What IBM Doesn't Know

Posted 29 Jun 2001 at 13:42 UTC by brlewis Share This

According to Dan Frye of the IBM Linux Technology Center, three forces come together to make a "disruptive technology" that changes the status quo.

  1. Desire to Learn
  2. Skills
  3. Open Culture

That was one of the points in his keynote address at the Usenix conference in Boston, Massachusetts on 2001-06-28. Any two of these forces will not result in a disruptive technology. He had a slide for each combination of two, demonstrating that combination's inadequacy.

He asserted that the success of Linux stems from these forces coming together, and not from the other factors to which we hackers attribute success. After all, the same sort of thing happened in the 1980s with desktop PCs. The status quo in desktop computing was disrupted as many people started developing and sharing software.

During the question-and-answer (Q/A) session, I pointed out that what created those three forces during the 1980s was necessity: Certainly you have a desire to learn about the only desktop computers cheaply available (IBM-compatible). Out of necessity you'll develop the skills you need on that platform. Out of necessity, you won't want to go it alone in dealing with the platform, so an open culture will develop.

I asked Dan what he thought drove those three forces for free software like Linux. Obviously it wasn't necessity; a platform for desktop software was already available. I knew the answer to my own question, because a previous speaker (GNU's Robert J. Chassell) had already enumerated what lay behind those three forces.

Dan Frye didn't know. He said that extensive research would be needed to determine the causes, as much psychology/sociology as technology was involved. If I recall correctly, he suggested you could get at least one PhD thesis in the course of finding the answer.

I said I thought the answer was actually quite simple, and I could talk with him more about it later. (A Q/A session is not a forum for engaging in lengthy dialogue with the speaker *cough* dan geer *cough* :-) Unfortunately, Frye disappeared before I had a chance to speak with him.

Here's the answer. Where do I pick up my PhD?

  1. Desire to learn stems from freedom to study code. I can look at someone else's work and say, "Hey, I could learn to do that!"
  2. Skills stem from freedom to modify code. The best way I can develop skills is by hacking on an existing project rather than starting from scratch myself.
  3. Open culture stems from freedom to redistribute code. I can collaborate openly with anyone, without any license hassles.

What drives the success of Linux and other free software is freedom. Dan Frye is correct about desire to learn, skills, and open culture coming together to create disruptive technology, but he needs to see that these are only the immediate effects of freedom.

Necessity?, posted 29 Jun 2001 at 17:21 UTC by apenwarr » (Master)

I wouldn't say that these forces came out of necessity for the creation for the PC any more than they did for Linux. Many people chose to use Macintosh or Unix boxes rather than PCs for a long time, so it was clearly not necessary to learn about PCs. You don't need skills on your chosen platform - ask any incompetent Windows sysadmin. And there's more than one way to work with other people. An "open culture" (the way I think of it) is a pretty extreme case of co-operation.

The PC culture was originally sort of open, and sort of not. People (rather, companies) were willing to work together, at least until they got big enough to not have to work together. There was always very little sharing of source code or hardware docs.

As for Linux motivations...

In my experience, desire to learn most definitely does not stem from "ability to study" anything. Rather, it's the other way around. If you really want to learn, you'll find a way to study it. Period. That's why we have reverse engineering, biology, and semiconductor physics. People who aren't motivated won't be much more motivated if you dump lots of new information in front of them. Motivated people, however, might become more motivated in the right conditions.

Skills: I've only sometimes found that modifying other people's code is useful for building skills. Most other people's code is unreadable crap. Code that's well-written (and thus worthy of studying) is often the stuff that I don't need to modify, because it already works.

Open culture: you can have that without the freedom to redistribute code, as long as you have enough collaboration tools (like Usenet, or the old fidonets and other networks of the past). On the other hand, sharing code does happen to make it much easier.

So anyway, I admit that I just shot down your list of motivations without coming up with one of my own. But I don't have a PhD either :)

what's freedom got to do with it?, posted 29 Jun 2001 at 22:33 UTC by dutky » (Journeyer)

Of the three items listed — desire to learn, skills, and an open culture — only the last is driven by underlying freedoms, and even for that, there is a far more important motive force.

In all three cases, the driving force is, essentially, cost. The Linux revolution, as with the personal computer revolution before it, was driven by the reduction in cost of capable hardware, which put into the hands of a huge mass of people, the means with which to explore new and interesting ideas. Similarly, the transition of the internet from an expensive toy for Universities, the government and the military, into a utility for all sorts of people, provided the means by which a vastly more open and comunicative culture could be constructed. The skills, if they were not either inate or a result of the desire to learn, resulted from the several decades of frenzied interest in computer science that preceeded the rise of Linux, which were also a product of the availability of cheap computer hardware to a large number of people.

Certianly, much of the knowledge and skill required to write something like Linux existed for years before Linus came along and did the obvious thing. Linux itself was a response to one of a number of unix clones, which had been available since the mid-eighties (Minix, Xinu, and Coherent are the ones I recall). The thing that made Linux really take off, however, was the ready availability of a capable and inexpensive hardware platform in the 386-PC and a recent drop in the price of memory (which, throughout the late eighties had been in a shortage and priced accordingly). You might even be able to lay some of the blame for the continued failure of the GNU Hurd project at the lack of a reasonable target platform. IBM-AT clones and MMU-less 68000 boxes just don't cut it when compared to VAXen and Sun workstations.

At some lower level, freedom was certainly an issue. Without the freedom to purchase and own whatever computer hardware you could afford, and without the freedom to publish you thoughts on and experiences with that hardware, the personal computer revolution would never have occurred. It is certainly likely that IBM, to this very day, doesn't understand this: they didn't seem to understand it too well when they broke in to personal computing in the early eighties, and it is unlikely that the corporate culture can really grapple with the concept today. Still, all the freedom in the world means very little if you can't exercise it in some concrete way. If the interesting computer hardware is unaffordable, freedom will not help.

<ASIDE> The converse is also true, even if the hardware is affordable, without freedom it doesn't mean much. This is part of what is so offensive about the proliferation of closed hardware — such as the iOpener and the Tivo — even if the hardware is cheap. The most exciting aspect of personal computer, and the aspect that set it apart from other technical endevors, was the way it could be used to realize and inspire the ideas of individuals. Closed hardware, as much as closed software, is a direct threat to individual intelectual liberty because it erects boundries around the types of ideas it is permissible to think or to express. <ASIDE>

Desire to learn stems from freedom to study code?, posted 30 Jun 2001 at 01:08 UTC by joe » (Master)

This logic seems arbitrary. I don't have a desire to learn about butterflies merely because I can go to the library and

oops, posted 30 Jun 2001 at 01:11 UTC by joe » (Master)

Hit post too quickly there. It should end ...read about butterflies. I might want to go the library and read about butterflies if I was interested in butterflies though. This is not a good analogy though, more fool me for hitting the post button before I thought better of it :)

Generalizations, posted 30 Jun 2001 at 19:28 UTC by julian » (Master)

What *IBM* doesn't know... ah, generalizations are great. ;)

critiquing my article, posted 5 Jul 2001 at 17:49 UTC by brlewis » (Journeyer)

In critiquing my article, it's important to keep in mind the context. New technologies are being introduced all the time, but only some of them turn into "disruptive" technologies, changing the landscape. Some technologies gain widespread desire to learn, skills, and an open community. Other technologies don't.

Obviously, many different factors can affect one's desire to learn. But the important question in this context is not which of those factors is the biggest. The important question is which of those factors distinguishes GNU/Linux from technologies that did not gain widespread desire to learn.

SCO Unix with an educational discount provided an affordable, solid technology much sooner than GNU/Linux. But it never became a disruptive technology. I assert that neither price nor superior technology was the differentiating factor. It was freedom.

Desire to learn: Magic tricks look cool, but most people don't want to learn to do magic tricks. They look difficult or impossible (that's the idea), so why waste time trying to learn something so hard? Computer programs can be just as mysterious, but sometimes the source code can be like a book of magic tricks. The curious read it and say, "Hey, I can do that!" It makes the difference between wanting to learn and thinking that learning is out of reach.

Skills: For a case study in how freedom to modify existing programs can bring a lot of non-programmers into programming, follow the comp.infosystems.www.authoring.cgi Usenet group for a while.

Open culture: Freedom to redistribute code is certainly not the only factor that builds open culture, and probably not the biggest. But I still think it's the differentiating factor that made GNU/Linux take off when other technologies didn't.

Re: Generalizations, posted 5 Jul 2001 at 17:51 UTC by brlewis » (Journeyer)

Somehow, "What IBM's Dan Frye Doesn't Know About Disruptive Technologies" just didn't seem as appealing a headline.

More than just "SCO", posted 7 Jul 2001 at 21:02 UTC by argent » (Master)

The availability of UNIX as a common software development platform was key to both the success of the IBM PC and Linux. Why the IBM PC? because MS-DOS was designed to leverage from the growing base of UNIX software. Up until Windows, every new feature in MS-DOS made it more UNIX-like and less CP/M-like. Microsoft shipped a C development system that made it easy to port UNIX software to MS-DOS. Microsoft (later SCO) Xenix became near ubiquitous as the "next step up" from MS-DOS for small businesses before networking. At one time Xenix on the TRS-80 model 16 had more users than any other UNIX system, and if you believe Microsoft's ads more than any other micro- or mini- based multiuser OS.

So you had MS-DOS at the low end, and Xenix at the high end, and a common stream-oriented C API using the same compiler for both.

Why Microsoft abandoned Xenix in the mid-80s, I don't know. I suspect they decided they couldn't control that market, so they'd replace it with something they *did* control.

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!

Share this page