Re: Nethack! (was): AdaGames

["Marin David Condic" <dont.bother.mcondic.auntie.spam@[acm.org>]

Its hard to collect metrics without having at least *some* of the developers
feeling that they will be rewarded or punished on the basis of the data you
collect. Hence, there's always *some* incentive to cheat. However, at least
in the studies I was involved in, there was a lot of grumbling about
metrics, but I never saw any obvious attempts to inflate or otherwise doctor
the measurements of code size.

Whatever you use: SLOCs, Semicolons, Halstead Bits, Source Bytes, Object
Bytes, anything. They all have weaknesses & can be artificially controlled.
But what choices do we have? We need some means of at least roughly
approximating code size or we might as well give up on trying to measure &
understand what we do. (I've been told by just about all of the developers I
ever collected metrics from that this would be just fine with them. However
after years of doing it, we were definitely able to see improvements in our
processes.)

MDC
--
Marin David Condic
Senior Software Engineer
Pace Micro Technology Americas    www.pacemicro.com
Enabling the digital revolution
e-Mail:    marin.condic@pacemicro.com


"Wes Groleau" <wesgroleau@despammed.com> wrote in message
news:3CAC7AC6.A6DBDC32@despammed.com...

>
> This is true as long as the programmers were not
> thinking that there would be any reward or punishment
> for a higher or lower count.  Take a well-written
> module of a reasonable size in any language, and there
> are usually ways to greatly change the metric without
> changing behavior.
>