The following correction was printed in the Guardian's Corrections and clarifications column, Saturday November 4 2006
Google's open source chief, Chris DiBona, completed a degree in computer science after he joined Google. We described his degree as "unfinished" in the interview below. We also quoted him as saying that at Google "We won't allow the use of GPL 3 code," when he actually said "We won't allow the use of GPL 3 addenda in our code."
Technology Guardian: You were behind Google Code Search (google.com/codesearch), which lets people search within public source code?
Chris DiBona: Yes, and it's been really funny because a lot of people have come up with clever queries to find comments in code like "this is broken" or "shouldn't work". It's been fun.
TG: What do you do from day to day?
CD: I work with a small group on compliance with open source licences that we use. And on the APIs, to make sure that they are properly open, and licensed, so that we don't turn round and tell people "you can't do that" and they say "but you never told us that before".
TG: Does Google use its own custom-built version of Linux?
CD: Google uses the Linux kernel on our own hardware; we make our own hardware, it's very much hand-built, Intel chips and stuff. Plus we have a stack of software running on top of that. We don't use Apache for our web server, but we have some of the Apache libraries in there. And here's the great thing about not running on Windows. If you want to change your operating system and you're using Linux, well, you can just do it. But imagine if we had to tell people with whom we're competing and who don't like us very much how many machines we were using, and how much we were using them, well ... the thing is with open source and with Linux, we can control our own destiny. It's a very powerful position to be in.
TG: How much would it cost in licences if Google ran on Windows?
CD: We've never said how many machines we have, there are estimates out there but I'm not going to say if they're right or not, but you know, think of a big number, a big number, and then multiply that by $1,000 (£526) each for a copy of Windows Server. It's safe to say it would be hundreds of millions of dollars at least.
TG: Do you use the BSD licence, GPL, or what?
CD: We prefer now to use the Apache licence; the BSD licence is pretty good, too. They are readily comprehensible. With open source you can use other licences; we have released code under the GPL, but we're not religious about using it.
For example, we released a library to interact with Google Talk under the LGPL and BSD licences. The thing about the Apache licence, though, is that it's really easy to use. The GPL says that if you use it and link something to it, then you have to make that [thing you linked to] available as well.
The Apache licence says here's some code, do whatever you want with it, but just preserve the notice in the program saying that you are using code from Apache. That's easy to comply with. The problem with open source software is that tracking its use can be very difficult sometimes. At Google we have the resources to track how a piece of open source software gets used within the organisation, but many smaller organisations don't.
TG: Is that holding back adoption of open source?
CD: No, I don't think so. For all the challenges of the GPL, we think that about 40% of open source software is licensed under the GPL.
TG: What do you think of Larry Ellison bidding to undermine Red Hat's business model by offering to support the Red Hat version of Linux, through Oracle?
CD: It's fascinating to watch. Although this isn't new for Ellison: he did a lot in 1998 saying that he would support Linux for Oracle. So Larry Ellison has been playing with Linux going way back.
It fits his strategy goals: I think Oracle wanted to control their destiny too, so they wanted to have Oracle on Windows, Oracle on all sorts of platforms. It's hard on Red Hat's stock. But we'll see. It's not trivial to support all that stuff. Though Oracle, I think, makes more money from services than licences on its actual software.
TG: Why are you reading operating system licences if you're not a lawyer?
CD: I went to Google for an interview and they asked me "if you worked here, what would you do?" And I said I'd make sure the licences were in compliance. They said OK, and I got us into a very good place in terms of compliance. Then the founders asked me to solve a problem: that computer science students weren't doing any programming over the summer vacation.
They gave me a target number of students to get doing stuff, so I worked out how much you'd need, and that produced the "Summer of Code" (code.google.com/soc). The first year was 2005, which cost $2m (£1.05m) and got 400 students, and this year it cost $3m and got 630 students.
TG: You said in another interview you were impressed by one entry last year, a database of genes. What was there that made you say 'wow' this year?
CD: The most powerful one, the most important, was to enhance libgaim, which is a library used in instant messaging (IM) systems. The person added QQ support, which is the biggest IM system in Asia. That's important because we can now talk to friends in China.
The great thing about that project is that it means that library becomes part of 20 IM programs. But there were so many good ones this year. They came from 90 countries, 70 currencies - this matters, you'll see - 456 schools ... British entries, there were 20 accepted from about, I think, 150.
TG: Google famously says employees should spend 20% of their time on a pet project - what's yours?
CD: Oh, it works out more like 4%, because I'm so busy. It sounds frivolous ... I try to get new imagery into Google Earth, such as the Firefox crop circles, and the Burning Man festival. Unique aerial things, basically. But they only show up at a particular magnification.
TG: What else are you working on?
CD: We're also working on large scale data repositories, with academics, to take large scale datasets and move them between people; and we keep a copy of the data. There's a digital sky archive, which is about eight to 13 terabytes. There's a lot of schools where if you tried to download 8TB it would take forever. These, you can just load them into a big hard drive and take them over.
TG: Do you talk to other companies like Amazon or eBay about how they handle lots of data and open source?
CD: They always want to know what Google does. But I don't talk to many very large companies any more. If Amazon or eBay or Yahoo! called, I wouldn't turn them down. But everybody is an open source user.
TG: What's Google's position on version 3 of the GPL?
CD: We worry that the new version is going to reduce the influence of the Free Software Foundation on open source, and that's bad. The provisions in v3 make it extremely hard to maintain compliance. That's because it allows the addition of free-form written addenda to the licence with specific permissions.Let's say I'm company A and I release my code under GPL 3, and say, "Here it is, but acknowledge my trademark." Now, what if I don't want to acknowledge your trademark? I've got to remove that code you put in. As people remove and add addenda, the software itself becomes extremely difficult to understand, and functionally incompatible.
TG: It sounds like the tower game where you pull wooden blocks out of the stack and somehow it still stands up.
CD: Yeah, and software's like that anyway. It would be one thing if there were bitflips, like the Creative Commons - "license this with attribution", or "without attribution" - but GPL v3 as it stands doesn't. And then there's the DRM [digital rights management] stuff which is a little inconsistent. I'm no fan of DRM, but it's telling people what they can do with their software. It says you can have DRM, but that you must provide the signing keys for it. What's the point of that? You might as well just offer the keys, in which case why have the DRM at all? But at Google we can deal with this: we won't allow the use of GPL 3 code. That's the great thing about software - you don't have to use it, right? You can build you own or just not use it. Nobody's forcing you.
· Curriculum Vitae
Age 35
Education Computer science degree (unfinished), George Mason university, Washington DC
Career
1993 Information management administrator, US State Department 1995 network adminstrator, Howrey & Simon lawyers, Washington
1996 Computer security specialist, Tandem fault-tolerant computers
1998 Private consultant in cryptography
1998 Marketing executive, VA Linux and editor/author Slashdot
2003 Co-founded gaming development company Damage Studios and Konstrux Technologies
2005 Google, open source program manager
Family Married, one child (and one expected)
Blog http://egofood.blogspot.com/
· If you'd like to comment on any aspect of Technology Guardian, send your emails to tech@theguardian.com