Linus on robots, fame and getting to ‘yes’ with 2.4
Bill versus Linus was helpful then, but this is now...
Torvalds interview part 1 To his millions of fans he's known simply as "Linus". From Southport to Santiago he's celebrated as the man who kick-started a computer revolution. But enough about The Register's publisher Linus Birtles. The only slightly less-famous Linus, Linux kernel creator Linus Torvalds, kindly gave us the best part of an hour of his time for what these days is a very rare interview, at Comdex. Particuarly as he's expecting his third daughter this very week.
What was guiding your decision making in the 2.4 kernel? You had a lot of choices, ranging from SMP to small devices
A lot of it's not so much that it's guiding my decision making - I try actively to avoid having specific goals, and with 2.4 the only goals really were SMP scalability and getting the network stack up to snuff. And that was mainly because of - as you know - Mindcraft. What's happened is that people are working on all kinds of things and some of them made it and some of them didn't, and I get patches from people I trust to do things that I think are really important.
And sometimes the patches were kind of wrong, and I'd say this is an important thing but let's go with it, even though it's flaky at the time - let's make sure we get this in for 2.4. The thing that really guides me is what people work on and what people are interested in.
So it's on a case by case basis, on merit how well executed something is...
Yes it's really how well thought out something is that's more important to me. If someone tells me we're doing something wrong and this is how we should do this, and by the way here's a patch that shows what I'm saying - then if that makes complete sense to me, then fine. The patch doesn't have to be perfect.
But as I'm actively trying to get 2.4 out the door I usually answer 'that makes sense to me but let's wait til 2.5'. If say this patch triggers this bug and the machine does bad things and if I can see this patch won't hurt anything else, then these days I'm much happier. It's my job to be a judge of taste.
And I try to set deadlines for doing a release. And part of why it doesn't work, why it gets delayed, is the fact that I have a deadline is pretty meaningless to convince other people that yes, this deadline should be met. I don't have any way of really enforcing deadlines, the best I can do is kind of change the psychology of the people to make them realise that they want to help to meet the deadline. That's why I often try to say we will aggressively release in one year - which was the plan for 2.4 originally and I was kind of getting people in the mindset.
At any company - every single software project ever has been late, right? People kind of have deadlines but that they don't really become deadlines until you internalise the fact.
A year ago you had a prominent role doing keynotes, and kind of evangelising Linux to a new audience. Do you feel that people understand what it's all about and you let other people do that now?
It's not that I think that the work is done. I'm very unmotivated to do evangelism - that's not what I do, basically. Most of the questions that people have now these days are questions that I'm the not right person to answer. Most of the questions these days are no longer about the kernel. That and the fact that I've been doing it for too long means I'm trying to distance myself, and let other people jump in.
Did you think that the Linus personality cult was helpful?
It was helpful in the sense that if you think about the press about two years ago - even a year ago - with the Microsoft Department of Justice thing - it was important when Linux was not that well known. People on the street hadn't heard of Linux and the personal story made it a lot easier to write about. It was Linus versus Bill Gates kind of angle and I think that was very important. Because a lot of stories couldn't have been written. It wouldn't have made sense to anyone who hasn't heard of open source and Linux before.
I think one thing has changed and that I don't feel motivated to go to conferences anymore, is that people do know about open source. Not everybody, but enough without having to drag it through this personalisation process. That's useful.
But everybody wants to be famous right? Everybody wants to make a difference. I wanted to be a scientist. I wanted to be a famous scientist - the Einstein of the 21st Century, something like that - but it was not to be! But I got into programming. So in a sense I dreamed of it.
Do you regret not having so much privacy?
It's not a problem. It's five days a year - it means at a backstage thing like here, I can relax, someone will get me a Coke, and when I go home I can go back to changing diapers again! It's not like I don't have a private life in general.
Actually most of the stuff is fairly positive. I don't like going to conferences because it's just too much stress. I don't like traveling and all these kinds of places you don't know and don't really care about. I mean, why would I come to Las Vegas?
But I like meeting people. It's certainly motivational to know people care what we do (laughs) so I don't have any problems with how does it feel to be famous. It's very nice. And most of the time, I'm not.
There's a lot of interesting free software work - on clusters and clustered file systems - that could really change how people put systems together on a big scale...
It probably will because that's so expensive to do now. That's where open source and an evolutionary approach really shines because it makes something that expensive economically feasible.
But at the same time I'm more excited about the gadgets. Never mind the clusters - the most exciting gadget by far was the robot - so...
Well that's taken care of my next question - what new hardware have you seen that makes use of Linux, that really bakes your cake....
The robot is fun. I love the TiVO, I have it at home. If I had known how useful it was I would have bought it even if it wasn't running Linux, right? The things I like are everyday. To me a workstation running Linux is everyday - that's not surprising and it's cool that it's doing 3D graphics and stuff like that - but the really exciting stuff is the stuff I would never have thought of myself. And I have this suspicion that the people who did think of it are kind of - you know crazy - but that's OK.
Sponsored: Benefits from the lessons learned in HPC