Oral History of Linus Torvalds
CHM Ref: X4147.2008 © 2008 Computer History Museum Page 12 of 41
Booch: You use a couple of phrases I wanted to key upon. When you say you connected to the
university's computers, of course connecting back then is very different than connecting now. And you
mentioned the Internet. The Internet back then was very different than it is now. Tell us about it.
Torvalds: So, the connecting part was over a MODEM and dial-up.
Booch: Do you remember the speed of your MODEM?
Torvalds: I started at 300 baud, so it wasn't really horrible, but I pretty quickly upgraded to 2400. I
stayed with 2400 for awhile. I mean the Internet was the same. It wasn't graphical, but I mean…
Booch
: So you were on the newsgroups and such…
Torvalds: Yeah, I was on the newsgroups.
Booch: What newsgroups did you hang out in, primarily?
Torvalds: I followed comp.arch. [a computer architecture newsgroup
–
Ed.] One my time-waster
newsgroups was alt.fan.warlord, which was this-- I think it may still exist as a newsgroup, although I
guess nobody sees the alt category anymore because it's all child pornography, whatever. But back then,
alt.fan.warlord was this newsgroup where you made fun of people's signatures, right. I don't think
signature files really exist anymore, but back then, there were people who had like these 40 line signature
files with really ugly, butt-ugly ASCII graphics in them like, and the whole newsgroup was just about
finding people with the most outrageous signature files and scoring them by how bad they were.
Booch: Did you have an outrageous signature file?
Torvalds: No, no, no, I never had one. I think being, seeing everybody laugh at other people's signature
files made sure that you kept to-- the official limits was, if it's more than four lines long, it's in bad taste.
But really, if it was more than one line, it was pretty debatable. It was all just a time-wasting thing for me.
I mean, I…
Booch: You were also into a bit of computer gaming back then, as well, weren't you? Prince of Persia, if
I'm not mistaken?
Torvalds: Well, that was actually before I-- so before my machine was useful, before I had MINIX on it,
there was nothing I really could do with it. I was doing some programming, but it was in DOS and it was
not-- so for a couple of months there, my computer was basically this doorstop waiting for Unix to arrive in
the form of MINIX. And during that time, I played Prince of Persia
and…
Booch: Were you very good?
Oral History of Linus Torvalds
CHM Ref: X4147.2008 © 2008 Computer History Museum Page 13 of 41
Torvalds: Not really, no. I was okay. I mean if you play anything for two months, you're good, I mean.
But I was never that interested in gaming. But it was actually one of the few times in my life when I
played a game for more than a week or two. So…
Booch: Let's take a break here, we need to switch tapes. We'll catch our breath and from there, we'll
move…
END OF TAPE 1 / BEGINNING OF TAPE 2
Booch:
…interesting juncture here, because, let's see, January '91 you bought your PC and you started
in earnest working upon it. And as you were telling me, working from the terminal emulation program
wanting more and more. So talk to me about how you began to start writing an operating system? What
led you to that? What form did it take in those early days?
Torvalds: Well, once I got to realize that I had to expand in my terminal emulation package, all of that
had been bringing in an assembler initially. And I realized, okay, it was fine for getting to really know the
low-level CPU stuff and it was fine for the small things I needed to do for the terminator emulation
package. But if I wanted to expand it and actually do a file system and do things like save and uploading
to the university, I needed to just turn things around a bit and not work at that lower level. And I kept most
of the low-level code, but I started working on trying to-- reorganizing the things so that it was-- acted
more like a real operating system. And I had obviously been reading operating system books anyway,
since that's how I knew about MINIX in the first place. So…
Booch: So there was Andrew's book, and were there others?
Torvalds: That was Andrew's book. There was Bach what is his first name [Maurice], I forget, Design of
the Unix Operating System, or Design and-- well, was it? Yeah, something like that. It's a long time ago.
My mind is going, Dave. Those were the two main ones. Tanenbaum's book about MINIX, which was
kind of nice for the concepts, and then Bach's book about Unix, which was more about the practical
implementation of Unix and listed the actual system calls. In fact, at the end Bach's book is a list of
system calls and my copy of that has small stars next to them because I started going through the system
calls. And as I implemented one I would put a star next to it and say, okay, done, what's the next one?
And it wasn’t really organized, but it was, like, I was literall
y looking at that book to see, okay, what do I
need to write so that I can then move my terminal emulation code up a layer so that it would actually be a
real process and it could do read-writes and actually open files and things like that. And that took a few
months. That was also actually when I started talking to people about this project, because I had the
book, the Bach book. But that was actually a fairly old book even at the time. And what I wanted to do, I
wanted to be more standards-oriented with more modern standards and POSIX in particular. And I found
some of it-- I mean I may have used the Bach book as the kind of basis. But then, for example, the Sun
machines at the university were much more modern and had more of the modern interfaces. So the
amount of [man, manual] pages there helped me some. But I was starting to ask around. So where's the
POSIX standard for what are your system calls really supposed to do, instead of having to try to figure it
out from really a book or from somebody else's man pages that are-- I mean, that is probably better, but it
was still not what I was really looking for. And I was then approached by one of the people at Helsinki
University of Technology who said that, "I'm sorry, but the POSIX standard is not actually available for
free. You have to pay…," like, I forget what it was. But it was 50 bucks or something like that. And the