Say what you will about Windows (and I will even join you), but Microsoft can put out some alarmingly competent docs. I say "can" and not "always do", because where they are lacking, I agree it can be frustrating. That said, in one such scenario the solution for me was to get in touch with a PM at MS and say:
"You designed [system #42]. I'm trying to accomplish [nifty goal] that is an obvious extension of what you suggest that [system #42] should be used for. You said some examples are included with the most recent Windows SDK. After downloading it, I found no such example and the problem seems a little intractable. What gives?"
The response contained an attachment of the overlooked/unincluded examples, and some clarifications about the system I was working with and even explained that some of what I was doing was explicitly NOT possible, for very good reasons (for the record, an earlier version of Windows Search).
I say this as a desktop Linux user, where at times my only recourse for "Why won't [application x] properly fullscreen?" is to go and do just what you suggest: an easter egg hunt on google. Furthermore, the end-all-be-all solution is the same as it is with our Windows example: get in touch with someone at the project and ask for clarification. The difference lies in that with one case, it's someone's job to provide clarification for the system they designed. In the other case, the possibilities are too varied to speculate on (e.g. unmaintained project, antisocial developer).
What you are talking about sounds more like windows developer than windows hacker. Still, almost all of my programming experience has been based on greping source trees in place of documentation. When I do get good documentation (it does happen) I don't know what to do with it and still look at the source.
What really sets open source apart for me in terms of hacking is that I can change the way any component operates, instead of having to jerryrig what would be a simple change using external APIs.
