Oral History of Linus Torvalds
CHM Ref: X4147.2008 © 2008 Computer History Museum Page 24 of 41
Torvalds: That
was fairly early, yeah. For some reason people think-- I have this name of being a nice
person because people, I don’t know. But at the same time since the beginning I’ve had very strong
opinions, and I’ve always enjoyed arguing with people, so my way of writing emails and my way of
expressing myself is actually not all that pleasant. I am argumentative, and I don’t like being proven
wrong, but I’m
pretty happy to say something pretty extreme and then somebody points something out
and I said “Okay, you’re right. I was wrong. It’s not quite that extreme.” So we had a lot of flame
-wars
[ph?] over the years. We still have a lot of flame-wars. The most famous one is probably the first one
when I told basically reacted badly to Andrew Tanenbaum telling me that [what] I did was worthless and
bad.
Booch: He had strong opinions too.
Torvalds: Yeah, he had strong opinions too. I just replied not very politel
y saying that “He shouldn’t
complain about other people’s code when his own code is so horrible.”
Booch: He did say some provocative things.
Torvalds: I
answered in my not-so-
polite way. Sometimes it’s me who starts it so I’m not going to
-- in fact
we’ve
met since and we’ve had email correspondence since. I actually still think his book is one of the
best books about introductions to operating systems. It’s very down
-to-earth and actually having
examples is a great way to teach people in general, and a lot of other operating system books are not.
They’re talking about concepts and not explaining them very well.
Booch: Perhaps it would be fair to say you were both very passionate people.
Torvalds: We are. Yeah.
Booch: You had a similar run in with Richard Stallman as I understand over the whole
things.
Torvalds: Also another very passionate person.
Booch: Yes, Richard is indeed.
Torvalds:
Yes. You know, we’ve never really gotten along, although we’ve usually been able to not butt
heads too s
trongly either. So with Richard the problem has always been that I’ve been a plodding
engineer in Richard’s mind, and Richard to me has always been this religious fanatic. We’ve clashed over
that a few times.
Booch:
You’ve clashed over naming as well too
, I understand.
Torvalds:
I tried to let that thing go, and he wanted everybody to say “
GNU
Linux” instead of just “Linux.”
I said “Whatever. You can call it whatever you want. I’m not going to go there.” And I let that whole thing
go. I don’t think I clashed too much over that. We’ve clashed over other things. Engineers or software
Oral History of Linus Torvalds
CHM Ref: X4147.2008 © 2008 Computer History Museum Page 25 of 41
programmers in general seem to be pretty headstrong, and then when it comes to things like licensing,
people get really excited about that.
Booch: And then as I look at the next few years there was this growth pushed by commercialization
because it really took off over the coming years. You had Debian, you have Red Hat… come on to the
scene. You still however were still pretty much in control of the Kernel around that time.
Torvalds: Yeah, I mean part of it was again that everybody was perfectly happy having somebody to
integrate it all. And a lot of the work that the distributions ended up doing and all the commercial efforts
ended up doing-- I mean much of it was in the Kerne
l, but at the same time a lot of it wasn’t. A lot of it was
in making installation easy. A lot of it was in the programs around the Kernel. And sometimes those
programs needed Kernel facilities so that obviously there was this feedback cycle. But there was never
really any conflict there. And part of it was that I very consciously did not want to work for a Linux
company. I didn’t want to make it so that let’s say I’d work for Red Hat and then that would
-- even if I
would have stayed neutral personally, it would still have given this appearance of non-neutrality.
Booch: You were certainly asked, I imagine, but you resisted their advances?
Torvalds:
I think everybody was so happy with the situation that even the distributions don’t
--
I didn’t
think they really even wanted me to work for them because they see the advantages too. And I was
actually-- I got this offer from Transmeta [Corporation] which suited me perfectly because the whole
reason I got into doing Linux in the first place was that I was interested in low-level CPU details and here
comes a CPU company and says hey, “Do you want to work for us and by the way, we’re even lower level
than anybody else. The things you need to do are kind of below the surface.”
Booch: What year would Transmeta have been?
Torvalds:
I started in ‘97 and Transmeta was founded I think in ‘95, but when I joined it was really pretty
small.
Booch: And they pretty much gave you license to just go do Linux, basically?
Torvalds:
It wasn’t even that. Part of my contract said that Linux is not my job. I’m allowed to do it but
when I do it, even if I do it on work time, it’s still incorporated [ph?] to me. And that was just again
-- it had
already grown big enough, and it was important enough to me personally that I wanted to make it clear
that this was my project and I could do Linux on Transmeta time because people were already back then
using Linux a lot in the whole EDA business, so doing all--
when you’re a chip company, these days Linux
is one of the major platforms for doing d
esign, but it was already becoming fairly popular. If you’re a
highly technical person and you wanted a Unix workstation for free or cheap, you’d go with Linux.
Booch:
In fact, in the ‘90s when did you discover or have this revelation that there was this
crossover that
all of a sudden people [were] using Linux for real things? Did that ever hit you in that timeframe?
Torvalds:
Actually, it wasn’t ever a big deal. I mean I think when Oracle announced that they were
porting Oracle to Linux, I forget when th
at was, but that was like around ‘99 or something like this. Maybe
it was earlier. That was a big deal because that was like Oracle was kind of the definition of a Unix