April 24, 2019

Tim O'Reilly on Open Source and Linux

Simon Cozens Interviews for Linuxplanet

  • September 29, 1999
  • By Simon Cozens
LinuxPlanet: I guess I don't really need to ask this, but, by way of introduction, who are you and what do you do?

Tim O'Reilly:
All right, heh. I'm Tim O'Reilly, and I'm the founder and CEO of O'Reilly and Associates. A lot of people know us as a book publisher, but we tend to think of our core competency as really being technology transfer. In other words, we pay attention to what the leading developers, administrators, interesting people in the technology industry are doing, and we write it down. And we do that basically through books; we also manage a number of websites, xml.com, we also run conferences. So we're a general technology information company.

LP: How about you personally, how did you get involved in the Open Source movement?

Tim O'Reilly:
Well, you know, it's funny, because we really have been involved from quite early on. One of our very first books was about UNIX, and while UNIX is not strictly speaking an open source technology by license, you know, Open Source technically applies only to certain licenses, but I think in a broader sense, the Open Source movement is about independent collaborative development, and a lot of the early UNIX development was in fact done that way; it started out at Bell Labs, but it happened in universities, and there were all these programs that were developed by people who were just saying, ``Hey, we need this stuff''.

And so fairly early on, I started working with UNIX as a consultant--my company did documentation consulting, and as we started working with UNIX we noticed that a lot of really cool programs didn't have any documentation, so we started writing it. And so, you know, our business really grew out of the fact that Open Source was happening. For example, there was no good documentation on UUCP, and we were using UUCP, which is the UNIX dial-up telephone network, and I wrote a book about it, and that really became very very widely used and helped UUCP and USENET spread. We did documentation for the X Windows System. A lot of these were important Open Source projects, and because there wasn't a company behind them, there wasn't anybody to do documentation.

So we were involved from the early days, but we didn't really start getting very very active until really about probably 1997. Now prior to that, we'd been very active in the early days of the web and getting the web off the ground; that also came from this technology transfer concept--we switched from using UUCP to using TCP/IP around 1990, 1991, and again we noticed all this stuff wasn't written down, so we did the first real book about the internet, which got a lot of attention. In fact, we've heard stories that the Mosaic project was inspired by a piece of junk mail from O'Reilly, that's how they learnt about the web.

But moving on, starting about 1996 or 1997, I started to be bothered by the discrepancy between what I saw in our book sales, and what was being written about by the technology press, and in particular, things came to a head for me around Perl. Our book, Programming Perl, which we'd originally published in 1991, had continued to grow in sales in 1994, '95, '96. With the web, it was really exploding. In fact, the computer book buyer at Borders told us that in 1996, Programming Perl was Borders' most profitable book in any category, and that was just really interesting for me, because in that same time period, there was no mention of Programming Perl in the computer trade press. Instead there was all this talk about ActiveX, you know, Microsoft ActiveX, which nobody used. And I decided I wanted to do something about that, so I started talking more about Perl, and you know, I'd already been talking about the Internet, so it was just an extension of talking about the Internet, just saying, well, here are these interesting technologies that are widely used on the Internet that nobody's talking about.

Anyway, in early 1998, I organised sort of small private meeting of Open Source developers--at the time we were calling them `Free Software' developers--and the reason was also that in my work of documenting a lot of these programs I realised that many of these projects had a lot in common, but the people didn't necessarily know each other, they didn't talk to each other. So I knew a lot more of them than knew each other. Actually, going back a bit, in 1997 we decided to hold a Perl Conference, and The Perl Conference was a huge success and it was really exciting for me to have all these developers who knew each other by email but had never met in the flesh. So the Open Source Summit was really an extension of what had happened at the Perl Conference; I went, "Gosh, Linus and Larry Wall really ought to know each other", you know.

So, I invited the leaders of many of the major Open Source projects to an all-day private meeting just to say "what do we have in common?". But then I also thought "this is a really good PR opportunity", so we scheduled a press conference afterwards. And it's funny because the subject of the meeting was "well, what do we have in common? And, by the way guys, there's gonna be a press conference afterwards so we have to tell them something." But what was really interesting in that meeting was that one of the things we discussed was the problems with the term `free software', that it had an ambiguous meaning.

For example, Linus said, ``I didn't realise that `free' has these two different meanings in English.'' So we started talking about libre versus gratis, and so on, and those problems, and Eric Raymond at that meeting proposed that we use the term `Open Source' , and there was some heated debate but reventually everyone agreed that we would go out there and sign off on that, and so that meeting was the formal introduction of the term to the world, and we ended up not long afterwards with a lot of major, major press, you know, New York Times, the Forbes story came out of that, and so I realised that it was very important to start promoting these things.

I suppose I've continued to be a bit of a contrarian, even with respect to Open source; there's been a lot of attention paid to Linux, and I've always been the one who keeps trying to say, "Hey, don't forget some of these other technologies". You know, for example, I really like to point people at Paul Vixie, who most people have never heard of, but in fact, maintains one of the most important Open Source programs out there, BIND, the Berkley Internet Name Daemon, that's the program that makes the whole Internet work, in terms of at least the naming, all those URLs, all those email addresses, none of them would work without Paul Vixie's stuff, and yet most people have never heard of him. Again, I think Linux is wonderful, but I also think that it's very, very important for people to get below the one buzz-word and to hear about some of these other projects. So, in addition to Linux, it's really important for the world to hear about Perl, for the world to hear about Python...

LP: Well, maybe not Python...

Tim O'Reilly:

...for the world to hear about many of the sub-projects: Samba, a lot of the Internet technologies--Apache, Sendmail.

LP: I know O'Reilly gets criticism, especially from Richard Stallman, that you embrace the Open Source movement, but people have to pay for your books. What do you think of the Open Content model, is that something you'll be looking into?

Tim O'Reilly:

Well, I've always said to Richard that we will publish books under whatever license an author wants to publish a book under, and in fact, we GPL'ed out first book--actually, I don't know if it's GPL'ed actually, it's really the license of the Linux Documentation Project--must be four or five years ago, the Linux Network Administrator's Guide. Now what we found was that sales of that book were less, in our opinion, than they would have been without that license. and so when I've talked to authors, I've told them that experience and I've asked them, "well, what do you want to do?"

And you know, for example, we're working on a book on Samba which is going to released under a completely open license, we're working on some GNOME and GTK documentation, we've just released a book Learning Debian GNU/Linux which is completely free. If the authors are willing to do that, we're willing to do that. But I do think that there are some interesting differences between free software and free documentation, and that is, you have a certain amount of free software that's developed for political reasons: you know, people who are hard-core Free Software Foundation people do it because they're trying to duplicate the functionality of non-free programs. You have an increasing of free software that's written by people who have a business model specifically around Open Source, for example developers who are funded by RedHat or VA Systems, or for that matter, O'Reilly--we funded a fair amount of recent work on Perl. Not only do we pay for Larry Wall, we paid for a major effort to re-integrate the Windows and Unix versions of Perl. We also worked to get the Unicode and XML support into Perl, and so on.

But in general, an awful lot of Open Source software actually was written by people who were trying to solve some particular problem they had. Larry Wall didn't write Perl in a vacuum, he did it to solve problems in his job. You know, Larry Augustine just mentioned the work on the printing sub-system in Cisco. This was a business problem for Cisco. You look at the origins of Sendmail, you know, Eric Allman was just trying to solve a problem that he had; he was connected to the ARPANET, all these other researchers at Berkley weren't, he thought it was easier to write a mail forwarding program than to give 700 people accounts on his machine. So software is written to solve a problem that you have; very few people write a book--at least in the software world--to solve their own problem, they do it for the benefit of other people. Now people do do it for reputation, that's certainly true, publishing has always been a reputation game. But a lot of times, the kind of books that people want to write, you have to pay them, and sometimes they want to maximise the revenue. And unlike Richard, I'm a great believer in `whatever works'. You know, my goal is to maximize the amount of useful information.

I also don't believe that necessarily you get a lot of benefit in the same way from making documentation completely free. At the end of the day, a book is its own source code, it's not as though the content is hidden in some way. There's certain a role for people patching things, but people don't, for example, routinely do all their own patches to the Linux kernel, they send in patches which are then integrated by somebody. Well, that already happens with books; we get comments all the time which we're integrating, so we already have a lot of the benefits of an open process with books regardless of whether or not they're redistributed for free, and you get some negatives.

That being said, we're doing a lot of experimentation, we're basically trying it with a number of books and we're looking to see what happens. If the sales are significantly less, then we'll consider that a bug.

In a certain way, again, a lot has to do with `what are the objectives of the author?' For example, if they're saying `we want this program to spread, we want it to have good documentation'--for example, that's one of the issues with Samba, although I have to say, the book is being written by an O'Reilly staff author, so we're in a better position to say `gee, we're willing to fund this' than perhaps an author who says `gee, over here I have consulting work and over here I have writing a book, which one do I do?'

I guess that's a long enough answer to that question.

LP: Yeah, I notice that the DocBook reference is going to be released free?

Tim O'Reilly:

That's right, the DocBook stuff, yes. Actually that's another project that we funded originally. We started working with DocBook back in 1988, in fact, or 1989.

LP: All around the conference we're seen more and more commercial software, closed source software for Linux. Do you think that as Linux develops its business reputation, we're going to see more of this?

Tim O'Reilly:

Yes, oh, absolutely. I think we're going to see a huge amount of closed source software ported to Linux, and what will be really interesting is whether that software ends up with real Open Source competitors. You know, there's some sense in which Linux as a platform and Linux as an Open Source development technology are not the same, and that boundary will be blurred, I guarantee you.

I guess I see Open Source as a wave front, a weather front that's moving through the industry, and technologies that are most productively Open Source at the front end where you're really doing new and interesting things. People just don't care after a while.

I'm not sure that all those technologies that people are offering, whether it matters if they are open or closed. Well, it's going to be interesting--you look, for example, at Oracle on Linux, you know, that will compete with postgres SQL or MySQL. We'll see how it goes.

Most Popular LinuxPlanet Stories