Don’t get me wrong.  I like the old model for making money on software. 
A couple of clever people have a bright idea and cobble together a solution on
spare time and credit card debit, releasing their solution to the world and
charging obscene amounts of money for each copy.  That is the essence of
the Silicon Valley dream – digital prostitution (you got it, you sell it, you
still got it).

But of late so much market momentum has shifted to Open Source that even the
oldest and
most cynical in the industry must give pause.  And in one case, we may well
be seeing the beginning of a revolution and a conspiratorial insurgency.

Here is the basic fact of life about the software industry: nobody buys
software.  It is the same as drill bits.  Last year millions of 1/4
inch drill bits were sold in America, but nobody surveyed wanted a 1/4 in drill
bit.  They all wanted a 1/4 inch hole.

Software is the same way.  The operating system is nearly irrelevant as
a purchasable item because what people want is  applications.  If the
application runs only on Macs, people will buy Macs.  Sun Micro built their
fading empire on growing a base of applications, as did Microsoft.  And
people buy application not because they need a CD-ROM filled with highly
organized bits, but because they want to accomplish something using the software
(like surfing the web to learn how to drill 1/4 inch holes).

Recently, IBM — the once and future 900 pound gorilla of the tech industry
— announced that

a client for Lotus Notes would be available on Linux.  The IBM
spokesman (if such creatures can be believed) noted that "he has personally been
‘hammered’ for the last several years at Linux and Lotus Notes conferences on
the issue of when a client would be available for the Linux desktop." 
Perhaps he was being hammered by

IBM employees who have been shoved into using Linux desktops and likely were
disconnected from their groupware grope.

The fact of the matter is that no enterprise will deploy Linux on the desktop
(outside of IT) unless their applications — or
reasonable equivalents
— are available.  The cost savings on software would be many orders of
magnitude less than the lost employee time and productivity.  Thus, common
applications must be ready for any sane CIO (if there is such a thing) to commit
to even a modest test deployment.

But we may well be on the cusp of such a moment. 

OSDL conducted a survey of tech professionals to determine what enterprises need
in a Linux desktop for it to be a practical alternative.  The results
largely matched early work I did with SuSE when "Linux desktop" was a new
concept.  Notice that with few exceptions, the list includes only
applications, and the exceptions are primarily application development oriented:

There are several take-aways here from a marketing and market perspective:

1. The Linux desktop
   
   "whole product" definition is incomplete, but not by much.  The two
   remaining holes are the toughest to fill and will take the longest time.

2. As always, there is a
   
   check-and-egg issue in that vendors don’t want to build Linux desktop
   applications or clients until there is a significant roll-out of Linux
   desktops, and CIOs will not roll out Linux desktops until there are
   sufficient business applications. 
   
   Catch-22.

3. Internal application cannot be built until CIOs get their in-house
   development staffs trained on Linux development, and that won’t happen until
   there is a compelling reason to roll-out Linux desktops … and that won’t
   happen until there are enough external Linux desktop applications (see
   above).  Catch-23?

(Note: This all ignores the elegant solution provided by my pal
Jeremy White and the good
folks at Codeweavers )

The market is near, but not yet at, a tipping point.  To get to the next
phase, there needs to be pressure from a few desperate, determined, or demented
stakeholders to move key applications to Linux.  In short, someone will
have to pony-up money and/or manpower to port existing applications to Linux, or
recreate them from scratch.  Neither will happen quickly as the
cost/benefit to the vendors is risky and most enterprises would rather invest
that kind of money internally.

The alternative is for vendors to Open Source some of their development work. 
If they open client code programming to communities with a vested interest in
creating Linux desktop applications, then the vendor can get the work done
relatively cheaply and quickly.  The downside is they face extra support
costs on the back end for no new revenues in the short term.

The other alternative is to port clients to web browsers.  With
AJAX
offering elegant ways of creating clients that are portable across most (if not
all) platforms, many vendors have a "port once and for all" opportunity.

The key is for vendors to recognize that they must support a Linux desktop in
the long term because of Asia.  Asia is growing like an economic weed, and
they are starting nearly from scratch.  Many Asian enterprises have no high
switching costs away from Microsoft or Apple, and thus are adopting Linux
desktops rapidly.  This means whatever applications you sell into Asia must
have a good Linux desktop story.  If not, those markets may be closed to
them, and competitors who do support Linux desktops will prosper.

Asia is the Linux desktop tipping point … and its coming faster than many
people want to admit.