[xapata]

> after you fire someone, they don't impact your numbers anymore

They would if your measurement is incidences per month. Otherwise you could fire someone every day and think you're doing a good job at office culture. I'd rather not have any harassment. Of course, you'd need to be careful not to discourage reporting of incidences. Measuring things is tricky.

> Find an offender, fire them

What's your technique of "finding" problems? Could it be improved? How would you know?

> no gray area

Of course there is. How do you know when someone accidentally insinuated sex versus intended a salacious proposal? Or accidentally touched someone's rear while passing by? Because it happened 6 times in 2 days? What about 2 times in 6 months?