Date: Dec 2 1998
From: David Downie

I have a question for Tim O'Reilly.

Aparently Stallman attacked you at one of your conferences, and pointed out that ALL of the arguments for open source SOFTWARE can be applied equally to DOCUMENTATION ie BOOKS. ie YOUR BOOKS.

How do you respond to the argument that all of your support is just to line your own pockets through your refusal to release your books under the GPL? You have the best unix books, Open Source is unix, it is in your best interests for the whole world to learn unix and hence buy your books.

It is an interesting argument, and I haven't heard what your response was (although I did hear it was quite lengthy). I would appreciate it if you could outline your rebuttal either personally or on your web page (and point me to it).


David Downie


>I have a question for Tim O'Reilly.

>Aparently Stallman attacked you at once of your conferences, and pointed out that ALL of the arguments for open source SOFTWARE can be applied equally to DOCUMENTATION ie BOOKS. ie YOUR BOOKS.

This isn't quite right. He argued that to the extent that documentation can be considered a necessary part of a product, that it ought to be free, so people can modify it when they modify the software. I agree with him about basic documentation for a product--and, for example, if you look at the Perl distribution, there is a lot of good free documentation that is part of it. All free software authors need to take seriously the job of documenting their work, and users of a product can helpfully contribute documentation in the same way as they contribute bug fixes or improvements.

When he takes the next step and says that authors of value-added books have an *obligation* to make them free as well--there we part company.

>How do you respond to the argument that all of your support is just to line your own pockets through your refusal to release your books under the GPL? You have the best unix books, Open Source is unix, it is in your best interests for the whole world to learn unix and hence buy your books.

Richard thinks there is a moral imperative underlying the free redistribution of software, and now, by extension, other information. Richard feels that since there isn't any physical cost associated with copying software, limiting free redistribution is a form of extortion. I on the other hand feel that it's immoral to try to compel someone else to give you something they've created without compensating them in some way. That is, when software is freed, it is a gift, not the result of an obligation. I found Richard's comments at the Open Source Developer's Day, where he called John Ousterhout a parasite because he now wants to build proprietary tools on top of tcl, a defining moment. This is akin to children feeling that their parents owe them an inheritance, or people on welfare feeling that the government owes them a handout. Richard should be grateful for what John has already given, not castigating him because he doesn't want to give even more.

I believe that the imperative underlying Open Source is pragmatic rather than moral. That is, it's a great development methodology that leads to better software and leads to the formation of communities around that software. Eric Raymond's paper, The Cathedral and the Bazaar, sets out a lot of these pragmatic arguments for Open Source software.

The pragmatic arguments for open source software include:

  • You've developed the product to "scratch your own itch" and because bits are easily copyable, there is no incremental cost, and possible incremental benefit, from giving other people access to your work.

  • By releasing as Open Source a tool that you originally developed for your own use you gain a cadre of co-developers who will give you back bug fixes, additional functionality, and the gifts of their own work on the same terms. This is a value-exchange, just not a monetary one. And yes, there are "free riders" on the system, but even these free riders, who contribute no code, contribute their attention and the good word, thus spreading the use of the package and finding you more potential contributors. (Effectively, they are carriers of your meme.) They also give you their esteem. As Raymond points out, in a "gift culture", you get status as a result of what you give away. Status has always been important in the hacker community.

  • By releasing Open Source software, you contribute to the overall health of the programming community, who will give you products of their own work. You benefit from the entire ecosystem, not just from the input of the community into your own specific project.

Now, let's look at those arguments as they apply to books.

First, very few people write a book to "scratch their own itch." Instead, people generally write books for two reasons, to serve other people (a noble goal), or to earn something (either money or esteem, or both) for themselves. So one of the fundamental motives underlying free software is often missing. If someone wants to write a book because they believe in the Open Source or free software movement, more power to them. I'll be happy to publish that book if I think it's a good book. But I'm not going to tell my authors that they have an obligation to do so!

In fact, we have published several books under copyleft (most notably the Linux Network Administrator's Guide) at the request of the authors, and are prepared to do so for other books, provided that the authors are aware of the negative impact that we believe this will have on their sales (and consequent royalties.) For instance, when we published the Linux NAG, it was republished by both SSC (as a standalone book) and Macmillan (as part of a "Linux Bible" or some such), neither of whom paid royalties to the author. Their costs were therefore lower--but more to the point, they took away some of the market for the book. The Linux NAG has always done less well for us than other Linux books for which we'd expect comparable sales, and perhaps partly as a result, the author was demotivated to continue updating the book. (We've now got someone else working on a new edition.)

This experience seems to undercut the argument that just like free software, freely available documentation will "spread the word" and result in wider use of the product, thus greater book sales. This does not seem to be the case in practice. Apart from the fact that there are significant costs in book publishing and distribution that are not necessarily there in software, you have to take into account the differences in distribution methods for the two products. While most free software is downloaded directly by the user, or put on a machine by a vendor, books are typically purchased through an extremely inefficient system. Only about 10% of our books are sold directly to the end customer. The greatest majority are sold through indirect channels like bookstores, libraries, and corporations. And the buying patterns of these middlemen are not under the control of the publisher. There is far more product than available shelf space, so books that sell below a certain threshhold are returned and not reordered. So, for example, if there are three versions of the same book, they may well divide the market when they are first released, leading the middlemen to decide that there's not enough demand to keep any of the titles in stock.

In order to combat this problem, we're working (with a number of other publishers) on a license that will allow binary redistribution of a book's contents with the software, but will reserve the right to print, publish and sell printed copies to the original publisher. But mind you, even with that license, we're going to be experimenting, not going at it wholesale. The point is that if Open Source is really a better model, it ought to lead to better results. If it doesn't, then we need to learn from experience where it fits and where it doesn't.

Giving away books as open source would certainly seem to meet the third benefit I listed above, of contributing to the overall health of the community, but in fact, I'm not sure that even that is true. I believe that without the economic incentive that a publisher offers, far fewer books would be written, and that the net gain for the community would be less.

At bottom, this is true even of software, I think. There are value-added features that the Open Source community doesn't seem to be good at providing, and that commercial companies can do a better job at. New companies like Sendmail, Inc., Scriptics, and ActiveState are exploring a hybrid business model, which relies on Open Source as an important technique for spreading the word, fostering innovation, improving product quality, and giving back to the community, but reserves certain products as value added so that they can make money.

At the end of the day, I see Open Source as a free market economy of ideas. If you look at free markets, there are all kinds of value exchanges, not just exchanges of money. But if you take away value exchange altogether, and mandate forced sharing, you vitiate the very thing that you hope to create.

I believe that the generosity that is at the heart of the Open Source community is a kind of enlightened self-interest, not a moral imperative. It behooves all of us to find the right balance between what we give away and what we keep for ourselves. An awful lot depends on your goals. You pick the hat to fit the head.

I give great kudos to Richard and to anyone else who gives away their software without any thought of personal gain. I also give great kudos to people who figure out how to build something valuable enough that other people want to pay for it. And I see no reason to tell people they should move to one side of the balance or the other.

Well, that's not quite true. I do believe very strongly that there are some things that are common goods, that belong to everyone by right. A good analogy is the environmental movement. Even if we don't own the environment, we have some rights in it. And someone who clear cuts the forest and doesn't replant it, or pollutes and doesn't clean up after themselves, is wronging the rest of us. In a similar way, someone who takes something that was created by others and tries to take away the freedom that was granted by the original creator is the equivalent of an environmental criminal.

But here, once again, I part company with Richard. I far prefer the BSD style licenses to the GPL. If I build proprietary added-value on top of someone else's work, I don't take anything away from them or their users. They have no obligation to use my added value. Where the "environmental crime" comes in is when you try to actually deny people the right to use the original package. This is the kind of thing we saw talked about in the Halloween Document, where Microsoft was talking about subverting open standards.

In short, I feel that undercutting an open standard so that other people's products don't work is wrong. But simply adding value, and offering a derivative product is fine--as long as that was the intent of the creator of that product.

A really good example of this is the X Window System. It was designed with the express intent that people would build proprietary products on top of it. So when GPL fans deride the license because it allowed people to build things that weren't contributed back to the community, they are deriding the intent of the creators. It's a little bit like Puritans complaining because other people are having fun.

That being said, there are definitely issues with the BSD style license that the GPL addresses. In particular, it is possible for the market to fragment more easily. But like software, licenses need to evolve to more perfectly meet our needs. A lot of good work and thinking is going into this area, and I'm confident that we'll eventually find even better ways to encourage both the common good and the economic incentives that can be an important stimulus to the creation of value.

And once again, I think this is in the end a pragmatic or scientific issue, not a religious one. Open Source is like gravity. Our goal is to understand the laws that make it work, not to decree how we'd like it to work.

My goal is to encourage people to write great books that really help people, and I'm charting the best course I know how towards that goal. So far, I think the results speak pretty well for themselves.

--Tim O'Reilly

David Downie responds...

Return to: Ask Tim Archive