% fortune -ae paul murphy

Linux: The meaning of success

A couple of weeks ago we - meaning the talkback contributors and I - had a fascinating series of discussions about some of the barriers facing Linux. This week I'd like to put together some drafts for pieces of a future essay under some title like "A prescription for Linux" to be offered to the larger zdnet audience at some later date.

The first big issue to address in a prescriptive effort like that would involve the definition of success: what is it we want to see happen?

I know what I'd like to see: a competitive software market with people choosing their operating systems and tools by trading off cost against functionality. In other words, a rational market in which the total cost of a decision to use something reflects that product's serviceability, there are clear business rewards for excellence, and any monopolies that get established are based on temporary technical advantages.

Right now the market doesn't look like that, and one consequence is that Linux seems to have nowhere near the market share it was expected to have by now - in fact, we don't count installs properly (and there's something that has to be fixed) but the numbers we do have suggest that Linux has achieved less than half the late 2005 market share predicted for it five years ago. Thus, if I have it right, most 1999 growth forecasts for Linux ranged at or above 35% per year, by 2003 that had fallen to about 25%, and today most estimates are in the 15% range.

(As an aside: does anyone know of an honest, public, compilation of data on Linux installations and major player forecasts? - some of the ones I found look, in retrospect, loony to the point of self-parody.)

So what happened? why did the oomph disappear from the Linux market sometime in late 2002 and early 2003?

This is not a Linux quality issue, nor is it related to the availability of third party software -that just follows the OS market. If we define good in terms of reliability, suitability to purpose, and flexibility in user hands Linux isn't the market technology leader -Solaris and the BSDs are - but it's more than good enough - has been since SuSe 7.0 in 2000 - to compete very effectively with Microsoft's Windows brand products for both server and desktop configurations.

My guess is that what happened was IBM. Specifically, I believe that the accellerating market growth Linux experienced from about 1997 to mid 2002 was largely due to the mass media's endorsement of Linux as a kind of politically correct Unix - something that was driven, as I've explained elsewhere, by the "myth of Linus" - the socially correct image of Linus Torvalds breaking out of Finland's Sherwood Forest to single handedly defeat the evil international corporations dominating computing. It was nonsense to begin with, of course, but it became untenable for even the most determined media carpet bagger when IBM started to intensively publicize its worldwide commitment to Linux.

Regardless of how it happened, however, what we seem to be down to now is two markets, both defined by negatives: the Unix haters and the Microsoft haters. The smaller of these, populated mainly by people facing a forced march away from an operating system environment that's either dead or dying like MPE, OS/400 and AIX, or intentionally orphaned like VMS, HP-UX, and Tru64, is characterized by an "anything but Unix" attitude and a willingness to pretend that Linux isn't Unix in order to satisfice both their needs and their prejudices. The larger one, however, consists almost entirely of Microsoft haters - some individuals, but mostly governments - who are against Microsoft rather than for Linux.

Both groups, of course, represent market distortions and are really rather sad, but this problem is something the community has to first face up to, and then do something about, if Linux is to become a key player in a real competitive market.

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.