Epstein’s New Paper

by on April 20, 2006 · 14 comments

Richard Epstein has a new report on intellectual property. Epstein is a brilliant legal theorist (seriously–several of his books are classics of libertarian scholarship) but unfortunately, I think he analysis of IP issues–especially technology related IP issues–is hampered by his lack of familiarity with the underlying domain. Take this passage about open source, for example:

One ongoing question is how well open source stacks up against traditional proprietary software. Much depends on the scale of the enterprise. The decentralized methods for open source work well with small systems, but are difficult to maintain as the network expands–a problem that any proprietary system also faces in integrating backwards to existing products while introducing new products. In addition, loose cooperatives must organize to fend off outsiders claiming that the entire system incorporates their trade secrets or IP. The present SCO litigation, for example, puts the entire Linux system at risk on these grounds, prompting the formation of a litigation committee to coordinate the common defense. Right now at the heart of the movement lies a commercial joint venture spearheaded by well-established firms like IBM, Intel, and Hewlett-Packard, which develop service and proprietary programs that operate on top of an open source infrastructure. The new development gives ample testimony that no loose assemblage of voluntary contributors will be able to carry the day any longer.

To be honest, I’m not sure I follow the third sentence. I think that to the contrary, in many ways open source scales better than proprietary development models, because it takes advantage of the decentralized, spontaneous processes to solve problems rather than relying on hierarchical, top-down processes. Of course, generally speaking, large corporations like dealing with other large corporations for the IT needs, so it’s not surprising that IBM does a lot of business selling open source software (along with some of their own proprietary software) to Fortune 500 companies. But that’s not because open source can’t solve the technical problems of large companies. It’s simply that “open source,” as an idea, doesn’t have a sales force and can’t meet with corporate IT directors. IBM does, and can, so it tends to get the IT contracts. But most of the value was created by the volunteers who built the underlying software.

He then claims that “at the heart of the open source movement” are IBM, Intel, and HP. He doesn’t elaborate, but I assume he’s equating “open source” with “Linux.” This is misleading for several reasons. First of all, those companies might be spending the most money on Linux-related products, but they’re hardly the core of the Linux community. Linux is still developed by a decentralized group of mostly-volunteer programmers from a wide variety of institutions, led by Linus Torvalds. They probably don’t seem significant to Mr. Epstein because they don’t have PR departments or billion-dollar balance sheets, but they’re the ones who control the direction of the core product. The work of IBM, Intel, HP, and their ilk is largely focused on making Linux work better on their particular systems, as well as building software on top of Linux to meet the needs of particular clients. Obviously, that’s often helpful to the overall project, but it hardly puts Big Blue “at the heart” of the Linux effort.

But the broader point is that Linux is just one out of dozens of major, successful open source products that are used by millions of people every day–and most of them receive far less corporate support than does Linux. Most of them are programs that Epstein has probably never heard of–projects with names like Apache, Samba, Perl, Python, gcc, MySQL, KDE, Gnome, FreeBSD, OpenSSH–but that make up the “plumbing” that make the Internet work. Each of these projects has a core team made up of, well “a loose assemblage of voluntary contributors.” Some of them get corporate support, but that support is incidental to the projects’ viability in most cases. I can’t think of any recent developments that prove that the open source model will not “be able to carry the day any longer.” To the contrary, the open source development model continues to demonstrate its vitality by churning out spectacular products without significant corporate subsidies.

Now, obviously it wouldn’t be fair to expect a 50-something law professor to be intimately familiar with products like gcc and FreeBSD. Linux is the product that gets the most press, and IBM is the Linux contributor that gets the most attention, and so Epstein naturally assumes that IBM is the biggest driver of open source software.

It’s an understandable error, but these kinds of blind spots are dangerous when you’re doing public policy analysis. If you misdiagnose the source of innovation, you’re likely to misunderstand the institutions required to promote it. Computer geeks are the ones closest to the ground of high-tech innovation. When they’re shouting from the rooftops about problems with our IP system, I think the law professors of the world ought to pay a bit more attention to what they have to say.

  • Ned Ulbricht

    Now, obviously it wouldn’t be fair to expect a 50-something law professor to be intimately familiar with products like gcc and FreeBSD.

    Why isn’t it fair?

    Back up to Professor Epstein’s claim about large-scale software projects: I think it’s reasonable to expect the professor to have read Dr. Fred Brooks’ The Mythical Man Month before making any kind of claim about software engineering. That book is a classic in the field and “everyone” has read it.

    So, why isn’t it fair to expect Professor Epstein to know something about Apache, Samba, Perl, Python, gcc, MySQL, KDE, Gnome, FreeBSD, OpenSSH?

  • http://www.blindmindseye.com MikeT

    The PFF would do well to actually bring on people with software development experience. That might temper some of their excesses when it comes to commenting about IT. DeLong, for example, always misses the irony that IPCentral’s blog is hosted on a pure open source software stack and that Movable Type was developed in Perl. I’d like to give them the benefit of the doubt and say that it’s a fundamental ignorance problem.

    The only part that Epstein seems to have gotten right is the software patent threat, and the need to have big companies like IBM bringing their “nuclear patent arsenals” to bear on companies like SCO. That, however, is caused simply by the fact that we have software patents in the first place.

    It’s not hard to find out that Google uses Linux and Python on pretty much all of its production servers. Python is one of their official languages! The company is a living testament to how open source is fit for a corporate enterprise setting. Few companies out there have the sort of massive infrastructure that Google does. Comparing Google to most of them owuld be like comparing the entire US military to the boy scouts. So… if open source software is good enough for Google, why is it still in doubt?

    You know what, though? I wonder why Epstein never bothered to contact a few Computer Science professors to get some help in fleshing out his understanding of the scope of open source development. Surely that would have been a lot easier than to risk a complete underestimation of the scope of open source in use today.

  • Doug Lay

    I’ve never read a thing by Epstein that indicates more than a very superficial, glib understanding of technology. Basically, a man in love with his own economic/legal theories, and if reality doesn’t fit those theories, it must be reality that’s wrong.

  • Mark Levitt

    I think the broader problem with this paper is its complete lack of any meaningful evidence to back up his fairly simplistic claims.

    The whole paper, in lots of words, basically says: “IP rights are good because they allow people to make money. I’m going to prove that by empirical argument. Here’s my opinion on why I’m right.”

    I expected to find some sort of analysis of sales or market size. Perhaps some study of the number of patents issued vs. the amount of venture capital available.

    Where was this empirical examination he claimed was neccessary?

  • Ned Ulbricht

    Now, obviously it wouldn’t be fair to expect a 50-something law professor to be intimately familiar with products like gcc and FreeBSD.

    Why isn’t it fair?

    Back up to Professor Epstein’s claim about large-scale software projects: I think it’s reasonable to expect the professor to have read Dr. Fred Brooks’ The Mythical Man Month before making any kind of claim about software engineering. That book is a classic in the field and “everyone” has read it.

    So, why isn’t it fair to expect Professor Epstein to know something about Apache, Samba, Perl, Python, gcc, MySQL, KDE, Gnome, FreeBSD, OpenSSH?

  • http://www.blindmindseye.com MikeT

    The PFF would do well to actually bring on people with software development experience. That might temper some of their excesses when it comes to commenting about IT. DeLong, for example, always misses the irony that IPCentral’s blog is hosted on a pure open source software stack and that Movable Type was developed in Perl. I’d like to give them the benefit of the doubt and say that it’s a fundamental ignorance problem.

    The only part that Epstein seems to have gotten right is the software patent threat, and the need to have big companies like IBM bringing their “nuclear patent arsenals” to bear on companies like SCO. That, however, is caused simply by the fact that we have software patents in the first place.

    It’s not hard to find out that Google uses Linux and Python on pretty much all of its production servers. Python is one of their official languages! The company is a living testament to how open source is fit for a corporate enterprise setting. Few companies out there have the sort of massive infrastructure that Google does. Comparing Google to most of them owuld be like comparing the entire US military to the boy scouts. So… if open source software is good enough for Google, why is it still in doubt?

    You know what, though? I wonder why Epstein never bothered to contact a few Computer Science professors to get some help in fleshing out his understanding of the scope of open source development. Surely that would have been a lot easier than to risk a complete underestimation of the scope of open source in use today.

  • Doug Lay

    I’ve never read a thing by Epstein that indicates more than a very superficial, glib understanding of technology. Basically, a man in love with his own economic/legal theories, and if reality doesn’t fit those theories, it must be reality that’s wrong.

  • Mark Levitt

    I think the broader problem with this paper is its complete lack of any meaningful evidence to back up his fairly simplistic claims.

    The whole paper, in lots of words, basically says: “IP rights are good because they allow people to make money. I’m going to prove that by empirical argument. Here’s my opinion on why I’m right.”

    I expected to find some sort of analysis of sales or market size. Perhaps some study of the number of patents issued vs. the amount of venture capital available.

    Where was this empirical examination he claimed was neccessary?

  • http://abstractfactory.blogspot.com/ Cog

    Ned, it’s “not fair” because lawyers seldom know much except the law, and most of them do not consider it necessary to know more.

    Well, OK, that’s an exaggeration, but only a slight one. The legal profession regularly holds conferences on “technology law” wherein not one bona fide technologist is present. The editors of law review articles are overworked law students who have neither the time nor the resources to obtain peer review (or even basic sanity checking) by domain experts. As long as the article’s properly footnoted and appears to exercise legal doctrines in a plausible fashion, grave errors about reality (that messy real world that exists beyond the boundaries of statutes and case law) can easily slip through. And, of course, if you’re not a domain expert, then it’s not necessarily even clear which facts need a backing citation.

    Tim: you say that the people deciding the “core direction” of Linux are mostly volunteers. I’m a little skeptical of this. I believe the most prominent kernel hackers (Linus, Alan Cox, etc.) are mostly in the employ of companies that pay them specifically to work on the kernel. See, e.g., this article in WIRED (a little outdated, but I doubt that kernel hacking’s grown less professionalized since 2001). Most successful large projects with which I’m familiar are similar. Most of the core KDE hackers, for example, are paid explicitly to hack on KDE.

    This doesn’t make Epstein any less wrong; it just makes him wrong in a different way.

  • http://abstractfactory.blogspot.com/ Cog

    Ned, it’s “not fair” because lawyers seldom know much except the law, and most of them do not consider it necessary to know more.

    Well, OK, that’s an exaggeration, but only a slight one. The legal profession regularly holds conferences on “technology law” wherein not one bona fide technologist is present. The editors of law review articles are overworked law students who have neither the time nor the resources to obtain peer review (or even basic sanity checking) by domain experts. As long as the article’s properly footnoted and appears to exercise legal doctrines in a plausible fashion, grave errors about reality (that messy real world that exists beyond the boundaries of statutes and case law) can easily slip through. And, of course, if you’re not a domain expert, then it’s not necessarily even clear which facts need a backing citation.

    Tim: you say that the people deciding the “core direction” of Linux are mostly volunteers. I’m a little skeptical of this. I believe the most prominent kernel hackers (Linus, Alan Cox, etc.) are mostly in the employ of companies that pay them specifically to work on the kernel. See, e.g., this article in WIRED (a little outdated, but I doubt that kernel hacking’s grown less professionalized since 2001). Most successful large projects with which I’m familiar are similar. Most of the core KDE hackers, for example, are paid explicitly to hack on KDE.

    This doesn’t make Epstein any less wrong; it just makes him wrong in a different way.

  • http://www.techliberation.com/ Tim

    Cog,

    An excellent point! I may have over-stated my case a bit. I think the concept of a volunteer becomes kind of fuzzy when you’re talking about the brightest open source hackers. Most of them do appear to work for companies that pay them to more or less work on open source software full-time. But I’m not sure that proves that IBM or Red Hat or whomever is subsidizing Linux development. A company doesn’t hire a hacker like Linus in order to subsidize Linux. They hire Linus because Linus is an extremely talented and well-known hacker who can help out their company in a variety of ways. If Linus decided to stop working on Linux, it’s not likely that would have much impact on his earning prospects.

    The more important point, however, is that no one company, or formal coalition of companies, controls or supervises the Linux development process. Structurally, it remains a “loose assemblage of voluntary contributors” supervised by the benevolent dictatorship of Linus Torvalds. Many of them happen to be paid to do it full time, but (as far as I know anyway) they still coordinate in a fundamentally decentralized manner.

    Again, I think Linux is somewhat anomalous in that for a variety of reasons it has attracted a disproportionate share of corporate support. Apache might be a slightly more representative example. By my count, out of more than 50 contributors there are five engineers from IBM, one each from Red Hat, Transmeta, CNet, HP, Novell, Ask Jeeves, and half a dozen or so from academia. The rest–easily a majority–list themselves as “independent” or work for companies I’ve never heard of.

    This is a “loose assemblage of voluntary contributors.” Removing the support of the big software vendors would certainly slow the development process down somewhat, but Apache could get along just fine without them.

    And actually, I think this kind of head-counting over-states the impact of the big companies’ contributions. Much of the strength of the open source development process comes from hundreds of small contributions from hundreds of programmers who find and report bugs (with, ideally, a patch to fix the problem). This bug-fixing occurs practically for free under the open source development model, yet a commercial company would have to pay an army of testers to perform the same function.

  • http://www.techliberation.com/ Tim

    Cog,

    An excellent point! I may have over-stated my case a bit. I think the concept of a volunteer becomes kind of fuzzy when you’re talking about the brightest open source hackers. Most of them do appear to work for companies that pay them to more or less work on open source software full-time. But I’m not sure that proves that IBM or Red Hat or whomever is subsidizing Linux development. A company doesn’t hire a hacker like Linus in order to subsidize Linux. They hire Linus because Linus is an extremely talented and well-known hacker who can help out their company in a variety of ways. If Linus decided to stop working on Linux, it’s not likely that would have much impact on his earning prospects.

    The more important point, however, is that no one company, or formal coalition of companies, controls or supervises the Linux development process. Structurally, it remains a “loose assemblage of voluntary contributors” supervised by the benevolent dictatorship of Linus Torvalds. Many of them happen to be paid to do it full time, but (as far as I know anyway) they still coordinate in a fundamentally decentralized manner.

    Again, I think Linux is somewhat anomalous in that for a variety of reasons it has attracted a disproportionate share of corporate support. Apache might be a slightly more representative example. By my count, out of more than 50 contributors there are five engineers from IBM, one each from Red Hat, Transmeta, CNet, HP, Novell, Ask Jeeves, and half a dozen or so from academia. The rest–easily a majority–list themselves as “independent” or work for companies I’ve never heard of.

    This is a “loose assemblage of voluntary contributors.” Removing the support of the big software vendors would certainly slow the development process down somewhat, but Apache could get along just fine without them.

    And actually, I think this kind of head-counting over-states the impact of the big companies’ contributions. Much of the strength of the open source development process comes from hundreds of small contributions from hundreds of programmers who find and report bugs (with, ideally, a patch to fix the problem). This bug-fixing occurs practically for free under the open source development model, yet a commercial company would have to pay an army of testers to perform the same function.

  • http://www.blindmindseye.com MikeT

    Cog,

    It is fair because he’s trying to formulate policy. If you’re going to formulate government policy toward software and are fundamentally ignorant of it, you need outside help from someone in the know. Why is he any different than some pimply teenager who knows a little about Linux? If he doesn’t really know much about the very model that he’s critiquing, that means that his opinion is for all intents and purposes based on ignorance.

    Another thing, the people behind most of those projects started getting paid well after they started them. I don’t know if I really think that qualifies for the same thing. KDE was well into development when Trolltech and SuSE started picking up the slack.

  • http://www.blindmindseye.com MikeT

    Cog,

    It is fair because he’s trying to formulate policy. If you’re going to formulate government policy toward software and are fundamentally ignorant of it, you need outside help from someone in the know. Why is he any different than some pimply teenager who knows a little about Linux? If he doesn’t really know much about the very model that he’s critiquing, that means that his opinion is for all intents and purposes based on ignorance.

    Another thing, the people behind most of those projects started getting paid well after they started them. I don’t know if I really think that qualifies for the same thing. KDE was well into development when Trolltech and SuSE started picking up the slack.

Previous post:

Next post: