←back to thread

371 points greggyb | 3 comments | | HN request time: 0.631s | source
Show context
FuriouslyAdrift ◴[] No.41983561[source]
Everyone forgetting about Lisa Brummel and "stack ranking"?

That nearly ruined Microsoft...

https://www.seattletimes.com/business/microsoft-ditches-syst...

replies(6): >>41983837 #>>41983909 #>>41983931 #>>41983969 #>>41993963 #>>42000415 #
bradlys ◴[] No.41983909[source]
What does Microsoft do now? Most every major tech company I’ve seen uses stack ranking - even if they don’t use that name. Hell, a lot of startups I’ve been at even use that. The founders and executives love it as far as I can tell - why else would they do it?
replies(1): >>41984296 #
1. andrewla ◴[] No.41984296[source]
At Microsoft in particular, stack ranking has always been used in the sense of trying to put together a rough ordering of employees at similar levels.

But "stack ranking" in scare quotes at Microsoft referred to the specific practice of the 20/70/10 rule -- the top 20% were the standouts, 70% was fine, and 10% was "this person needs to be eased out". This was applied for any org with more than a certain minimum number of people, and led to a very very toxic review process.

replies(1): >>41986118 #
2. bradlys ◴[] No.41986118[source]
This is pretty much what I see at almost every company I’ve been at in the last decade though… The review process is always toxic and has always lead to my peers in the industry being more likely to sabotage than help me since that’s the best way to look good in reviews. That with 80% of my peers being permanent H1B means they will do whatever it takes to stay employed.

I would be happy to know big tech companies that aren’t doing this but I don’t know any?

replies(1): >>41986811 #
3. andrewla ◴[] No.41986811[source]
I would argue that Microsoft's original practice (without the 20/70/10) is actually pretty good. Have managers make subjective evaluations, merge them together at higher level meetings, and then work out compensation from there.

There's a big cottage industry of trying to back everything up with data, to provide actionable feedback, etc., and these end up being giant time-wasting cover-your-ass exercises, which always end in an uncomfortable non-working system for everyone -- "I did the thing you asked but my review is the same as last year, why aren't I getting promoted". Mentorship and growth has to be more than just "here are your goals". Peer reviews can be okay, but only if you force people to make judgments -- "rank these three people against each other" rather than "give these people a rating 1-10 in each of these five areas".

The subjective evaluation process doesn't work unless you trust your managers, though. And that invariable means that it doesn't scale.