There isn't any real objective way to measure performance though - there are things you can quantify (lines of code written, code reviews performed, etc) but those don't encompass the entire job, and once it becomes clear the metrics you use people will work to optimize their numbers at the expense of the other aspects of the job that are harder to quantify/measure. Who has time to mentor junior colleagues when you gotta bump your LoC count for the quarter?

I think we need to embrace the idea that whether you are given raises, promotions, continued employment, etc is largely a function of whether your manager likes you, likes having you on their team, and wants to reward you. And if they don't like you, it doesn't matter how good your numbers are now they will find a way to get rid of you sooner or later. It's better to just get it over with, give the employee a generous severance package, and part ways (or transfer them to another team that's interested in them).

You failed the moment you said objective. Any such system needs to acknowledge right off the bat that these assessments are subjective, subject to human error and re-enforcing bias etc. Otherwise you're just high on your own supply.

Indeed, the objective from your manager is to launch X feature (as designed by the design and product teams and in collaboration with three other teams) and achieve Y% usage within the quarter of Z.

However, if the company’s priorities change, significant on-call issues arise, the design team experiences a layoff, the front-end team undergoes an exodus, and the product team abandons the project, you will not meet your objective. Any one of these, or more could happen.

Consequently, you will be considered a low performer. You could have kept the ship afloat, kept the trains running on time, and deployed your part of the project on time and under budget, but still be a low performer on paper.

They aren't, in the platonic form.

But what I think most of us have experienced, in most companies, is that instead of driving compensation and promotion/firing the middle-management decisions around those things are made and then post-hoc justified via the performance system.

Which is another symptom of HR not having real power and instead simply being used to implement decisions made by other leadership.

We have no idea how to measure management performance at scale, at basically any level, in any kind of reliable way. It takes PhDs and rare datasets to even start to make any kind of serious data-backed statements about what works and what doesn’t in management, and how to hire, train, and promote effectively. It’s basically all a black box that’s so hard to measure that the most any company does is pretend they’re using objective and relevant data for it, when in reality they’re mostly looking at noise.

Keep that in mind when anyone talks about how vital performance evaluations are. Management can’t even figure out non-bullshit ways to evaluate their own work.

In reality, "objective" could be a fairly subjective target that depends on your relationship with your manager in a company with bad culture. Just spend 10 minutes browsing on Blind.

> it's crucial to have an objective (as objective as possible) system for measuring performance both to reward your high performers and to understand your low performers whether to provide support or to document reasons for firing them.

This isn't how it ever, ever plays out.

Also, most companies implementing this clearly are doing so out of some sort of compliance, whether it be SOC or otherwise. So yea, lots of times it actually is total performative bullshit. Congrats on finding the .000000000001% of cases where it isn't, if you are being serious here.

You’ve articulated, correctly, a need for performance reviews. We have a need to measure performance, but we don’t have reasonably objective ways to do it that simultaneously measure what we actually care about.

There’s a kind of tradeoff here, where on one end you have the most useful performance measurements, and on the other end you have the most objective performance measurements. When you’re doing things well, the best you can do is fall somewhere on that line. The entire line is bullshit, you just get to choose what kind of bullshit you have.

Large orgs tend to skew towards objectivity because it minimizes liability, and large orgs are full of people trying to avoid being liable for mistakes.

LLMs wrote this

The underlying goal behind performance reviews is noble and admirable.

But performance reviews are still bullshit. You could replace them with 'vote off the island' style votes every quarter instead and we wouldn't be any worse off.