[robryan]

I get where the first point is coming from in that the best developer can be many times more productive than an average one but that doesn't mean the average one can't make a contribution. Many great programs have come from people that have admitted that they don't find themselves to be the best programmers but picked a goal and worked hard towards it.

[slantyyz]

Reading his points, I wonder how much experience he has and what the scope of his experience is.

My experience is that if you've got multidisciplinary teams populated with reasonable people who have a good grasp on real world constraints (usually a delicate balance of compromises between scope, quality, time and money), all of his four points break down.

On the measurement point: While I'm sure some of the more clueless companies might use some arcane and pointless measurement systems, I haven't seen any half-decent dev shop use any of the measurement practices he talks of.

[hello_moto]

That's because half-decent dev shop thought that they're full of good developers and their best developers managed to convince the management that measurement and metrics suck, just because software development is different!

Really, we need some kind of measurements or metrics. Otherwise, how else do we know we're progressing or improving or at least not sucking more.

I agree that metrics is hard to do and easy to game. That's why don't put a specific number or targets. And also don't tight metrics with employee performance. Put it on team's performance (still not perfect, but I'm sure it's better).

One example of metric that probably might have made sense is that if your shop is doing Agile (code review, code coverage, unit-test, # of bugs etc.) is to make sure that these numbers don't go down in every iteration. Make sure these numbers either stay the same or go up (in the case of improvements or new development). If the numbers go down for 2 consecutive iteration, someone or some team doesn't care about quality. Track improvements, not specific number.

Side note: I would suggest people to read Standish Group Chaos Report to understand how bad our industry was and how far have we improved over the years.

[slantyyz]

"metrics are are hard to do and easy to game"

That is the painful truth with employee measurement in general (hm, perhaps another knock against software dev being different). I would still say that the junk "objective" metrics provided in the original article aren't generally used in any competent dev shop.