Re: Microsoft says Internet browser has bug

Robert S. Thau (rst@ai.mit.edu)
Sat, 17 Aug 1996 13:12:35 -0400


The thing I want to know, is, Microsoft hires every quality programmer in
the known world. How does this happen? What is the mentality in Redmond
that lets this shit seep out? Don't they check their products? What
actually goes on?

Well, I've never been inside Micro$haft, but a little experience with
industrial development gives me a few ideas about how this sort of
thing can happen, which are consistent with what little real info
escapes from Redmond. (Not to say that these things necessarily *do*
happen on every project, or that they happen only at Microsoft ---
anyone who believes otherwise is hereby sentenced to memorize five
chapters of the Dilbert Principle --- but that they're likely causes).
Stuff like:

*) Design by clusterfuck, with lots of people throwing in their
own neet ideas, and no one keeping track of the overall architecture.

*) A process for setting product cycle goals which puts the focus on
end-user visible glitz, necessarily shifting it away from
internal technical excellence.

*) A high-pressure environment which guarantees that anything which
doesn't have an explicit focus shone on it by management will get
short shrift.

*) A corporate culture which assigns little priority to bugs which
*most* users will rarely encounter or detect, going all the way
to the top (viz. the RISKS digest a ways back which quoted Bill
Gates, in an interview with a German computer hobbyists'
magazine, variously denying that Microsoft ships buggy products,
and sneering at the idea of spinning a new release solely to fix
bugs).

After the first of these, small technical details like this http-auth
glitch are likely to be lost; the rest does nothing to help them get
found once that happens.

(The above are, of course, just manifestations in the *particular*
environment of software development of the same factors that in
*general* justify the well-known approximation of the IQ of any
committee as the IQ of the dumbest member divided by the size of the
committee).

rst