Jim DeLong received several thoughtful responses to his TechCentralStation article on Open Source: "Today Linux; Tomorrow the World?”

Major points are summarized below, together with his initial responses. Within the next few weeks, PFF will be publishing a paper in its Progress on Point series that takes up many of these issues in more detail.

The letters in full are posted at the end, with names removed, as we have not asked permission from the correspondents to identify them.

Reader Comment: The productivity estimate that a programmer can produce 1,000 lines of finished code in a year is too low.

DeLong’s Response: The figure comes from a classic work on software production, Frederick Brooks, Jr., The Mythical Man-Month, first published in 1975, but revised in 1995 and still a respectable seller. It should be noted that this estimate is for productivity of programmers working on large programs that require considerable testing, debugging, and integration. There is no consensus on the issue, admittedly, but I think the range of estimates is probably between 1,000 lines and 5,000 per year. Most of the time goes into the follow-up.

Comment: Rights in intellectual property have expanded too far. See, for example, the time extension upheld in the Eldred case.

Response: This is an interesting question. The statement that rights have expanded is made repeatedly, but it is not necessarily true. As Ralph Oman, former Register of Copyrights, wrote (Letter to Editor, The Washington Post, March 11, 2002, p. A20):

During the past 90 years, to solve political controversies and to hand out economic freebies to sympathetic supplicants, Congress has sweet-talked authors into giving up their right to say yes or no to a use of their works -- the essence of a property right -- in exchange for a longer term. A long list of special pleaders now gets free use of copyrighted works, including small businesses, veterans' groups, bars, scholars, restaurants, fraternal groups, marching bands, Boy Scout troops, nursing homes, libraries, radio broadcasters and home tapers. Another long list of powerful industries gets to use copyrighted works in exchange for a small government-set fee, whether the author likes it or not: cable and satellite companies, record companies, juke-box operators, public broadcasters and, most recently, Internet companies.

The authors went along with this coerced subsidization because Congress held out the promise of a longer term of protection. It would be a switch-a-roo worthy of Lucy yanking away Charlie Brown's football if the Supreme Court removed the carrot [in Eldred] and left the authors with the dirty end of the stick.

The real problem with the copyright extension litigated in Eldred is that there are hundreds of thousands of copyrights out there of varying ages, most of them lying fallow, and it is extremely expensive for someone who wants to use any of them to ascertain their status and/or ownership. This transaction cost issue does need to be addressed. I think Lawrence Lessig’s proposal to require periodic re-affirmation by a copyright holder deserves a close look.

Comment: (1) Linux stayed afloat without corporate money for years, right up until 1998, when IBM came in; (2) Programmers have good reasons for liking open source software; (3) The programmers are concerned simply with creating a system that works for them, and the rest of us can go suck eggs.

Response: The first comment is partially true, but Linux and other open source programs certainly piggybacked on the corporate/ university/ government money spent on Unix in the 1970s-1990s; they are not systems built from scratch by hobbyists in their off-hours. As the second point, there are indeed excellent reasons for programmers to like open source. To be able to code, they must be able to see the code. Finally, it is true that programmers have a view that this is something they are doing, and anyone who doesn’t like it need not come to the party. However, open source has moved beyond this point, out into the world at large, and many people are concerned with making into a system that works for non-nerds. There is considerable inconsistency in trying to be both a hobbyists’ preserve and a mainstay of the world of commerce and industry.

Comment: (1) The transaction costs of negotiating rights inhibits creativity, in contrast to the situation in the physical world; (2) There is a gridlock of patent and copyright rules.

Response: This is an excellent point. Physical goods often contain a considerable component of intellectual property, whether in the form of patents or know-how. When you buy that physical product, you automatically get the right to incorporate all of that intellectual property into a subsequent product. The transaction costs are zero.

Comparable institutions for pure IP do not exist, and it is serious gap. Interestingly, the loudest complaints have come from the music industry, which has had a very difficult time negotiating the rights and permissions necessary to set up paid downloading services. Rights in music are very fragmented, and very hard to track down.

The patent point raises a serious issue. See the discussion in the recent FTC report. An excellent product of the open source movement will be the pressure it puts on software companies, and technology companies generally, to address the problem. One caution, though – intellectual property issues raised by the General Public License are equally complex.

Comment: "Most open source programmers are not looking to build a workers paradise. All we are really interested in is a programmer's paradise."

Response: This may represent the view of many programmers, but a sizeable number of others view the enterprise in overtly political terms. See, for example, a recent article featured in NewsForge: The Online Newspaper for Linux and Open Source, entitled "Something New in Free Software," (Jan. 30. 2004), which starts with the line, "Intellectual property rights and their protection have become the favorite battleground for global capitalistic expansion," and soon continues with, "This is why a central concern of the movement for global justice is to shake off the shackles that markets would like to use to imprison knowledge."

Comment: I doubt that billions was spent on Unix.

Response: Oh, I think so. Looking at Eric Raymond’s The Art of Unix Programming, one is struck by the incredible variety and intensity of the Unix effort over more than three decades.

Comment: Isn’t open source rather like the basic model of science in general?

Response: Yes, and this is a fine analogy. But in the scientific world, the distinction is made between basic research, which is a public good, and applied research, which can and should be produced by private commercial entities. The distinction does not always work well, of course, but it provides at least rough guidance. If the same distinction were made in the software world, a lot of open source programs would fall on the applied side of the line, I think.

Comment: The business model is to give the software away and make money off the services. And IBM seems to be making it work.

Response: This is certainly the model applied by Red Hat and other distributors. And, indeed, it might well work. That is for the market to sort out. My principal focus in the TCS piece was on the proposition that open source provides a general model for production of IP. It works, so far, in the software area under certain conditions, and it is important to understand what those are. Primarily, I think, Linux has provided a long-sought but ever-elusive Unix standard on which a number of commercial operators could agree.

Comment: The Berkeley Software Development (BSD) license is superior to the General Public License (GPL).

Response: I agree completely. The BSD allows a maker of proprietary software to incorporate code as long as it publishes a notice that the product contains BSD’ed material. The GPL requires that any software that contains any BSD’ed code, or is derivative of it in any way, must also be made open source under the GPL. The biggest problem with the GPL is that it is incredibly difficult to interpret how far this “viral” (Eric Raymond’s term) provision extends.

Comment: Linux has a developer core of full-time, paid employees.

Response: Yes, it does. That was one of my points: it is a myth to think that open source software is being produced by armies of unpaid volunteers. So if programmers must be paid, there must be a way of producing a revenue stream from which to pay them.

Comment: Open Source software can be sold for a profit.

Response: In theory, the first copy can be sold. But once it has been sold, then the licenses allow the buyer to go into competition with the original seller, and its buyers can go into competition with it, and so on. Open source software cannot by itself be made into a commercial proposition.

Comment: Open source software is superior.

Response: Again, the point is asserted regularly, but not proven. In fact, some experts hold the opposite view.

Comment: Commercial enterprises are spending billions in support of free software..

Response: Yes, and for good reasons. It provides an open standard, and they do indeed get to profit from the investments made in the past by others, particularly investments in Unix. Also, programmers and IT professionals like open source software because they can see the code, and the commercial enterprises are responding to this as a market force. A good result of the open source movement is that it is forcing proprietary companies to rethink their policies and find ways to make code accessible to users.

Comment: The Berkman Center and the Free Software Foundation do not support the use of tax revenues for the production of IP.

Response: I disagree with the commenters’ interpretation of the materials available from those organizations.

Comment: Open source is producing billions of dollars in value by reducing software costs.

Response: Open source is certainly introducing an element of healthy competition, which is why many of the customers are willing to put money into the Open Source Development Laboratory and other open source institutions. This is a fine thing. Again, my point is that the phenomenon of open source software is highly-context specific, and care must be taken in generalizing beyond that context.

Comment: DeLong is not a programmer, and does not know what he is talking about.

Response: DeLong is certainly not a programmer, and there is more truth in the second part of the sentence than I wish there were. A severe "two cultures" problem exists. I am not expert on programming, but I know a lot about law, economics, and industrial organization, topics on which the programmers are weak, and because the open source movement is now inserting itself into all those fields, the programmers cannot demand immunity from dissection. The obvious solution, of course, is dialogue -- such as this exchange, which seems to me to be a good one. The points made by the commenters are excellent, and it is not characterized by the flame wars that distort many Internet discussions.

My longer paper on open source issues, which will be ready in a couple of weeks, is labeled "Version 1.0," because -- in the long-standing tradition of software development -- I am sure it will have bugs that need correction by its reader community.

Comments in Full

Sir, your article was very interesting and opened a new line of thought for me regarding copyright. I support revisions of the copyright and patent laws because I believe too much power is in the hands of the holders. I sympathize with the "pirates," not because music should be free, or someone makes too much money, or because the artists do not sufficiently benefit, but because time benefits are too extreme, too much is lost or at least held back from would-be consumers and packages do not reflect consumer desires. In other words, the public is being taken advantage of by a gradually accumulating system of law and privilege that works to the detriment of innovation and opportunity, in my opinion. I can see that you believe in the market, however I do not understand whether you believe that it should remain static. Do you agree that there is a problem or do you stand with the status quo?

"IBM, HP, Sun, Dell, and others are putting in billions of dollars, and this is what keeps Linux afloat."

IBM got in the boat in 1998. The others you mention followed afterwards, some quite reluctantly. What, pray tell, kept Linux afloat for the first six years of its development? It was those scant volunteer resources, methinks.

Open Source is flourishing because the producers, the programmers, want it to flourish. Plain and simple. The industry has embraced Linux because they must. Faced with the reality not of their own making, they cannot afford to turn away.

Most open source programmers are not looking to build a workers paradise. All we are really interested in is a programmer's paradise. The rest of you can go suck eggs. But feel free to talk among yourselves.

I just read your Linux article on TechCentralStation.

Some things I don't agree with -- more later. However, the real problem with strong IP protection is the transaction costs of doing derivative work.

If I want to build a material thing -- a better mouse trap, say -- I can go into a hardware store, buy the springs, the little pieces of wood, etc. I can put them together and start selling my product. If it's a success, I can get quantity discounts. At no point do I have to sign a contract with the spring manufacturer, or negotiate the percentage they will get from my use of their product as a component.

In the material world, we've evolved a market system that puts more or less uniform prices on a product, for all buyers, for all uses.

In the IP world, this is not the case at all. If I want to sample a song, or take a frame from a movie and make it into a funny poster or T-shirt, forget about it. The overhead of negotiating the rights to something like that make the whole thing pointlessly expensive. Even if I anticipate huge returns, the time it takes to negotiate all the contracts for a complex project would greatly extend my development time, and reduce my returns.

In the software world, yes, there are utopians who talk about "information wants to be free". But then there are working programmers like myself who would like to do large projects, using pieces of work already done, who are completely shut out by the high overheads associated with obtaining the rights. We want to build software out of commodity parts, the way physical industries seem to be able to. "Open source" offers a way to do that. The strict IP model, where authors control every aspect of how a product is used, does not.

Until this overhead is reduced, the only experimental software you are going to see is open source. Even middle-sized commercial companies cannot afford the time and risk it takes to assemble pieces, where each piece requires a lawyer, a negotiation, and an expensive contract. Only Microsoft and other giants can afford that process. They know this, and so are perfectly happy to push a strict IP regime that crushes all chance of small rivals doing something clever. The occasional company that slips through the net can simply be bought up. If it refuses to yield, there will almost certainly be a patent infringement case that can be brought.

The current system of draconian patent and copyright laws will almost certainly push software development outside the U.S. It is definitely in the interest of the third world to ignore these rights, and to do their own development with open source. Software development is not a capital-intensive process -- nerds with cheap PC's can develop anything now. I expect to see a lot more work moving overseas.

As for other points in your article:

- I don't have a number, but I strongly doubt billions were spent on Unix. The original work was done by a handful of researchers at Bell Labs. It has been added to by graduate students working for peanuts on grants at universities across the country. Linux has similarly grown by hundreds of small projects, unfunded by anything but spare time. Only very recently have heavyweights like IBM thrown serious money. They are in it because they need an alternative to Windows, or else they all become Microsoft followers.

- 1000 lines a year is a gross underestimate of what programmers produce now. At one time, when debuggers were primitive and many programmers working in low level languages like C, this might have been an average over all members of a project (programmers, managers, documentation people, quality assurance, etc.) For small groups working alone, 20,000 lines a year would be much more typical.

- You say that an open source model is not appropriate for something like music and movies. Big Hollywood productions require millions, but most of it goes into special effects, or the salaries of big name actors. Something very similar could be produced for much less. And artists, singers, actors and so on are notoriously poorly paid. The thing to keep in mind is that 99% of these creative types do it for fun, have day jobs or (if they are lucky) make minimum wage-level livings from their art. The Hollywood movie and the big name pop star are the tiny, rare exception in the world of art and music.

One view is that software is going the same way. We do the fun stuff we care about in our off hours, and have a day job doing boring work for some large software house that wouldn't know an innovation if it bit them.

And that's a shame.

This is in regard to your recent article at http://www.techcentralstation.com

I suggest you do a little more research before writing such an article.

Have you ever coded? administered a Unix or Linx Server? Would you even know one if it smacked you in the face? I doubt it. Open Source Software is a hard concept for many people to understand and appreciate if you don't use it. I will give you an example. I once bought a wireless lan card. This cards driver was compiled only to use an interface that was called eth0 or eth1 my interface was called wlan0. Since this driver was Open Souce I was able to patch it to look at eth0, eth1, and wlan0. A small triumph but something you could never get a proprietary software company to do. By the way most Open Source Software is developed on *bsd or Linux because it is cheap and you don't need a $40,000 dollar machine to test your code on. Also 1000 lines of code is way way low.

Mr. DeLong,

Open Source is not the answer to the world questions. It is as you've said, Open source can only exist in an acedemic envirionment. You are correct. Open Source was never meant to create profit (not to say that it cannot be done). However, the place that Open Source excels is in acedemia. As qoted by Eric Raymond, on what he call's Linus' Law: "the more eyeballs, the less bugs there will be", in short: peer review on a global scale.

Open SourCe is the biggest scientific collaboration on the planet. Think back to acedemia, the essence of science is that all theories are false until proven other wise. Open Source is very closelink to Computer Science, wouldnt' you say so? If so, follwing the basic scientific theory that I've just mentioned, Open Source is merely a global extention of that.

Open source might not help you make money, but it will have you cut cost, by giving users access to a global network of in dividual who are able to approach a problem in a thousand ways, and disect it, at almost little or no cost. Software creation now cost hundreds of thousand of dollars to create, it will cost millions more to debug. This is why companies like IBM are willing to give away software, because they get a community of people to work for them for free & return, they give them source code. They can sell services to the software that they give away for free. The reason being, while individual developers can fix their own issues, corporations want to point a finger at someone responsible.

Do you know that 60% of IBM' srevenue comes from software? that for evey $1 of software they give away for free, they make $4 off the services!

(& you said you can't make money with free software! :) )

I've hope i've helped to clearify your issues with this e-mail.

Dear James,

I found your article on Linux to be interesting and informative. I agree with a majority of what you said. Namely that open source is not a full alternative to the evil software companies. Open source has it's place, just like commercial software has it's place.

There are some things I'd like to convey. In reading an article about Larry Wall, creator of PERL, he talks about the type of people who are involved in open source and they generally are not pro-capitalism. In a personal experience, I was visiting a Linux user group at one time and it was about the time that RedHat went public. One of the gentlemen there replied "Well, I guess it's time to move to FreeBSD." I didn't get the joke. Well, I did. What he was saying is that as soon as a business gets hold of something it's instantaneously evil. Many of the people involved in open source are inherently socialist. Which is fine by me as long as they release their software under a BSD-style license (PostgreSQL, Apache, Jakarta, etc.) rather than the GPL. The GPL is full of crap.

One thing note. Your quote "Very roughly, a professional programmer can produce about 1000 lines of polished code in a year. A distribution of Windows or Linux has 30 million lines.)" I'm not sure how that is calculated but I find, for myself, that number is extremely low. I'm no super programmer by any stretch of the imagination, but I do that in about 2 weeks. Maybe my code isn't polished enough.

Dear Editor,

I can only imagine that Harvard Law School would require a class on logic and philosophy. Mr Delong should be well aware of logical fallacy called a Straw Man Argument. In this article, he makes an incorrect and distorted picture of Open Source (and by implication Free Software) and then burns that false model in effigy.

Firstly, he make the comment that, “The theory is that Linux and other open source programs are written by hordes of volunteers...” The GNU/Linux operating system has a developer core of full time paid developers. These people are paid buy corporations, both directly and indirectly. Linux Torvalds himself works for the Open Source Development labs, who's funding is provided for by over forty leading technology companies (www.osdl.org/about_osdl/members/). Several other core developers are paid by GNU/Linux companies like Red Hat, Suse, and Mandrake. The Apache Foundation pays for the development and maintenance of the Apache Webserver, which is sponsored by over thirty of the largest technology companies (www.apache.org/foundation/members.html).

Next Mr Delong makes the following statement, “If a program is to be sold, this source code is kept confidential,...“. Clearly he seems to be under the impression that Open Source (and Free Software) can not be sold for money and for profit. This is simply untrue. Open Source software created under the BSD License has been used in proprietary software for years. For example, Apple's OS X operating system is built on an Open Source foundation of code from the BSD Operating system and the CMU Mach kernel.

Microsoft itself sells a product called Services For Unix, which contain code released under both the BSD License and the GNU GPL license. The GNU GPL has a requirement that the source code must be provided along with binaries, of which Microsoft has been dutifully in compliance.

There is no need to explore his of vaporware claims because the facts show otherwise. Despite his sophistry, GNU, Linux, BSD, Apache, PHP, Samba, MySQL, PostgreSQL do in fact exist. They are proving every day that a distributed internet driven model creates better software. IBM, HP, Novell, Red Hat, Dell, Cisco, Apple and many others are literally seeing billions of dollars in value from Open Source and Free Software. This value comes from hardware sales, support contracts, services, and add-on commercial software. With regard to marginal costs, give me a few old computers and an internet connection. I can build you an enterprise web cluster with redundant database, an Java eCommerce engine, and a hardened firewall. If you bought this software, from a company like Microsoft, it would cost well over a hundred thousand dollars. If Mr Delong would like a support contract to run all this, I would be happy to provide my services.

After making an effort to sound reasonable, Mr Delong can't help himself by playing the Socialist card. The two cases he sites, the Berkmen Center citation is for a conference session to debate the use of tax revenue. From a general policy stance, the Berkmen Center does not advocate such a stance. Again with the Electronic Freedom Foundation citation, several options are discussed with the goal of paying artists for their works. It should be noted that neither organization is consider to be an “Open Source Theorist” as such. Free Software and Open Source are a politically diverse group. Contrary to Mr Delong's characterization, we tend to lean towards Libertarian views. We simply don't need government to create (or pay for) our software.

In closing I mention IBM again, as they did a billion (with a “B”) dollars in GNU/Linux business last year. A billion dollars is well beyond the realm of mythical academic dreaming. Free Software has created a vastly improved market. It has created billions of dollars in value by reducing software costs for companies like eBay, Google, Merril Lynch, Burlington Coat Factory, and Amazon.com. I propose that the mythical dreaming here is done largely in Washington think tanks.