I love building product and solution roadmaps almost as much as I love talking to customers. The reason is simple - once you get executive buy-in that the path you've articulated is one that will deliver on your revenue and profitability goals, you've got the rough equivalent of a shield against the depredations of enthusiasm and entrepreneurialism that come your way on a daily basis.
Sing with me here - you *do* get new ideas coming at you every day, don't you. You might even be the source of some of them. Think of each of these ideas as a threat to you, a disruptive force that might bring good, but if left unmanaged, will certainly deliver a world of hurt.
Without a roadmap, you have no defense other than "I thought we agreed to such and such" and "the development team is already working on so and so". Imagine strapping some cardboard to your arm before going up against an opponent with a Spiky Atomic War Club. Not too effective.
The goal is to have everyone feel like he/she "owns" the roadmap - that the path it describes is the one that will lead the team (and the customer) to the promised land. Once this happens, the need for the roadmap doesn't diminish, but the protective effect it has is no longer quite as urgently necessary.
This is not to say that roadmaps don't change - they do. But without a benchmark to measure changes against, you're at the mercy of hope and luck, neither of which are valid management strategies.