3 ms·
I always thought having a pool of 20 metrics and randomly picking 4 every quarter/year for bonuses/promotions/whatever might get people to stop optimizing for s
by ByteJockey 5y ago
I always thought having a pool of 20 metrics and randomly picking 4 every quarter/year for bonuses/promotions/whatever might get people to stop optimizing for single thing.
- bongoman37 5y agoThis is fine until someone sues because they had similar or better metrics than someone else but the other person got in in one year and they were refused. A lot of the reliance on metrics is CYA by administrators.
- MaxBarraclough 5y agoA more deterministic alternative: a weighted sum of the pool of metrics, where the weight can be gradually adjusted over time.
- Beldin 5y agoIf you have 20 independent metrics, this could work. Typically though, the metrics aren't independent as they all try to measure "doing your job well". In that case, if you gamed the "wrong" metrics, you still come out ahead. Not as much as for gaming the right metrics, but still.