I read that article yesterday. It's an extremely well done article. Unfortunately, it doesn't actually say what the summary claims.
At the center of the cultural problems was a management system called âoestack ranking.â Every current and former Microsoft employee I interviewedâ"every oneâ"cited stack ranking as the most destructive process inside of Microsoft, something that drove out untold numbers of employees.
When the millionaire mint ran dry, the problems began:
And so, the bureaucratization of Microsoft began. Some executives traced the change to the ascension of Ballmer, but in truth Microsoftâ(TM)s era of fast cash was almost certainly the actual driving force.
Empowered by a dysfunctional incentive culture instigated by His Billness, though some defend it.
The Case for Stack Ranking of Employees
From the posts I read, the stack ranking at Microsoft is political and not based on valid accepted metrics that define performance. But Iâ(TM)m inclined to fault the measurement system more than stack ranking.
What a complete idiot. He presumes that such a metric must exist, and completely misses the boat on absolute rather than relative performance norms. As soon as the norms become relative, you're tying your sneakers to outrun your team mate. If that's not political, I don't know what is. There are people who might not be star performers by any specific metric, but who enhance the productivity of any team they join. Guess what other company adopted stack ranking? Enron.
I believe I once read an essay by Drucker where he said if the person who was worth hiring in the first place is underperforming, most likely that person's boss has failed to put that person into the right context.
And software is the worst of all industries to institute such metrics. Any crank an employee can turn at 1000 rpm is better off scripted. The surest route to efficiency is repetition (the athletic model from he cherry picks his favourite aspects). Human repetition is bad repetition, yet metrics never catch up to non-repetitive cultures.