Yeah, I agree. An asker is often unsure of what they need to ask and doesn't realize that what they haven't said is crucial to dealing with their real problem. An apparently simple regression question changes completely once it's revealed that the response is binary, or cumulated values, or that there's error in the predictors. Or the question will explicitly ask about testing (because that's what they know about) but the actual problem is one of prediction.

Often the signs that point to these things are only the subtlest hint in the original question.

"What are you trying to do?" is pretty much essential for a lot of the questions.

Especially when a lot of detail is abstracted away.

Answering when I am tired (which is sometimes the only time I have) can lead me to miss even obvious things in the question, though. I've managed to skip whole lines of text a few times.

This sounds like the xy problem, which Meta.SO has an entire tag on.

Good exchange. It reminds me of a lesson this site continually teaches me: an important component of creativity in statistics is the ability to interpret a question in multiple distinct ways.

Out of multiple interpretations come counter-questions and out of the answers to those come clarifications of the OP's real need.

If you can conceive of only one interpretation, then you have no motivation to probe for those clarifications.