Hacker News new | past | comments | ask | show | jobs | submit login

At Sun we experienced a new metrics fad every other year. Once they tried to use bug report clearance rate, so a friend said "I'm going to code myself a minivan" at a meeting. That metric didn't last.

You have to have a great deal of sympathy for the management of software-heavy companies. They've no idea how to manage software development. Nobody really knows how to. It sucks. Oh, well, I suppose there is a way: mind the bottom line, do not over-hire, and have good managers of the old sort: the type that know how to keep tabs on their direct reports without being overbearing. Business schools won't go for that.




Dilbert had this punchline in 1995: https://dilbert.com/strip/1995-11-13

It's possible one of Scott Adams's moles was at this meeting, since they seem to be omnipresent and omniscient.


Ahhh, I'd missed that. Thanks for setting me straight!


Goodhart's Law in action: "When a measure becomes a target, it ceases to be a good measure.”

https://en.m.wikipedia.org/wiki/Goodhart%27s_law


This sounds true and I've seen it in action. But I always wonder about the implications.

Must we keep our most important measures target free? That sounds insane, although I can see how it would improve the measurement.

So the questions is how are we supposed to set targets without our most important measures?


> "I'm going to code myself a minivan"

Could you explain this term? I thought i was a native English speaker but this has lost me.


"I'm going to get myself a large bonus by clearing many bugs".

The Dilbert strip has the additional suggestion that the programmer could introduce bugs in order to clear them, in an example of the Cobra Effect.


Or the AI vaccuum cleaner problem. If the target is to scrape maximum amount of dust then it would clean-dump-clean to maximize efficiency.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: