You should get paid for making a game, but you shouldn't make a game to get paid.
I think I'm going to print this out and tack it on my wall.
Non-linear progression through a gameworld is not something you're likely to see nurtured outside of text adventures any time soon. It balloons the cost of asset generation while simultaneously decreasing its point (because a healthy proportion of whatever you generate won't be seen) - that's a lose/lose situation to any publisher thinking of investing millions.
That assumes both a brute-force approach to asset creation/utilization as well as branching-tree style non-linearity. The latter has long been dismissed as a bad approach (although ignorant game designers keep trying to do it,) while the former is a problem that the big companies are finally waking up to (or at least Will Wright; everybody applauded Spore, but nobody else seems to get it really.)