> If the metrics don't properly measure whether the job is being done then change the metrics.

Nobody has ever come up with a good set of objective metrics to judge software engineer performance. So the best we have is still the subjective opinions of your managers and peers.

Like in the cases of US courts defining obscenity or fair use, there isn't necessarily a set of metrics which can be used to perfectly taxonomize something.

Imagine I sent a manuscript to a publishing house and they rejected it for being bad. I wouldn't expect they got to that conclusion by comparing it to a set of metrics, I would assume they have people in authority whose judgement is the decider on whether something is "good" or "bad".

The original comment was regarding employees currently being judged via metrics bringing up whether certain jobs can or cannot be judged using metrics is pointless.

Your analogy only works when applied to the hiring stage, as that is when the publishing house decides to work with you. If the publisher accepted your manuscript, assigned you an editor, gave you a target publish date, and gave you advance and then suddenly booted you and said “your work isn’t good” you’d have some questions, and rightly so.

This sort of thing happens all the time? Many manuscripts and screenplays are stillborn. Movies make it halfway through production before the plug is pulled. Software projects fail left and right, with millions of dollars spent (sometimes billions!)

Human endeavors sometimes fail to live up to expectations.

They meet these metrics while they are under formal process just before termination. I used to work with a couple people clearly working multiple jobs who switched focus when they were PIPed.

If they are refocused on their job and now meeting metrics why terminate them? People can become unfocused for a variety of reasons beyond working other jobs. Life happens. If they don't remain focused and again don't meet metrics they have already been given an opportunity and should then be terminated.

> If the metrics don't properly measure whether the job is being done then change the metrics.

For workers who work with their heads, "metrics" is a fantasy. How do you measure a better writer?

Well the comment I was responding to specifically called out employees meeting metrics and still not being considered good employees, so your point is a little moot to my comment but I will reapond anyways.

How do you measure a better writer? It depends on what the purpose of the writing is. If it is an author directly selling books then you measure by books sold. If it is an online publication you can conduct surveys to determine the impact of a particular writer on subscription or view rates. If it is a techincal writer doing product documentation you can measure based upon meeting schedule, number of defects and by conducting customer surveys.

It is possible to both have some metrics and not have them be the only way you determine if an employee is doing a good job. Because some things can't be measured, and some can.

There are no objective criteria as to what is "good" writing vs "bad" writing.

> If it is an author directly selling books then you measure by books sold.

This is a fairly lousy metric. It depends enormously on the marketing campaign and the ability of salesmen to sell it.

For example, I read an article about the author of the "Slow Horses" book. It languished for years selling at a rate that was indistinguishable from zero. Then some journalist read it, wrote a glowing review of it, and it took off. Now it's a best seller, with sequels, and a miniseries.

Good writing is writing that allows the publishing house to achieve their end goal and bad writing is that which doesn't. The end goal is the same as for other businesses to make money. If you don't sale books you are a bad writer for their purposes.

What metrics do you propose that aren't susceptible to Goodhart's law?