Hacker News new | past | comments | ask | show | jobs | submit login

Joel is absolutely right.

If you ask people to enter their time, they will lie. Sometimes not on purpose, but they will lie nonetheless. And, they'll eat up 10-30 minutes trying to remember what they did and enter it in in a human-readable way. And they'll piss away some time/focus shifting from productive stuff to time-entry stuff and back throughout the day.

Of course, for time/materials folks-- lying is GOOD. It almost always means that you bill for more hours than you spend on behalf of the client. If you bill for an hour of time, you almost certainly aren't subtracting the 7 minutes you spend emailing your girlfriend and the three minutes you spent twittering about the bowl of oatmeal you had.

But I truly think that one of the most important things a person or business can do is understand how people spend time (and how that changes over time).

(note: I run RescueTime, which passively records how people spend time/attention. I also ran a time/materials consulting biz for 8ish years and constantly was nagging geeks to enter their billable time. Yuck!)




I've seen just one performance measurement system that works really. It must be done in a shop that does Test First development and maintains a comprehensive unit test suite. The shop must also use "story points" or something similar. To measure a programmer's performance:

    1) Review a random sample of their unit tests
    2) Count the number of passing unit tests that they write
    3) Count the number of story points they complete
It's harder to game this system. If you write lots of trivial, worthless tests, then (1) suffers and you don't complete more of (3). Also, story points originate with the user, so it's harder to game those.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: