Unix and Beyond: An Interview with Ken Thompson | Hacker News Reader

Unix and Beyond: An Interview with Ken Thompson

genius.cat-v.org

51 pointsluriel22 comments

This is one of my favourites. So many nuggets in there. A very nice interview. Programmers like Thompson are the only thing that keeps me interested in computers.

It's really sad that the loudest voices in computing are no longer thinkers like Ken Thompson. Look at some of comments in this thread. Pathetic. It's like Slashdotters insulting W. Richard Stevens after his passing. Extreme stupidity.

How many programmers these days can build from the bottom up? How many can start with a blank canvas?

It's a little troubling to me that he has to work for Google (as much as I love GOOG their work is not exactly stuff of Bell Labs: how can we serve more ads?), but I guess you have to do what you have to do.

The comments about his music collection are great. "Illegal downloading!" :) The legal department just looked the other way.

He also more or less says, in this field, there are really no new ideas. This is hard for today's programmers to accept I guess: the idea it's all been done before. But oh how I wish today's self-proclaimed "productive" programmers - who are alamringly ignorant of history and even contemptuous toward any code does not meet their strange notions of "freshness" - how I wish they would take Thompson's comments to heart.

Now, here's a question: How similar is Go to Sean Dorward's Limbo? If we put them side-by-side how many similarities would we see? If no one answers, I may just do this myself. I think it would be interesting.

If you do compare Limbo to Go, compare them to Alef as well.

Hypothesis: Limbo --> Alef --> Go?

Is this wrong?

Probably Alef -> Limbo -> Go: http://www.youtube.com/watch?v=f6kdp27TYZs&3m45s

> It's a little troubling to me that he has to work for Google (as much as I love GOOG their work is not exactly stuff of Bell Labs: how can we serve more ads?)

google has some very interesting systems problems to be solved that are almost entirely divorced from the fact that the monetary goal is serving ads, or even that the core goal is organising the world's information. the company is seriously attempting to push the boundaries of "how can we make large clusters of people, machines and networks more efficient and productive", and thompson's work on go is squarely in that realm.

(disclaimer: i work for google, but i was a fan of that aspect of the company long before i joined)

I agree with your gentle disagreement with your parent.

Google serves ads on top of search results for the same reason that ATT sent bills for phone service: to make money. If you focus on Google's ad serving as their reason to exist, then you might as well focus on ATT as a billing company, as those are the direct ways that both make money.

And in that vein, Google and ATT (well, the old ATT) are much more alike than different. They both took existing fledgling technology and essentially re-invented it into a profoundly reliable and life changing system; in fact they all but invented a new science.

There are orders of magnitude difference in depth and impact between twisting two wires together and the physical and information science discovered and invented by ATT, and there are equally orders of magnitude difference between serving up an html page of a computer's directory and the physical and information science discovered and invented by Google.

Either one is a fine place for Ken Thompson or anyone else to work.

Good points. But... why did Google kill Google Labs? Why not have a separate entity for basic research, like Bell Labs? There is no shortage of funding.

I do see a major difference between selling ads to advertisers (and organising the world's personal information for profit) and selling phone service(s) to customers, but maybe that's just my perspective - I've been using the telephone and the web much longer than most people working at Google - I've seen how things could be done differently.

You seem to have misunderstood what Google Labs was. It was not an organization doing basic research. It was (crappy) infrastructure for making neat little experiments available to the public without making them production quality.

Yes, I thought of it as a cross between beta testing and A/B testing. Or throwing spaghetti on the wall and seeing what sticks. Certainly not basic or serious research, which is probably going on right now somewhere in Google.

Yeah, I knew that. I didn't mean to imply it was a separate entity for doing basic research. What I meant was the fact they closed it seemed to suggest that the idea of research for its own sake is not really somethng Google is interested in. At least, that how it looks from outside the Googleplex. Inside it may look different. (And yes, I am aware that employees are allowed to publish. But that research is almost always for the ultimate purpose of furthering ad sales, though it may not be immediately obvious to the outside observer.)

It's also interesting how Google employees, at least the ones who post on HN, seem to be isolated from what is going on elsewhere in the company. Is there cross-pollination between functional groups within Google? Great research organisations often have this quality. (Prepare for waxing poetic from young Google employees who think they are changing the world...)

"labs" was an unfortunate name, google labs was about product experiments rather than basic research. what you want is http://en.wikipedia.org/wiki/Google_X_Lab, whence emerged (in the "what has google done for us lately" department) the self-driving car and the augmented-reality glasses.

> Now, here's a question: How similar is Go to Sean Dorward's Limbo? If we put them side-by-side how many similarities would we see? If no one answers, I may just do this myself. I think it would be interesting.

Quite a few, and you might note that Sean Dorward is now also at Google working on Spanner.

And for details on Limbo, see this paper by Dennis Ritchie: http://doc.cat-v.org/inferno/4th_edition/limbo_language/limb... and this one by Kernighan: http://doc.cat-v.org/inferno/4th_edition/limbo_language/desc...

And as others pointed out, is also worth checking Alef: http://doc.cat-v.org/plan_9/2nd_edition/papers/alef/

May 1999. Not exactly news. Chances are high that we already read it 13 years ago. I'd be more interested in an interview about the things that he worked on more recently, such as the Go language, or what he's planning to do on this winter vacation.

Have you seen this one from 2011? http://www.drdobbs.com/open-source/interview-with-ken-thomps...

Thanks for the link. Here's a video from 2012 starring him: http://www.youtube.com/watch?v=sln-gJaURzk Not an interview, but he sometimes says a sentence.

Thompson: I view Linux as something that's not Microsoft—a backlash against Microsoft, no more and no less. I don't think it will be very successful in the long run. I've looked at the source and there are pieces that are good and pieces that are not. A whole bunch of random people have contributed to this source, and the quality varies drastically.

My experience and some of my friends' experience is that Linux is quite unreliable. Microsoft is really unreliable but Linux is worse. In a non-PC environment, it just won't hold up. If you're using it on a single box, that's one thing. But if you want to use Linux in firewalls, gateways, embedded systems, and so on, it has a long way to go.

"My experience and some of my friends' experience is that Linux is quite unreliable."

I don't think any software is "reliable". Well tested/battle hardened? Yes. Reliable? No.

This is the kind of ridiculous statement that comes from Slashdot. Yes for a Mars rover or medical device we might want some insane level of reliability. But reliable enough for Google, IBM and Amazon and thousands of other companies is enough to take the title "reliable operating system", the rest is just playing semantics.

have you used linux in '99?? i did. it was pleasurable but in a certain academic or toy sense (if i had to be productive back then i might even have been prodcutive but not everything worked and it was surely missing many reliabilities). it did go a long way and it is now at a point what thompson would not attribute to it back then. he would surely conclude differently now.

>My experience and some of my friends' experience is that Linux is quite unreliable. Microsoft is really unreliable but Linux is worse. In a non-PC environment, it just won't hold up. If you're using it on a single box, that's one thing. But if you want to use Linux in firewalls, gateways, embedded systems, and so on, it has a long way to go.

Huh? Linux _IS_ used in servers, embedded devices, gateways, firewalls, cellphones, what have you...

What exactly is "unreliable" about it, except a very vague accusation?

in 1999 Linux wasn't really used for big production, but sometime around then the .com/linux bubble happened with a lot of companies selling/using linux as their main product (VALinux/Akamai/Redhat/many others).

I was still using Sun/Solaris machines back in '99 when this was written, which were considerably more reliable from an every day use perspective. Most of the negative comments and opinion came from people who'd worked with industrial grade kit and moved to cheap PCs with Linux on.

To be honest, if you go back that far, Windows NT 4 on Compaq Professional Workstations was probably the most reliable thing out there believe it or not so I'm not sure where the hell these comments came from.

Reply on news.ycombinator.com