However, this fails to take into account that the
government will instead
of spending $200 per machine for licensing of assorted pre-existing
commercial software, will instead have to spend
millions of dollars
hiring people to write the software that they need that does not exist
on
Linux. It also does not take into account the time it takes to
retrain the entire government workforce to use the new
operating system
and
software. Nor does it take into account the retraining of
the sysadmins and reworking of
network architectures.
Don't get me wrong, I use Linux on my machine. However, there is
an incredible cost in switching installed bases. Microsoft Word
has had years of market acceptance. It has had hundreds of thousands
(if not millions) of developer hours
put into its existing state.
It is far from perfect. Yet to bring another piece of software to
a similar state will likely take similar amounts of work.
Yes, Linux and other open source operating systems have an extremely
powerful developer group behind them.
Still, the nature of open
source does not lead itself well to accepting business requirements
from people. Nor has it had about 20 years of evolution of interface
and functionality. Open source leads to extremely good programs to
fill an arbitrary problem that takes the interest of the developers -
it does not fix problems that don't interest the developers.
The point I'm trying to make is that money will be spent one way
or another. It will either be spent on 'expensive' propriety
software licenses, or 'expensive' software developers, training,
and re-tooling of long ingrained processes. The government is a very
slow beast to turn around. And yet, I doubt that a few million
in software licenses will make much of a dent in the entire budget
that deals with numbers in the trillions (millions of millions).
I'd much rather have the government work efficiently than cheaply.
Good, Fast, Cheap: Pick any two (you can't have all three)
--RFC 1925 (The Twelve Networking Truths)
This can be applied to the government processes also.