I don't want any part of them.  They result in over-engineering, wasted documentation, never-ending discussion as everyone takes the opportunity to put their chef hat on, employees that get stuck going down a path while the world changes around them, and promises to customers you won't be able to keep. 

Planning is not bad and having a clear vision a necessity, but don't get caught up in the details of your roadmaps. Instead of a huge roadmap have a customer scenario end-state that you can use to comunicate what needs to be enabled, made simpler, or exposes an opportunity. 

Another way to think about it is that your goal should not be to pretend you are a writer on Alias that keeps trying to build up to a dramatic "end game" that takes five years and will likely disappoint.  Make sure everyone knows about your end game and try to keep it achievable in 2-3 years or less. 

The more you accomplish down a path quickly the more people will learn about and have your "end game" come into focus for them with a bunch of "aha" moments as the see the pieces fall into place.  If you can't communicate your desired end state and a roadmap for getting there in less then one slide... I don't want to be a part of it.  There's just so much low hanging fruit out there that your time will be better spent picking them off to get short term wins that start to make your desired end state more visible to folks.

What you need is smart people to identify the low hanging fruit, prioritize the fruit in order of how much closer each piece gets you to your desired end sate, and start picking them off one at a time. 

Another problem with the "5 year" thinking that's rampant here at Microsoft is when people say "Yeah, that would be cool, but what we should be thinking about is..." and the ... is usually filled in with the most overengineered solution you can think of that's designed to handle every edge case and adjacent problem you could imagine. 

We're hired and paid to constantly be thinking in line with big pictures that we frequently miss the little things that, when done en-mass, would have a much bigger effect on customer satisfaction than the overengineered solution would.. because you'd be 5 years late. 

Again, I don't have a problem with big solutions to big issues, but at Microsoft we often grab that big hammer too frequently when what you had was a pushpin of a scenario.

I guess, in the end, what I'd suggest is to question your choice of weapons used to solve problems.  Only break out hammers when necessary, don't believe that every problem needs a 5 year plan, and start accomplishing the little things that will have a huge customer impact... just make sure that you do have an "end game" in mind that your building towards.  And remember that an "end game" shouldn't take you 5 years.