Previous Politech message: http://www.politechbot.com/p-04007.html --- Subject: Comments on the Comments Date: Mon, 23 Sep 2002 16:16:34 -0400 From: "James V. Delong" <JDeLongat_private> To: "Declan McCullagh (E-mail)" <declanat_private> Declan My piece on the NYT editorial provoked considerable response -- next time, I will just look for a hornets nest and poke a stick in it. There are too many comments to answer individually, so I will comment generally. Viral An easy one: some objected to characterizing GPL as viral. I agree that a less-loaded term would be better, and would like to have one. Replicational , maybe? Fees Several comments claim that one can indeed charge for GPLed software, sometimes pointing to a clause early in the license allowing a charge to distribute. I disagree. The GPL allows you to charge a fee for the physical act of transferring a copy. You may also offer warranty protection for a fee. But you cannot charge for the software itself or inhibit its further distribution, which means that the person to whom you sell it is then free to distribute it at no cost, so you can sell only one copy to an enterprise. Also, under standard legal doctrines, if you charged so much for the act of transfer ($1,000 to ship a disk, for example), the license could be enforced against you on the grounds that you were engaging in a subterfuge. (Commercial distributors of Linux are engaged in an elaborate system of selling warranties and services, not the software itself.) Writing Aps The part provoking the most reaction was my comment about the legal risk of writing aps for Linux. I tried to phrase this carefully, but clearly I did not phrase it carefully enough. The point is a legal one more than a programming one. Let me try to explain my thought more clearly. As I understand it (those who said he s certainly not a programmer are absolutely correct), an application calls up various libraries from the OS. Would this be considered incorporating code in a way that would put the ap under the GPL? No one knows for certain, but in response to the problem the open source movement created the Lesser GPL, which clearly allows such use. But a program writer must be very sure about the licenses that cover the libraries he wants to tap, and, as several comments noted, there are other licenses at work here. Also, use of the LGPL is disfavored by the FSF. In addition, code-writers have a problem common to all writers. Once something is in your head, it is difficult to be sure that it is not reappearing when you write. (Ask Doris Kearns.) It would be very easy for a writer to insert code that is actually lifted from a GPL'ed program, totally inadvertently. The penalty for doing so could be drastic loss of the entire program. Were I a software company lawyer, I would be nervous about having people familiar with Linux working on proprietary programs. This issue is made more difficult by uncertainties under copyright law about when a work is derivative, an important term in the GPL. (I would be interested in knowing whether companies like Oracle and IBM separate the staff working on Linux from that working on proprietary programs.) There are also a lot of cross purposes at work here. My critics for the most part want to create proprietary programs to run on Linux. But the motives of the originators of the GPL were anti-proprietary. It is far from clear that this segment of the open source movement will acquiesce in any shift to a mixed system if they can do anything about it, and the licenses give them legal leverage. Risk is exacerbated because of the cloud of legal uncertainty surrounding the GPL. Who can enforce it (or, perhaps more important, who cannot)? How will the many ambiguities get resolved? For a proprietary program, issues can be resolved by contract. For the GPL, this is impossible; the title is too clouded. Many commenters are, in my view, naive about the functioning of the legal system. They assume that decisions on these licensing issues will be rendered quickly by judges of great intellect and high character. I have bad news -- the legal system is a tottering mess. Anyone investing in ap writing must consider the possibility of a now-wealthy tobacco plaintiffs lawyer filing a class action in a Mississippi county where the judge is his brother in law, and he can assert that all these big companies are stealing software that belongs to the people. If you do not think this is a real risk, check out the asbestos litigation now going on in West Virginia. So, while companies are indeed writing aps, they are running legal risks that are not trivial. They have good reasons for doing so, and they find the risks acceptable, but they are real. (It is also worth noting that Red Hat s 10K filed with the SEC comments that the dearth of third party programs is a major problem.) In my view, it is a tribute to the virtues of Linux that it has come so far despite these problems. Communitarian Model There is powerful appeal to the idea of a community of researchers working on the cutting edge and freely sharing information with each other. This story applies not just to Linux but to numerous other areas. It is, as some commenters noted, the history of science. The line people usually try to draw is between fundamental and applied research. Fundamental research is supported by universities, governments, and often the researchers themselves because its practical import is unclear and support by customers unlikely. Applied research, we think, can and should be funded through market mechanisms. I think it is necessary to draw this line with respect to Linux. As it expands, it gets into areas where more of the work necessary is not at the cutting edge, and thus cannot run solely on the intellectual interest of the programmers. Also, once commercialization comes into play, it is difficult to run a system that mixes volunteers and paid workers. The Red Hat 10K points to the movement s heavy dependence on a few named programmers, and on the programming community generally, and the loss if they were to become unavailable. It also reveals that its officers are paid as much as $400K, plus stock options. Sooner or later, the programmers will notice the discrepancy, especially if their forbearance does not result in an intellectual contribution to the software community but simply in lower prices to the world at large, or more pay for the marketing executives. Other Programs Some pointed out that it is misleading to focus on the GPL when there are many other open programs available. This is a valid point, especially because some of these, such as the BSD, do indeed foster mixed systems of open and proprietary models. Convergence Many people defended the quality of Linux and the ability it affords to fix problems oneself or by putting a query out to the world at large (without waiting forever for some damn tech support desk). I have no argument with these points. As I noted in the original piece, open code has many virtues. The focus of my thought was rather different while the open approach has virtues, so does the proprietary approach. The best overall system will encompass both, I think, and it is important that governments not go off half-cocked. I intended to defend the proprietary model more than attack the open source idea. Best, Jim James V. DeLong Senior Fellow - Project on Technology & Innovation Competitive Enterprise Institute 1001 Connecticut Ave., NW - Suite 1250 Washington, DC 20036 (202) 331-1010 TEL (202) 331-0640 FAX jdelongat_private http://www.cei.org/sections/section38.cfm ------------------------------------------------------------------------- POLITECH -- Declan McCullagh's politics and technology mailing list You may redistribute this message freely if you include this notice. To subscribe to Politech: http://www.politechbot.com/info/subscribe.html This message is archived at http://www.politechbot.com/ Declan McCullagh's photographs are at http://www.mccullagh.org/ ------------------------------------------------------------------------- Like Politech? Make a donation here: http://www.politechbot.com/donate/ Recent CNET News.com articles: http://news.search.com/search?q=declan CNET Radio 9:40 am ET weekdays: http://cnet.com/broadband/0-7227152.html -------------------------------------------------------------------------
This archive was generated by hypermail 2b30 : Thu Sep 26 2002 - 00:54:42 PDT