Famous Last Words: Ed Muth on Linux

A Bit of History

"UNIX is a simple language, easy to understand, easy to get started with. It's great for students, great for somewhat casual users, and it's great for interchanging programs between different machines. And so, because of its popularity in these markets, we support it. We have good UNIX on VAX and good UNIX on PDP-11s.

"It is our belief, however, that serious professional users will run out of things they can do with UNIX. They'll want a real system and will end up doing VMS when they get to be serious about programming."

Weak Value

"The more I study Linux, the weaker I think the value proposition is to consumers."

The value proposition today for Linux consumers is stronger than it was yesterday, and it will be even stronger tomorrow. This is constantly and unalterably true because Linux is consumer-built. The more it's used, the more it's improved. The more consumers there are who use Linux, the faster it improves; the more Linux improves, the faster consumers adopt it. This is positive feedback; this is exponential growth.

Seriously, just how much significant improvement has there been in the value proposition of Windows 95/98/NT/2000 in the last four years?

(Study Linux as long as you want, Ed. Take your time. Another four years ought to be enough.)

Shrink Wrap

"Thirty years ago, everything was custom, and the trend for the last three decades has been to make more software off-the-shelf," Muth said. A "design-by-community" ethos will not work in corporations, he said, where even highly customized applications such as ERP systems include much shrink-wrapped technology.

Now, all else being equal (support, features, quality, etc.), which "shrink-wrapped" downloadable package is going to have a competitive advantage in the future? The one that comes with a large price tag and restrictive license terms, or the one that comes free with unlimited distribution and modification rights?

(Ed, if you are under the impression that "design-by-community" software doesn't come in prepackaged distributions, you need to go back and spend some more time studying. Another four years ought to be enough.)

Integration

"People want more integration. They want to take a bar chart from Excel and put it in Word. On the server side they want strong queuing and security. This is all done through integration. Linux has a low degree of integration. Linux is basically a big step backward for those two reasons plus others."

Integration is a big deal for Microsoft for two main reasons:

• Free value.
  Integration means never having to build it twice. Once you create the integration infrastructure, interoperable widgets, and graphical doodads for your first product, all that stuff gets transferred over to your second product for free. Furthermore, because of the interoperability, the first and second product together have a value greater than the sum of the parts. Your costs go down, and the market value goes up. The difference goes to the bank.
• Consumer lock-in.
  If you can control the integration framework, you can keep consumers who buy into it from straying off the ranch. Given the choice between a mediocre but integrated datadiddler, and a better, but completely incompatible datadiddler, they'll go for the integrated datadiddler nine times out of ten.

This is the kind of integration that Linux has emphasized to this point, because it's the kind that mattered most to the people who were building the system. Now that programmers are turning their attention to user interface issues, they're bringing along the same attitude. Developers for the two contending desktop standards (KDE and GNOME) are coordinating efforts on interoperable drag and drop, for example, so that applications from one framework can integrate with applications from the other.

And that's the key point: promiscuous, ecumenical integration delivers the same benefits to the consumer as Microsoft integration, but without the exploitation and without the handcuffs. And that's a value proposition consumers can relate to.

(But Ed, I urge you to take comfort from the immaturity of the Linux desktop. Relax. Don't worry.)

No Technical Comfort

"The problem with that is there are fewer applications available for Linux, there's no long-term development road map, and there's a higher technical risk in using it," he said. "You could cut Linux some slack if it were sharply lower in cost per transaction than NT, but that's not the case."

Linux isn't a product. Linux is an organic part of a software ecosystem. The Microsoft "long-term development road map" had everyone using MSN in the year 2000. Then the ecosystem produced CERN httpd and NCSA mosaic, and we all know what happened next.

Linux doesn't develop according to a proprietary product plan. Linux just automatically grows towards providing the greatest consumer value, because it is consumer built. If anyone anywhere in the ecosystem has a good and valuable idea, the consumers will absorb it into Linux without prejudice.

Too Much Hype

"We're all in the business of wanting the customer to have the information needed to make informed choices. We haven't seen a flavor of Linux coverage that addresses that. Some criticalness is needed. Some people say positive things about Linux when their message is anti-Microsoft, but I wonder, in 36 months is this the next [Network Computer] or is it a viable OS? We don't see people question the Linux numbers."

So there is a hysteresis in press coverage. If it isn't a spectacle, the press will ignore it completely. If it is a spectacle, the press will pump it up into an even bigger spectacle.

While the latter is currently the case with Linux, however, this is still completely beside the point.

Unlike the NC, Java, or even Microsoft Windows, Linux crossed the media's attention threshold with no marketing strategy, no multi-million dollar advertising campaign, and no hype machine. It became the world's second-fastest growing operating system purely on the good word of mouth of satisfied consumers.

And if not one single word--true, false, or shaded gray--were ever reported about Linux ever again, Linux would still continue its strong growth, for exactly the same reasons that got it this far.

Shelfware

"We feel that 2 to 20 percent of Linux shipments turn out to be 'shelfware'" .

Consider that every single Linux shipment can be legally installed on an unlimited number of systems. If only 20 percent of Linux shipments are installed on only two computers, this in itself makes Ed's observation completely moot.

But, again, this misses the point entirely. Every single Linux shipment puts the source code and development tools into the hands of another potential developer. That means that 80% of Linux shipments are going to people who are using Linux to do something, people who have the power to make Linux to do that something better, faster, or more reliably if they need to. And if they do so, every other Linux user benefits.

Even if the installed Linux base were frozen tomorrow, the value of Linux would continue to grow at its current rapid pace. The fact that large numbers of Linux distributions are being shipped to satisfied paying customers ensures that the pace of development will only increase in the future.

No Free Lunch

"I find it hard to believe that some of the best computer scientists in the world will want to do their work for free," he said. "Without a long-term technical road map, without multimillion-dollar test labs, someone wants me to believe these visionary programmers and developers will want to do the best work of their lives and then give it away. I do not believe in that vision of the future."

When a small company manages to bring a competitive product to market without the benefit of multi-million dollar test labs, no one is surprised when Microsoft buys them out (Ed's vision of the future notwithstanding). And hardly anyone notices anymore when another vast swath of Microsoft's own long-term technical road map evaporates.

The astounding misconception here has to do with where "free" software actually comes from. Ed's message is that "free" software doesn't make sense, and particularly doesn't make economic sense. At least not in his vision of the future.

Out in the real world, however, the best computer scientists in the world give away the results of their work for a number of very sensible reasons:

• They get paid to.
  This covers university professors, graduate students, and researchers at public and private research organizations (unquestionably some of the best computer scientists in the world). This also covers employees of many successful companies for which free software is a part of the business model. There is the "give away the razors" model of companies such as Cygnus. There is the "grow the market" model of companies such as O'Reilly. There is the "commoditize the rest of the value chain" strategy of companies such as Oracle and IBM. And there is the good, old-fashioned, "hurt the competition" strategy of Corel.
• If you want something done right, you have to do it yourself.
  Perhaps the majority of design-by-community software originates because the existing tools for the job didn't meet someone's price/performance standards. Tools are not products. Tools are what you use to produce your products. When someone develops their own tool (as opposed to their own product), they can either sit on it (which helps no one), or they can contribute it to the community software ecosystem (in which case they will directly benefit from any improvements made to the tool by the community). To put it in concrete terms: networking software companies don't develop Samba; network administrators develop Samba.
• It's a cheap substitute for an expensive education.
  Many very good and experienced programmers work on community-based software projects in their spare time to keep their skill-set sharp and up to date. This translates directly into economic benefits at promotion and job-hunting time.
• It feeds the ego.
  If Ed wants to argue that it is foolhardy to develop mission-critical business software based on ego gratification, then I merely offer Dave Cutler as Exhibit "A". Ego-based software development may not be "sensible", but it is firmly based in human nature, so it is at least something you can depend on.