I'll chime in on this thread finally.

"Let’s write a response that we will still be proud of, decades hence."

One of the oldest responses on the Internet is "RTFM", no matter how you slice it. It's also the most realistic response to anyone who is having an issue when it has been proven that the docs clearly portray how something should be done.

Now, I am in agreement with others that if the docs are followed but are either inaccurate, ambiguous or otherwise confusing, ask a question about it, and, I'm in agreement with sometimes you just need a quick answer so ask it (Rubber Duck Debugging), but I do not agree with asking questions when one has the time to read the docs thoroughly but don't. That don't go over too well.

As a (what I like to think as a somewhat) prolific software author amongst a few languages who is vehement on keeping documentation as accurate, legible and understandable as possible, I want to know when my docs aren't portraying the usage of the API or binary I've written so that it is completely understood. It takes me as long, or even longer to write my usage documentation than it does to write the code, and oftentimes longer than writing the code and tests combined.

When one has time to take a few moments to read the docs, then say they were scanned (without actually testing what the doc says) is actually an insult to any software writer who takes the time to write really decent documentation.

RTFM