I think his point is that saying we're in a bubble should mean that the right thing to do now is divest, because you can't precisely pick the date that it will end. It might not be immediate -- you might want to ride the bubble a little -- but you should be basically assuming that at any point, the over-valued companies are going to pop, and look to getting out of them.
If the right move now is to continue to invest -- then that means that the valuations are not bubbly.
So, 10 years to a burst is way too far out to cause you to divest right now, if we are going to have 10 years of growth before then. Even 5 doesn't mean we're in a bubble now, so Steve saying 5-10 years is the same as saying, "we're not in a bubble" (according to what I think Ben means). I think Steve understands that too -- and that's why he changes the subject.
If you say 10 years, people start pulling out 9.5 years in. But, knowing that, smarter people will pull out 9 years in, hoping to beat out the people pulling out 9.5 years in. Then...well, you get it. Game theory. And that's how bubbles become unraveled.