IDC's Market-Share Tally Still Low-Balls GNU/Linux

Michael Singer writes an interesting article today, "Linux to Grow Steady for Next Four Years," for Jupitermedia's Internetnews.com. He cooly states, "Linux is much more popular in the enterprise than previously thought, according to a new report out today." Given how grossly this understates the truth about GNU/Linux and its prospects, such tempered optimism deserves a cascading laugh track.

Mr. Singer properly cites the day's new information: IDC projects 2008 revenues for "desktops, servers and packaged software running on Linux" at $35 billion, and they project a 2008 server market-share split of 26% Linux and 63% Windows. He notes that these are better than IDC had previously expected for Linux.

Notwithstanding that Solaris 10 & OpenSolaris will together probably have somewhere over 10% of that split by 2008, these are anemic numbers based on what we practitioners understand about GNU/Linux usage. It is, therefore, ironic that the data are expected to elicit pleasure. We know, for instance, that Apache-on-Linux actually commands 67% of active Web server installations, as of Netcraft's December 1st, 2004 snapshot (based on responses of over 56 million Web sites in the Netcraft Web Server Survey). That's 67% TODAY, not 27% in 2008. This fact--that Apache-on-Linux soundly dominates the duty on working Web servers--is not disputed by critical observers. But the repetition of bent statistics is one way the incumbant has always benefited from disinformation.

IDC admittedly is not attempting to track usage. IDC counts revenues and units of software sold. They look at data originating from barcode scanners and inventory systems at places like Frye's or CompUSA and by tracking OS's and applications bundled by Dell, hp and others. What they do not count is software like the many Whitebox Linux, SuSE, RHEL or Fedora distros I put on client machines or otherwise were burned by thousands of IT staff around the world this year and installed on PC's or servers doing real work in government agencies (including the Department of Defense and the National Security Agency), universities, schools and companies of all shapes and sizes. Note that many of these GNU/Linux installations went on boxes that once had Windows installed, Windows instances which no longer exist but are still being claimed by Microsoft and IDC, further skewing the reported data.

The problem with the IDC data is that there is no systematic demand for accurate research about Linux--no one, apparently, paying for the information. Dad always told me that you get what you pay for.

Since you are wondering, the guy to complain to is Dan Kuznetsky in IDC Research. To be honest, there are few technology research people more professional than Dan Kuznetsky, or more interested in getting a true picture from the available data. Dan works hard to model the public information and in my estimation really cares about what he does.

Trouble is that the people who pay for IDC data are Wall Street firms, publishing entities and companies wedded to the old way of thinking, i.e., that open source is a marginal phenomenon. Quite a few of the consumers of IDC research can't be bothered to hear anything other than the party line about Linux, not least because it's viewed, correctly, as an un-investable segment of the IT industry.

This is a paradox. It's troubling because, if decision makers knew how pervasive the use of GNU/Linux actually is, they would accelerate their inquiry and adoption plans for open software standards like GNU/Linux.

This suggests another unasked question. Under what circumstances can Microsoft's Income Statement continue to tread water while GNU/Linux and other important FLOSS examples like Mozilla Firefox and OpenOffice grow in usage by over 100% year-on-year? The answer would be something like, software usage on aggregate is growing much faster than conventional measures say about software markets. That growth--perhaps all of it--is being supplied by Free Software.

This is the unreported revolution. Microsoft can churn its old customers and stay afloat but users in places like Dehli, Kuala Lampur, Bejing, Christchurch, Soweto, Munich, The Bronx and Brixton are being smarter about their IT consumption.

As Christian Einfeldt (via Maddog) said (paraphrasing), "Five-odd-billion people haven't chosen their operating system yet."