Re: [OT] Re: uk general election
On 4/21/05, geoffthur@ntlworld.com <geoffthur@ntlworld.com> wrote:
> Thanks for the reply Michael. I don't understand why it is so
> complicated myself, but apparently the latest CSA system contains 60
> million lines of code. One gets the impression that every time the uk
> Govt. has a new project to sort out it is sold to them as if it all has
> to start from scratch, using no off-the-shelf tools whatsoever.
This is in no way limited to the UK government, nor to governments in
general. The basic problem is "NIH Syndrome" -- "not invented here".
People don't trust software written externally, never mind the fact
that often the external software was written by considerably more
experienced programmers. I've seen some of the results of this in two
separate contexts.
As a graduate student in high-energy physics, I saw a mish-mash of
programs developed by individual experiments that all did essentially
the same thing. There were a few notable exceptions (CERNLIB being
the most prominent), but generally there was a massive amount of
reinventing the wheel.
In the US military, there's been a shift towards emphasizing what they
call "COTS" (commercial off-the-shelf) software and hardware. This is
at least an improvement. However, there are systems of very similar
functionality being developed by the research arms of all the
services, and they don't really cooperate in any meaningful way. The
result is multiple groups trying to develop the same thing, and then
later having to worry about getting the systems to communicate during
joint operations.
> The idea of treating the whole Govt. system provision problem in a more
> packaged and organic way - which is what I think you are suggesting -
> appeals to me hugely. How realistic it might be to do this, I don't
> know.
Never underestimate institutional inertia. That doesn't mean it's not
worth trying, though.
--
Michael A. Marsh
http://www.umiacs.umd.edu/~mmarsh
http://mamarsh.blogspot.com
Reply to: