[michaelochurch]

I'm going to take you seriously even though you don't deserve it.

Any rip on a state with 50 million people in it is somewhat in jest. Anyway, there are a lot of great things about "California culture". When applied to technology, an open-minded and experimental "Let's try it" mentality is great. Necessary, even. When applied to management without enough attention paid to the fact that some of the people posing ideas have bad intentions, some awful ideas get into implementation and it damages companies. It hurts people. So more conservatism in selecting what to implement is in order, and discussing ideas that might be harmful (with thousands of people) until they've been explored is a bad idea.

The problem with Google is that it's got the conservative New York culture technically (I mean, even Scala isn't allowed) and the California culture with respect to hare-brained managerial ideas like downslotting-- the exact opposite of how things should be.

Put it this way: technological and managerial innovation are utterly different. If you do a tech demo and it's slightly rough around the edges, that's fine. You're awesome for having the courage to put yourself out there. If you're putting forward suggestions that are going to affect the way thousands of people work, the traditionally sloppy (for tech, I mean "sloppy" in a positive sense) tech demo is not how you should be communicating.

[apenwarr]

I think I agree that the existence of downslotting was a mistake. However, the problem it was attempting to solve (you got hired at one level, but performed at a lower level, because you managed to fool people in the interview) is a real one and there is no good solution for it. So I appreciate that they tried to find one, and realized their attempt didn't work, and tried something else.

If management was easy, we could all read a book about how to manage a company and then all do it optimally. Since that's not the case, experiments are necessary, and I admire the attitude that leaves them open to things that might not work. I disagree that it was "obvious" that the downslotting mechanism would not work - or rather, that it was obviously worse than any other alternative, because once you have the "not performing as expected" problem, all your options suck.

And BTW, I'm in New York :)

[michaelochurch]

However, the problem it was attempting to solve (you got hired at one level, but performed at a lower level, because you managed to fool people in the interview) is a real one and there is no good solution for it.

These levels are a convenient fiction. Performance is way too context-dependent to believe that there's some "platonic" level for each engineer. This (http://michaelochurch.wordpress.com/2012/01/26/the-trajector...) is the best model I could come up with for the software engineer trajectory, and even it has a 0.2 to 0.4 point (out of 3.0) swing for most individuals based on technology choices, interpersonal topologies, motivational flux, etc.

What you mean is "some people don't work out". Right. So there are people you "manage out" (that is, try to get them to find another job and fire them after ~6 months if they don't get the hint and you absolutely have to) and there are people you work with to bring them up to speed, or to figure out what's blocking them. Typical management stuff. What doesn't work is to keep people around but at a lower level than they were promised in the hiring process. That just creates a class of miserable, shafted people who hate their jobs and the company they work for.

By the way, a lot of the idiots on this subthread think I'm airing personal gripes. I joined after slotting. I'm just pointing out what kinds of ridiculous results come from an out-of-touch management culture. What drove me insane at Google was being at a company where the engineers were so good at their jobs and yet the people making important decisions were so epically bad. The disconnect was shocking. I was watching an awesome company self-destruct in front of me.

Also, on downslotting: careers are sensitive things and once you make one overt move so directly against an employee's interests, you've essentially lost that person. Loyalty is pretty much binary. Once you make a move like that on someone, you now have someone whose full-time interest is career repair, which usually involves getting the fuck out and lining up the next job. If this is what you want, then fine. (If someone really is a bad fit, for that person to begin full-time job searching is the best thing.) That category doesn't encompass most of the company. Downslotting only makes sense as a mechanism for managing people out, and (1) there are better ways of doing that, and (2) you shouldn't be managing out over 50% of new hires.

The real reason for slotting, I think, was to put a better job title in the offer letter than people were actually expected to get, since the upper title is what was used. This works only because of Google's brand: what keeps it from doing major damage is that downslotted people have the Google name on their resumes and can get the fuck out long before they become "problem employees".