Hmm, kinda going in a bunch of directions here, but the "stalemate" thing is worth discussion:
1) The "stalemate" outcome is an intentional part of the original intent of the game design, and in fact used to be considerably more pronounced in older versions: one of the ways to lose in AIW is to be unable to make progress.
2) That said, I don't think it's much fun either. I don't mind can't-win scenarios as long as there's some satisfying way to measure how "well" I lost; like a pre/early WW2 scenario as Checkoslovakia or Poland: presumably if it were modeled correctly I couldn't come anywhere near near stopping Germany, but if I mixed the right concoction of audacity and natural-20s I might have the satisfaction of the dying rabbit seeing the look on the tiger's face as it realizes it no longer has any intact hamstrings and there are other predators out tonight. But just being ground into the ground by inexorably growing AI forces doesn't have that feel.
said, radically changing the game to remove the stalemate possibility is likely to have a net-negative impact on fun for the audience as a whole. It's easy to not realize the importance of a certain part of the game being a certain way until things start flying around due to a central change.
But I do recall some ideas I tossed around a couple years ago about a "Showdown" AI plot where basically some trigger would cause the AI to just go all-out offensive and if you survived you could win without much further difficulty. The idea was well received but we just never got around to doing it, and I wasn't ever all that clear on how to best do it anyway. One general problem is that it's basically making the AI do something really stupid: abandon its well-entrenched position. Stalemates win a lot of games for the AI, why would it stop? Because it's not the DM
So let me throw out an idea; this is just off-the-cuff and might be an impressively bad idea:
a) Add a new kind of hacking that takes place on an AI Homeworld (specifically, on the same planet as a living AI Home Command Station).
b) Doing this hack takes 2 minutes, during which the AI spawns some moderate retaliation but mostly just as a "does the human have enough forces to even consider this" check.
c) Once the hack is complete, the virus or whatever makes the AI "do something really stupid"
- All reinforcements stop for 30 minutes (I'm thinking for both AIs, otherwise the effect is kind of diluted).
- 30 minutes worth of AI reinforcements get spawned all at once as threat, from both homeworlds and potentially a few different places (maybe even some spawning like a counterattack-guardpost does). This will probably involve some exo-ish composition rules and/or wild-rolls to condense the available strength into not-too-many-individual-units to avoid this being a showdown between the AI and your CPU/RAM.
- During the 30 minute "there aren't
any reinforcements" phase, the "wait at wormholes" behavior is also suppressed, so it won't just pile up into a bigger wall (and will "flush" existing threat-walls). If that proves troublesome to implement (I don't think it will, but cannot be sure), just make the superspawns all zombies.
And in theory that would take care of it: if you die, well, the game's at least come to a conclusion. If you survive you have a golden window to smack the AI around. Of course, the response is probably "with what?!", which may indicate that the "30 minutes" needs to be longer.
Thoughts? I realize this doesn't address the overall mid/mid-late stalemate potential before homeworld assault becomes possible, but one step at a time