Thursday, January 17, 2008

I've lately been reading a bit about politics in France, a weird, topsy-turvy land where right-wing President Sarkozy is proposing new taxes to pay for commercial-free state TV, and inviting the heads of powerful labor unions to join him at the old Versailles hunting lodge that he's made his personal Camp David, while high officials of his own party complain they can't beg an invite. And where the left is maybe, just maybe a little too left-wing for my taste.

But there's one thing that Sarko's doing that strikes me as really questionable. Nothing to do with his personal style. I'm long past caring about that in politicians. (Though, by the by, it's a bit unfair for Ségolène Royal to complain that he's "an exhibitionist who lives like a billionaire". Unfair to billionaires, that is --- the problem with Sarko is that he lives like Donald Trump). Nor even the odd prattling on about the religious roots of civilization that has him frantically trying to explain that he's not trying to go back on separation of church and state.

No, what has me worried is that he's hired some American management consultants to evaluate his ministers, by establishing numeric targets for them to meet. For everything. Not just, say, the border guards, who have a quota of sans-papiers to turn away. But the culture minister, who's been tasked to get more people walking through free museums.

Well, there's an old story in computer circles about this sort of thing --- a story that never goes out of style, because each new generation of green, know-it-all managers keeps making the same damn mistake. They establish solid numeric targets for their programmers, in lines of code written, and in defects fixed. Thus rewarding the behavior that they want to encourage. What could possibly go wrong?

What goes wrong first is that programmers can maximize the first metric by quickly writing a lot of sloppy, buggy code. But, of course, that code is full of defects and barely works at all. Is this a problem? Not at all! They then get to waste time and earn money maximizing the second metric by fixing all the bugs that, with a little more care, they would never have introduced in the first place.

Government, of course, is a more complex business than the common run of programming. So, experience in the one field isn't necessarily a great guide to the other. But if a lot of French museums are suddenly doing shows of Tintin originals and the art of the soccer stadium, you'll know why.


Anonymous HP said...

Well, there's the project I worked on where someone got the bright idea of offering cash incentives to QA interns based on the number of problem reports they filed. It wasn't long before the list of PR resolutions looked like this: 1) Non-reproducible. 2) Rejected. 3) Rejected. 4) Works as designed. 5) Non-reproducible. 6) Oh, come on. 7) Rejected! 8) WTF?

We got plenty of bugs fixed, alright. Work? Not so much.

I'm sure there's a political analogy in here somewhere. Something about the surveillance state or something.

8:03 AM  
Anonymous Anonymous said...

NPR had an interview with a black song writer. He told of Gulf and Western who bought out the music company he worked for. Same story. The new management tried to get more songs out of the writers! They all said "No Can Do it don't work that way" and they all walked out.

10:21 PM  

Post a Comment

Subscribe to Post Comments [Atom]

<< Home