of developers who work for free at the same time. However, this depends on being able to get all of the good developers to join your community rather than have them split off their own competing project with its own code base and a more equitable license. For example, a commercial company, Progeny Linux, is adding networking power beyond that in Inferno to the Linux system under an Open Source license. That group includes programmers from the Berkeley "Sprite" project and other academic networking research projects (and me, so that's all of the promotion that Progeny will get here).

Tim O'Reilly suggests that the gated community might be a good "first step" for companies not willing to go all the way into Open Source at this time. I'd support that if I was more sure that a gated community would be a successful step. It might, instead, sour a new entrant because of a lack of community participation.

In conclusion, the gated community is an interesting concept, but I'm not at all sure it's a step forward or even a good intermediate step. Open Source, however, isn't about stopping other people from running their own projects the way they like. The only thing we ask is that people be careful about what they call "Open Source." While I'm not going to start participating in gated communities, I will continue to watch them for good ideas.

Bruce Perens, president of the Linux Capital Group, is a luminary in the Linux and Open Source software community. He is a highly regarded computer scientist as well as an authority on free software licensing. His views are widely published in media publications from Wired Magazine to the Wall Street Journal. He is the primary author of the Open Source Definition, and, as such, is well respected in the international network of Open Source programmers and computer scientists.





   Page 1 of 1