Hacker News Reader

Comment by Zimahl | Hacker News Reader

I believe it's a 'miss' for the manager if they can't do the performance reviews the right way.

I worked with two folks that had previously worked at Intel. The same procedure was implemented there. There was a top 10%, a middle 80%, and a bottom 10%. This isn't a problem when working on a team of 10 to 20, but you certainly don't want to be on a team of 3 to 4. In that case, someone has to be the top and someone has to be the bottom. Once you are on the bottom once it's pretty much a 'start looking' situation.

These types of systems have their obvious drawbacks. You don't want to work with good people because the better they are the bigger chance that you'll end up on the lower end.

View on HN

Yeah, it's clear to me what the disincentive is. On the other hand, maybe call their bluff. If you're the manager willing to go to bat for your team and take a stand against insane performance measurement systems, there's a decent chance they don't want to lose you anyways.

My read is that with the situation Yahoo is in now, that's exactly the type of manager they want to let go. From their perspective, those are the managers who don't listen to the directions they're getting from executive management.

It's hard to turn around a company if you tell people to do something and nobody does it.

You would be surprised. Some big companies are so broken that these processes are like a steamroller. If you're in management at one of these places, do not assume that because you are valuable and in a profit center you will be able to buck the trend. Better to just start looking for other opportunities now.

Large companies that use forced ranking: you will reap the fruits of your broken process after a few years.

The typical solution (although I am not a manager) is to bucket up staff until you reach some predetermined size (i.e., collate staff upwards in the hierarchy until you reach something like 100 people). This still really only works if you have a large population doing similar work, and it just so happens that they're all divided into small teams. You probably need to still do level setting (e.g., this part of the org worked on a harder/easier problem than this other part of the org). Not an easy problem to solve.

Indeed and at this point the game turns into articulate managers defending their people and less true performance driven.

With that sort of distribution, there have to be at least 7 people on a given team to have a >50% of a single bottom 10% outlier (0.9^6 = 0.53). Bell curves only work well with large sample sizes.

Reply on news.ycombinator.com