For example, we have a lot of legacy Delphi code. Delphi 7 has a bug in which it gives incorrect compiler warnings for some of it. I don't care one bit about the history, details and nuances of the bug. The 2 hours I don't spend examining that are 2 hours I can spend learning something useful instead.
It's not yes or no. It's not everything or nothing.
* There are cases where the fastest thing to do is ask your colleagues.
* There are cases where the fastest thing to do is ask your friends (if you have any).
* There are cases where the fastest thing to do is ask Google.
* There are cases where the fastest thing to do is ask StackOverflow.
But there are times when it's best (and not necessarily fastest!) to work on it for a while to see if you can either solve it, or at least get your mind around it enough to understand the solution when it comes.
The very best programmers/engineers/employees/founders/people know when it's which case from the above.
This is true regardless of which method he'd use to figure out a solution.
I do agree it's not binary -- I'm learning Objective C right now, and spent two hours last night tracking down a crash that was ultimately linked to not understanding retain and release semantics. That is worthwhile, and a situation where just asking someone for a fix wouldn't work.
But I think that's a learning issue. If you already know the majority of what's going on, and the rest is just finding a key, then I think searching/asking is perfectly valid.
And really, wouldn't you argue that a programmer worth being called such would always attempt to be involved with finding the answer?
It's not binary.