Subject: Re: Open letter to those who believe in a right to free software
From: Jean Camp <>
Date: Thu, 4 Nov 1999 23:19:27 -0500

On the cost of death and open software.

There is certainly a non-zero proability that bad software will cause
death, as it has done so in a few cases.

Software controls an increasingly large sphere of safety-critical systems.

There is some probability - increasing over time -- that open source
software would be more reliable. Perhaps one could even argue that the
difficulty of using open software decreases the liklihood of individuals
using software they do not understand in safety-critical applications.
(Like mathematicians being dumbed down by computers and statistical
packages enabling foolish people to claim important findings).  But this
later argument is not critical.

Including this variable could show how sensitive the model is to the cost
of life. I suspect not very initially but increasingly so over time
assuming software controls increasingly critical systems.

Thus the value of a life is not entirely tangential to the open source

Further if code is speech, then there are many values which cannot be
easily quantified and therefore a CBA will not be useful in terms of its
direct output. What may be very interesting in such a  model are its
relative sensitivities to the value of different outputs, in particular
claims of reliability. This would provide a nice framing for the technical
Linux v Windows debate. I suspect you have these results for your model, if
so please share them.

There is a nice model of switching cost by Joel West which may prove useful
since this is no doubt a large element of the cost. (meaing once you are
stuck with Windows crawling out of the muck of that pile of vile code has

It would not be hard to put in a dummy variable, set usually to zero, that
trivially shows that if, in fact, code is speech then free software is a
simple values question. This would not be particularly illuminating but it
would certainly be honest and resolve the debate about the model at some
level. Then you could argue if its one or zero, or how close it has to be
to true to be important.

People at Harvard making bad analyses! Oh heresy! Off to the stake! Dare
not suggest this to the alumni of the business school! Heathen!