Vision >= Solution >= Problem
I'll admit it, I'm not 100% agile since I tend to like solutions. I would prefer a problem to solve, but if a problem is intangible I often find a solution is a great way to explore the potential and help express the underlying problem.
It is worth noting that innovation and invention are solutions and not problems. When you apply for a patent, you don't patent the problem, you patent the solution, why? If there are multiple solutions to any given problem then wouldn't we want to patent the problem?
I also find that quite often the solutions I am exploring are the result of a problem that I have assumed was well known and in actuality is not; it is in these moments that using the solution to derive the lowest common denominator of a problem stated as a user story can ensure that everyone is speaking the same language.
But given a solution, do we always need to know the problem? My pragmatism (and a few lean principles, i.e. waste) would state that if the solution brings value then run with it, if not ditch it. If on the other hand the solution seems too constrained or lacking in features/functionality, then why not exploit the solution as the basis for a vision? Indeed many visions are just that, solutions to problems, expressed in a language alien to most developers.
So given a vision expressed in an abstract form vs a vision that is pinned against an actual solution which is easier to understand and less likely to be misinterpreted? I know which I'd prefer...
Do solutions have value, obviously; must we always mine problems, hmmm...