[rtpg]

I don't mean this to be cynical, but I do think that it's worth acknowledging that describing the problem is also, in itself, a tool to guide people towards a solution they want. After all, people often disagree about what "the problem" even is!

Fortunately not every problem is like this. But if you look at, say, discussions around Python's "packaging problem" (and find people in fact describing like 6 different problems in very different ways), you can see this play out pretty nastily.

[skybrian]

At a toy scale, using ChatGPT's Code Interpreter to do some programming for fun can be an exercise in getting what you want from an inconsistent worker by changing the problem definition (prompt engineering).

This is sort of like:

* writing an exam question so the person taking the exam is likely to get the answer you want

* guiding someone in a code interview that isn't going so well, without giving away the answer

* being in the back seat while pair programming, except you're not allowed to take a turn at the keyboard

[pgwhalen]

I don't think it's cynical, I think it's the point. Describing the problem is not easy, and to your point, is sometimes controversial.

One advantage of focusing on describing the problem is that it naturally lets you have an impact on what you believe to be the important parts of the solution.

[rtpg]

I just want to acknowledge that describing the problem is part of picking the solution, and it's not really _that much_ of a "I'm making the most neutral action and letting other people actually choose the solution".

Honestly the "real" hands off thing is letting somebody else also describe the problem and then probing it. But that might lead to a bit too much of an existential crisis for some people. And hey, if something works it works

[pgwhalen]

For sure, it’s only partly hands off. But he is an engineer after all, he should be doing something outside of just managing.