The second generation of the CM engine allows for specification of plans with triggering actions selected at random (non-even distribution). I have found it to make the CM a more competent attacker, and for the first time the scenarios do exhibit re-playability. Where they use to be like mystery/spy novels (once you read the book; that's it). I would wait three years before replaying a CM first generation scenario and almost always as the attacker; because at my age it takes me three years to forget where ATGs are located and such.
But PG has always had a far more capable AI than the CM series. CM units individually have behaviors, but the strat AI has been weak and the TAC AI did not exhibit emergent properties which would threaten the survival of mankind any time soon.
Now, GTOS (Graviteam's Operation Star) is a real pleasure. Is the Strat AI or Tac AI that strong? No, I don't think so, but it is adequate. But where it shines is that it is a two level game where the operational level is generally unpredictable and thus yield totally random and unpredictable battles. I love it.
So for PG:
* Variable force/reinforcement matchups.
* Variable objectives.
OR
* Variable super size map objectives.
* Creates linked scenarios with exit condition sliced from the larger map.
* Forces carry across battles.
---
Something I love about GTOS. It is very possible to win battles and lose the operation. Since force preservation and balance is critical. The principle of economy of force is essential to reaching the finish line. Too many war games end up being these last man standing contests. They may be tactical or strategic, but they lack a strategic feel when they are played like that.