BERLIN: AS PART of our visit to LinuxCon this week we’re going to ask five key players in the Linux story the same 10 questions to get an idea of where Linux has been, where it is and where it’s going.
And who better to start with than Linus Torvalds, the often outspoken creator of Linux itself. Torvalds isn’t actually attending the celebrations this year, but was kind enough to chat to the INQUIRER by email.
What's your first memory of Linux?
It's really hard to say, mainly because it's hard to pinpoint when the project became 'Linux'. Trying to just figure out how the boot sequence worked took me ages. OK, realistically several weeks.
It was my first PC, so not only did I have to figure out how 386 protected mode works (it's not trivial), but the rules for making a bootable floppy and learning how the BIOS worked. But it wasn't exactly 'Linux' at that point.
Or was it when I was going through Maurice Bach's The Design of the UNIX Operating System and starting to mark off the system calls at the end of the book as I implemented them? I don't know.
What do you believe has been Linux's biggest contribution to the world?
You really should ask somebody else. I am a bit biased. I think part of it is how it has allowed others to just build around it.
An OS kernel is such a critical piece of infrastructure in a computer, and everybody needs one, but very few people really care about the kernel itself. They really just want to do something else, whether it be a service, selling some piece of hardware, or writing programs on top.
And sure, there were other kernels around, but realistically they all came with a lot of baggage. Linux really made a lot of things possible for people and companies that just wanted to do something new and something that didn't really have anything to do with a kernel (except for the need).
But I also think that Linux made open source in general a saner environment. You had the 'unconscious open source' people (the BSDs etc) that didn't make a big deal about the openness per se, and you had the rabid 'freedom of software' people who didn't actually care about the technical merits because non-open source was evil and it didn't matter if some proprietary code was better or not. And I, and Linux, came in right in the middle.
I really think that the technology is what matters because without having the best technology, where 'best' obviously includes issues like price and flexibility, why would anybody care?
But at the same time, I do think the open source part matters too, for various reasons, but partly because I think it's the way to get that best technology. Linux wasn't the only such project by any means, but I think Linux very much helped drive that middle road.
When did you start getting paid for doing work on Linux and what was it doing?
Hmm. It was when I moved to OSDL in 2003, I think. That came about because I was starting to prepare for making the 2.6 release while still at Transmeta.
Transmeta had been very supportive of Linux, and used it internally on pretty much all the workstations, but Linux was never my 'real' job there. Although my contract did say that my Linux work was mine. I could work on Linux on work computers, and in fact. some of my early Transmeta days had been about fixing SMP problems that they had.
But the 2.4.x release had been fairly ugly, and I felt that I really had to concentrate pretty much full time on making the 2.6 release, so I told people that I was going to take a year of unpaid leave to just do Linux.
That went around a bit, and a week later I was basically contacted by OSDL to say: 'Hey, we'll pay you to do it,' and it started out as a situation where I was 'on loan' from Transmeta to OSDL instead of just being on unpaid leave.
A year or two passed of me doing Linux in that kind of dual role (still technically at Transmeta but paid by OSDL and doing Linux full time), and it became fairly obvious that that was just how it was going to be. So I'm still here. It's 13 years later, and OSDL is now called Linux Foundation and I'm still doing Linux full time.
What has been the biggest single challenge to Linux's success?
I'm really not a big believer in the 'silver bullet' or 'killer app' or 'single challenge' kind of thing. I think it's way too common as a storyline, but it's an oversimplification to the point of just being complete BS.
Real life, especially in technology, is about a lot of small details. You don't fail or succeed because of one big thing. You fail or succeed because of a lot of small things.
There may be something that stands out for timing reasons or because it made the news, and people will point to it and say: 'That's why it succeeded/failed,' but that's just the story. It's not real.
What's real is all the daily effort by hundreds of people, and the motivation and infrastructure that leads the end result to good things or bad. Sometimes the real story ends up being 'somebody else did it better'.
Sometimes the real story is that you had lots of people who put in a lot of effort on things, and it turns out that nobody really cared about the end result all that much, or that for various reasons it failed.
And it's really not even one single effort. I obviously lead and care about the technical development side, and I'm known for that, but even in just that area we have the stable tree led by Greg KH, which is just the 'next phase' of the technical effort that kind of bridges the development tree towards the different vendor trees (where the different vendors have their own technical trees too).
And that's all the technical development side of just the kernel, and ignores all the support, Q&A, marketing, yadda yadda. And on top of that, you have all the other projects - hardware, software, services - that end up using Linux in some minor or major capacity.
Single challenge? It's not a meaningful question. You can ask 100 different people, and they'll just talk about the one thing that they think is most important.
Some will swear to you that what they care about is the one big roadblock, or the one thing that will drive things to the next level. You should distrust people like that. If somebody tells you that one thing is more important than anything else they are being simple-minded.
What has been its biggest triumph?
What I really end up always being happiest about is how Linux has been able to cover so much more than I ever imagined or even cared about.
I'm fairly well known for saying I use Linux on the desktop, and that that is what I care about most, and it's true. But at the same time, what I'm perhaps happiest about is how Linux isn't just one thing.
I may not be a big embedded guy, and I may not even be that much into supercomputers or all the big iron that is so important to a lot of companies, but I really appreciate, and think it's very important, the diverse things Linux is used for.
When people tell me that they use Linux to do some random thing that I would never have done, and that I think is wonderful but that isn't my thing, that is very satisfying. I love seeing stories of Linux being used in all those situations that I never really intended it for, and that I would never have driven. That is a triumph.
I guess that's a bit strange and even ironic. I think the real triumph of Linux is all the things that I literally had nothing to do with, and that I explicitly had no interest in doing. The fact that others can do that I still think is just wonderful.
Do you believe that Linux has a future on the consumer desktop as well as the mobile and the server?
Yes. That future seems to look different from the one I envisioned but, hey, I guess that goes back to the previous answer. Linux still isn't very big on the 'traditional' desktop, but at the same time it does look like the 'traditional' desktop is becoming more and more like the old 'workstation' market: serious people doing serious things.
The consumer desktop would seem to be very much influenced by the mobile market, and you see more and more people using their phones first, with a tablet or laptop or desktop as a 'back up'.
So I think that Chromebooks seem to be a fairly natural thing, and they do seem to be successful. I think the whole 'Android apps on Chromebooks' is a sign of that whole logic.
Is that what I meant or expected from the Linux desktop? Absolutely not. It's not the same thing at all, but it looks like a real direction for the consumer desktop.
Why does the open source model work so well in a world of profits and shareholders?
I still think it ends up working best in 'infrastructure' things. The kernel is obviously almost pure infrastructure: nobody runs a kernel in order to run a kernel. It's never your primary goal. The kernel is just a piece you need but shouldn't care about.
But everybody cares about it a bit, and the open source model really works very well because pretty much nobody can really afford to go at it on their own. And if they did, nobody else wants to be beholden to them. Just look at the situation with Windows during the 90s.
And it turns out that there's a lot of infrastructure in modern technology. Not just the kernel, not by a long shot. Look at almost any piece of modern technology around you. There's just a lot of small details that have to be done. But if you look at what is actually touted as the distinguishing thing, that is often very much about just the surface issues.
It's the whole 'everybody needs a browser'. Christ, you have browsers in cars, in refrigerators, in ATMs - all these things that you don't even think of as computers.
But it's not just the kernel or the browser, it's simply a lot of software. All those people sell you the shiny new hardware or services, or whatever needs that software to make it all work, but what they are really selling is a fairly specific expertise where the software is just what enables them to do the piece they care about.
“Those commercial entities with profiles and shareholders probably really only care about the one per cent of the code that is their special secret sauce. The 99 per cent they want to make sure is something they can rely on, and make sure nobody else controls too tightly so that nobody else can put a squeeze on them. Open source is perfect for that.
And I completely made up that 99 per cent vs one per cent, but I really do believe it's largely true. I don't believe I'm exaggerating.
This is the first conference since Linux and Windows became officially 'buds'. How is that partnership going, do you think?
I have to say that at LinuxCon NA in August I got the feeling that Microsoft is actually being serious about the whole open source thing.
Don't get me wrong - it's been going on for a long while now where there's been an open source office, and it's not like the whole company has magically changed. But it felt like people really were being pretty sincere. It's interesting.
I used to make fun of Microsoft long ago, and I consciously decided to stop doing that maybe 15 years ago. Part of it was that it had been the 'cheap joke', but part of it was that I just felt that the whole Linux vs Microsoft thing had been overplayed anyway.
It's not like I ever cared personally, and Linux was never some kind of 'anti-Microsoft' thing for me in the first place. But it was such an obvious storyline, and it came up all the time.
I think that part of what has happened is that the storyline has gone away, so people don't even expect Microsoft and Linux to be too much at odds any more.
What does the next 25 years hold for Linux?
Hey, I always answer that the same way: I just don't know, and it's not how I work. I don't do five-year plans. I say I'm a 'plodding engineer' and I mean it.
I look at the problems we have today, and obviously I have some expectations of what hardware and usage will be tomorrow to kind of guide what happens today, but it is very much about the now.
My explicit planning window goes about one release out, so maybe two to three months. Some things are visible a bit further out, but they tend to be the exception and don't have a huge impact on what I do.
Any anecdotes to share?
Too much email writing right now. Just looking forward to clicking that 'send' button. Nothing comes to mind.
We’ll let him off the last answer as the other nine were so good. Another Linux lynchpin tomorrow. In the meantime, thanks so much to Linus for his time in this interview and over the past quarter century. µ
Turns out some companies had fixed it before it came to light
There's still a timeline for Timeline but not this time
'Lack of significant enhancements' is causing lacklustre sales
How SD-WAN can fuel your businesses' digital transformation