I look forward to a post that actually suggests some value, either empirical or philosophical, inherent in the use of software metrics. This one isn't it.
No, software metrics are not evil. Many things are evil. Software metrics are more on the lines of "pathetic." They are definitely useless -- although software planning metrics on a historical and statistical basis, a la Spolsky, may well be valuable.
We can only judge by results, because the IT industry presently has no real concept of blind tests. I would refer your readers to www.hacknot.info for an in-depth discussion on this issue. One of the points that "Ed" makes on that site, and I think it's important to the way we approach these nebulous measurements, is that we all (and I've become convinced that this applies to me, too) need to apply critical thinking in a logical way. Unfortunately, when you say:
"Better yet, the programmers would have great evidence to have the moronic manager fired."
... this is actually meaningless. First of all, you're proposing the lemma that there is, indeed, some sort of meaningful software metric, or collection thereof. (This being a lemma, it's not possible without further proof to say whether you are wrong in proposing it or the software manager is wrong in misusing it.)
Secondly, you define this as "great evidence." Even if you went forward from the lemma to the proof (QED), you'd still need quantifiable evidence based against, presumably, some sort of algorithmic expression of that proof. Which would probably fall within the domain of statistics. And would probably be questionable on the basis of methodology, even given a universal acceptance of the proof.
Thirdly, you assume that evidence of 100% failure to do one's job, as defined in the tedious legal crap that we all sign in order to do our job, is sufficient cause to be fired.
It's not.
There's no such thing as a silver bullet.
But if there was, I wouldn't reserve it for vampires -- because I <i>know</i> they don't exist. And I wouldn't reserve it for bosses, because I <i>know</i> my only control over them is to leave.
I'd reserve it for people who argue that software metrics have any conceivable value at all, without significant comparative work.
Flat View: This topic has 50 replies
on 51 pages
[
«
|
242526272829303132
|
»
]
Previous Topic
