Re: Design Methodologies w/Ada95

[mfb@mbunix.mitre.org (Michael F Brenner)]

   > They expect you to do unpaid overtime.

One of the biggest reasons that ordinary metrics fail to predict the
true cost of projects is that metrics for backlogs (such as overtime
that will never be paid, service requests that are on hold for ever,
documentation that will never be written, and testing that is too
expensive to do) are not created. 

Once you start accounting for the backlogs, you will find they
exceed anything that is correlated to line-of-code, complexity,
or any other volume metric. 

This is one of the primary justifications for doing true 
rate-of-return studies on the backlog of each project, and 
for keeping backlog and clustering metrics instead of volume metrics.

The converse of this is that if you are subject to an expectation of
overtime work that will never be paid, then you can publish
these metrics yourself on your door and on your status reports.

The first law of metrics is that as soon as they are posted on the
door the people who see them take whatever steps are needed to
improve them. Such as burning the poster at the stake along with
the door she posted the metrics on, improving management techniques, 
cancelling the project, actually making these metrics a part 
of the project management process so that the TRUE cost of the 
project gets measured, and, for future projets, making this true
cost the basis of the estimates and bids.

Seeking Truth,
Mike Brenner