In my experience it's by talking to lots of people, focusing on those who you suspect "should" have solved it and asking them in great detail about the problem itself. Additionally you'll find that in many cases those with the problem will be happy to tell you either jury-rigged attempts from the past or point you to those who have tried. Interviewing many people is key.
Thank you for answering. As I suspected, a good network is a real deal. Making a good network is too difficult outside a tech hub as SV. Fortunately, we have HN and other online communities.
