I keep coming back to an idea for a competitive RTS that features randomly-generated maps for most matches, and where many other details such as opponent faction and your own and opponents' starting locations are not known. Obvious such a scheme would have problems with balance in cases where a generated map turned out more favorable to one player than another. I've been trying to think about ways to ameliorate this (beyond simply having a really good set of generation algorithms), and want to know if there is a lot of precedent for one such idea (and to hear thoughts).
So imagine that a map is generated and certain amount of detail is revealed to the players; a partial preview. They then have some options for refining this preview: a strategic scouting phase. They might send out scouts to locate gold mines or other resources, or get a better idea of the exact topographical details. Finally, they might even have the option to modify the map itself: "destroying" a resource node hey have reason to think their opponent will be closer to, or otherwise values especially highly.
The idea would be that none of these changes could be completely game-winning. They would be at most little statistical shoves to make a map more favorable to yourself in a way that you can control, or remove an enemy advantage if you spot one.
In a more blunt fashion, this could be accompanied by a "veto" ability so each player can force a map regeneration if they desire. This option would have limited uses.
feasibility of a "veto" power dependent on how fast maps generate
In a hypothetical scenario with two players equally skilled at the "strategic scouting" phase, overall change to inherent map imbalance may be zero.
Anyone know of games where the first part - the ability to gain more info on and modify a map before proper play begins - has been tried? Of course, other thoughts are welcome. Aside from strictly balance issues, does this sound like a possible entertaining addition for RTS players?