Aside from the significant asymmetry of the distribution (of course) there's the spikes where people push themselves to hit a mark (i.e. 3:59 vs 4:00).
Programming is full of weird incentives that will distort themselves in any measure -- e.g. if A gets his/her work done in 0:20 and B in 7:30 and A then goofs off or helps other coders, the measured productivity difference might not be huge. Since most people who code are having their behavior distorted by measurement that's a big issue.
From experience -- the incentives against outperforming expectations are huge -- demonstrate you're 3x as productive as the guy next to you and from now on you have 3x the workload for 1.1x the pay. In practice what happens inisthe better coders pad their estimates and get trusted with tough problems rather than time-consuming problems.
http://m.runnersworld.com/sports-psychology/round-number-tim...
Aside from the significant asymmetry of the distribution (of course) there's the spikes where people push themselves to hit a mark (i.e. 3:59 vs 4:00).
Programming is full of weird incentives that will distort themselves in any measure -- e.g. if A gets his/her work done in 0:20 and B in 7:30 and A then goofs off or helps other coders, the measured productivity difference might not be huge. Since most people who code are having their behavior distorted by measurement that's a big issue.
From experience -- the incentives against outperforming expectations are huge -- demonstrate you're 3x as productive as the guy next to you and from now on you have 3x the workload for 1.1x the pay. In practice what happens inisthe better coders pad their estimates and get trusted with tough problems rather than time-consuming problems.