Okay, let's get to cases.
1) Doctors aren't software engineers and aren't likely to become software
engineers anytime soon.
2) Open source software is the only realistic option for the future of
medical software.
3) Open source software "business models" in anything except operating
systems and things very close to operating systems are poorly worked out at
best.
4) Open source software developers are attracted by high-use pieces of
code. If something is in widespread heavy usage, open source software
developers are attracted to it and want to contribute without, necessarily,
monetary reward. The reward is the usage.
5) Only physicians can come up with the vocabularies/semantic nets that are
needed in medical software, because only they know what the hell the terms
mean, assuming that *anyone* has any clear idea what the terms mean [blood
pressure, for example]. *Namespaces* [RDF for example], in passing, are
a/the way out of the ambiguity inherent in each physician coming up with a
slightly different meaning for the same term. Gradually the namespaces
will merge.
6) The barriers to large projects in medical open source software are
considerable, in great part because of *usage*. Unless a particular piece
of medical open source is adopted by a multitude of physicians as soon as
it is completed, the issue of financing development becomes quite
difficult. A small, sometimes very small, group of developers must be
willing to endure a very long development path to achieve even a very small
market share. No one on this list is talking about adoption at a
significant market share level.
7) Yet...are there examples of free [not necessarily open source] software
that physicians have adopted en masse as soon as they were released? Yes,
ePocrates is the best example [in passing to Alex, are you going to come
out with an open source version of ePocrates-like functionality on the
amazing little Agenda device? the drug information is freely available, in
fact you *must* use the FDA info]
8) The easiest way to achieve quick, widespread adoption is functionality
that is on the web, or, like e-Pocrates, can be hot-synced from the
web. Note that if one goes to an ".exe" one must develop for Windows,
because there aren't any Linux platforms out there in the physician
community. "Aren't" in the sense of significant market share.
9) Adding points 1-8 together what are we left with? It must be
web-based. It must be adopted by a large percentage of physicians the
minute it comes out. It must be open source :-) What is it? Well, what
do you think? I think the answer lies in Point 5, myself. That's the key:
a high usage web-based piece of software that permits/forces physicians,
without their even knowing it, to create vocabularies/semantic nets, yet
has immediate gratification. What is it? Well, what do you think?
I will add in conclusion that even if what I am proposing never comes to
pass, it would work if it did come to pass.
John
