Potential AI strategy for larger maps

[Leroy30]

I've been thinking about the mechanics of processing a large number of units on a large geographical map (as in absolutely ginormous).

Would like to get some oppinions on the feasibility of my idea.

The idea is that the further away you get from the camera view and the longer the particular units have not been in view or issued commands then the more their fate belongs to probability ad opposed to actual calculations.

Example. We have 10 spearman walking up to close proximity of 5 enemy spearman. Unless we actually move the camera to where they are then we simply use probability, derrived from variables such as proximity, time duration, unit strength, unit speed etc, to calculate the probability of them finding and fighting each other and the outcome of the battle.

This way units not being focused on could theoretically be less processor intensive because we would not need to calculate every second.

Also i think units mining resources could be given a similar strategy.

Anyone think im on the right track? I might have to take a look at the source later and see if i can work out how it currently works.