In the epic game at least, this really only applies when the human has access to fast (M>1), powerful units (compared to the AI) such as Knights, Cavs or Tanks (or UU-equivalents), which -- crucially -- have the ability both to retreat during combat if losing against M=1 (defensive) units, and also to run to friendly city to heal, either via friendly roads (which triple movement speed), and/or over hostile-but-flat terrain which doesn't restrict all movement to one tile per turn regardless of M-value. This allows units to be 'recycled', thus a smaller total army can still fight effectively. But in this Mod, all units are essentially 'slow', since a unit with M>1 pitted against another unit with M>1, will never retreat if losing -- and even if it wins its fight, it cannot escape to heal due to the high movement-cost of the terrains. So the 'smart' thing for a human to do (or at least, what I tried to do, which might not be the same thing!) would rather be to build a combined-arms stack(s) consisting of (fast) attackers, defenders, and bombardment units. Problem is, of course, that due to the Imp's Stealth-attacking abilities, the rebels' (my) stacks can't actually get anywhere, and still be useful on arrival... The Imps made no effective moves against Pisidia in my game to date (although that will likely change in the near future!). They did initially make some aggressive forays against Leucet, but eventually lost interest due to the stack of bombardment units and defenders I placed there. It was shortly afterwards that they started hammering Matallim (the city, misread the name earlier) instead... I was surprised to note that my (posted) save was/is actually only Turn 22. Matallim had fallen 3 or 4 turns before that, IIRC. On topic: If you don't want to nerf the Imps (which is fair enough, it's your mod after all), then another possibility might be to make an SP-version of the .biq which doesn't require (Strategic) resource-trading, by putting all the essential Strat-resources on both halves of the rebel-held area. That way, the human player doesn't have to worry (so much) about defending that trade-route in addition to their cities (because the AI doesn't recognise the importance of keeping that road open itself). The AI also doesn't send Workers into harm's way while at war, nor does it particularly care about breaking deals -- so it won't prioritise adequately defending its excess Strat-res from pillaging. I actually had to send a Royal Garrison + Worker into Metelim myself, to road + Fortress the nearest Traun-resource to Pisidia, just to ensure that they could supply it to me. Not that it did me much good, of course, because the Imps killed the first defensive stack I'd installed in the Hill-Fortress, and pillaged the road under it shortly afterwards. And although I did regain the Hill-tile, I lost a lot of units doing so, and couldn't hold it long enough to get a Worker up there: I lost it (along with a stack of 8 Ballistae) again after Matallim fell -- and only then did the Imps attack and pillage 'my' Traun! Strat-res are also generally more valuable to the AI than Luxes, so Astracet is at a trading disadvantage right from the off (just because we're Allies, doesn't mean the Metellim will give me mates' rates on their goods!). Putting, say, 2 different Lux in each territory might be a 'fairer' way of encouraging (Lux-)trading between the rebel sides, without crippling both if trading becomes impossible (which it probably will!). Even without resource-trading though, a solo-player would still be required to aid their AI-partner, since (presumably) the Locked Alliance means that the game will end in defeat if either of the rebel capitals is captured (and the associated king-unit killed)?