[asdfasgasdgasdg]

I suspect this will be an unpopular opinion here, but I think that if you set the thresholds conservatively enough, a tool like this could be useful in identifying people who are not contributing. Where I work there are searchable tool invocation logs and there have been cases where someone is only compiling code one or twice a month, who not too long after I discovered this made an exit.

If you're enough sigmas below the median, it might be worth a closer look at least. Human review would be necessary of course. There are a plethora of ways to contribute. But it doesn't seem controversial that there must be some signal to be extracted from logs like those described in TFA.

[uncomputation]

The problem with any metric is that old quote about “when a metric becomes a measurement it’s useless.” How do you recognize someone who commits the same “amount” of code but which is consistently shitty and has to be reworked later from someone who commits a lesser “amount” but guides the project overall to a better place? Or someone who is spinning gears all the time: compiling code, spinning up VMs, in meetings all the time, and yet whose absence would be unnoticed?

Even a system of assigning a certain number of tasks. Sounds reasonable enough, right? But in my experience if you need X tasks per Y, the tasks will soon begin to conform to the metric. “Oh I need 8 gold doubloons this week, let me assign all of those to fixing the white space in this file.”

[asdfasgasdgasdg]

Perhaps. It depends on how easy it is to adapt to the metric, how risky, and how much upside there is. It's kinda like antibiotics. Use them too often and the pathogen adapts. It would be an approach that would have to be balanced with others.

For what it's worth I do think it is very silly to announce you are using metrics like this. Ideally you'd not want people to know what tipped you off that they were just cashing checks. You would fire them gradually over several weeks or months, and explain it some way that doesn't easily point back to the filter metrics. This approach of announcing it is like exposing pathogens to a sub-curative dose of medicine.

Anyway, I don't get the sense from the other commenters that the main concern is for the wellbeing of the company -- that they might be at risk of being gamed. I don't think an employee should have an expectation of privacy in at least several of the spaces used in the article as a signal, if any.

[MattGaiser]

I think the problem is this can only work once before you set off a ton of system gaming.

[oh_sigh]

Funny, where I work, someone saying "I haven't 'prod accessed' in 3 weeks" is a humble-brag about being so important that they are only writing design docs and doing high level reviews.

[wanderer2323]

Tell me you work at Google without telling me you work at Google.meme?

[oh_sigh]

Job Title: Inefficient protobuf copier