Comment by JohnMakin
This shows very visibly in the devops/platform engineer/whatever-the-hell-we're calling-it-these-days world.
Often you will get a request, sometimes (or quite often) you have no idea what is driving it, like for example "reduce rate limiting for xyz service." Lots of ops guys will just blindly do this ticket shuffle, even very senior ones - maybe for their own sanity, maybe out of preservation - but the best ones I've ever worked with, will often question "Wait, why do you need that?" Then you find out there is some other really trivial solution to fix that underlying problem that doesn't involve as drastic of a change, or maybe even none at all. Especially if it doesn't involve code change on their side, you're not going to face much headwind in pushing back.
The reason this is important is that, no disrespect to developers, they often live in a world where they treat infrastructure as a blackbox (as I believe they should). The problem sometimes is they want to also control the behavior of that blackbox. So while the request may seem to solve the problem, often there is a much easier/simpler/safer/scalable way to fix whatever underlying problem got tossed over the fence to you.
The senior guys I've respected the most always will ask the "wait, why?"
I realized I was senior when jr. people would ask me things like "how do I make awk do this this?" and I responded with "what problem are you trying to solve?"