Sorry for not replying sooner. Had a lot of stuff at work, so I've had little time to play with FFH.
In my experience the AI suffers when dealing with Pyre Zombies (so do I, honestly), but it can deal with them. I've never seen what you mention happen in ExtraModMod, even when we are the ones in control of the Pyre Zombies. It's relatively simple to conquer the AI with them, but it also is quite simple with nearly any valid strategy. At which difficulty do you play?
I would also like to mention that Averax is Imperialistic in EMM, and the pyre zombie cap is 90%, not 95%. The scenario you are mentioning seems to fit with Erebus in the Balance; is it possible that your problematic games against the Sheaim have been played with that mod? More Naval AI has added a lot of incremental improvements to AI since EitB separated from it and in these years the AI has become better at handling wars. Besides the general improvements, I believe that some of the AI changes were specific improvements to this kind of units (via the bExplodeInCombat XML tag).
It was indeed EitB, as I mentioned.
I've tried a few more games, and it seems that Averax is also a part of the problem. He expands rapidly due to Expansive and Barbarian trait, so most of the time he's got a great power base to build a lot of Pyre Zombies to steamroll everyone.
That's not my experience, as I mentioned in my previous post. If the people I play with declared war to me as soon as they see Loki, that would serve my purposes too. I would just use Loki to annoy them and move it around their lands while ignoring their declaration of war and focusing on my economy.
You don't need Loki for that, you can do that with a scout. War weariness affects both, and unless it is someone like Elohim, it hurts both. Some nations will be even less hurt than Balseraph, those that have less war weariness from their palace.
Anyway, regardless of that, it's a bad feature in strategy games if there's only one right move and only one (poor) counter.
Even with that, it leaves the issue of AI vs AI.
Early game economy is vital and if they don't focus on it they will be defeated.
Quoted, bolded and generally emphasized for truth. A few wrong moves can cost you the game. Imagine how crippling it is when one Civ has the option of conquering several cities with one relatively inexpensive unit without declaring war and suffering little or no consequences.
Well, that's the whole point of Loki, messing with other players for your benefit.
Agreed, but taking over cities easily is too much.
I see your points, although in my opinion they are not as severe as you say. Nevertheless, I'm interested in which drastic measures you believe he needs. As it is happening with the Savage case, I'm always open to changing features if a good proposal that solves all problems and is more fun comes through.
As been noted, AI learned how to use Loki, but not how to deal with him. In my latest game, Balseraph have started between Hippus and Khazad. They took over 2 cities from Khazad, built 2 and took over 3 from Hippus. At the time when most civs had 2-4 cities, they had 7 and have crippled Khazad and practically destroyed Hippus. Hippus had just their capital surrounded by Balseraph cities on all sides. Hippus couldn't build another city as they couldn't pass through Balseraph territory.
I'm not really sure I'd like to have AI declare war on Balseraph as soon as they see Loki. That's bad design and poor strategy. The best would be to take away Loki's ability to take over cities. He should be able to:
1) Cause unrest
2) Destroy buildings
3) Steal gold
4) Steal tech
5) Steal map
6) Steal contacts...
that sort of stuff... Stuff that help Balseraph but don't cripple opponents. Even more of a super spy than he already is.
I'm not sure myself... Maybe ability to take over workers? And give all those abilities a significant cooldown so that players/AI have to use him in a balanced way, not just spam the best ability constantly.
I'm only sure he should lose the ability to take over cities. I'm not sure I know how to compensate loss of that ability. Maybe other people have additional ideas.
My original problem with Pyre Zombies was that they were able to kill with just collateral damage. The EitB changes fixed that and, as I mentioned earlier, in my experience they are powerful but not game breaking (against both human players and the AI). I'm not sure if the cap should be increased from 95, though... In ExtraModMod the cap is actually 90%.
Pyre Zombie explosion was heavily nerfed by my nuclear banhammer approach to balance in the first version (from which I hope to have moved away and learned from in these years). When the first feedback posts made me see the error of that change (among others), between the cap values they suggested I chose the lowest one. After that change the Pyre Zombies have always felt right for us, although we probably wouldn't even notice the difference if it went up by 5%.
Even with a cap, they're deadly, and it doesn't scale well. The more units on both sides, the deadlier they are. 3 PB vs 3 Axemen? Maybe the axemen can win. 10/20/30 PB vs 10/20/30 Axemen? Axemen have no chance.
In the first example, even if PB win, they will lose most likely two, which translates to 66% of the force and only slight advantage over the opponent. Having 2/3 of the losses of losing side is not so great.
In the second example, they may lose a few PB (2,3 or 4) which will significantly weaken entire stack of 30 Axemen to have the rest of PB slaughter the remaining Axemen. In that case, PB suffers 10% losses, opponent loses everything.
If it's Epic or Marathon game speed, both issues become even more pronounced. That's why I said, don't rely on EitB too much. Sure, MP is very good for balancing, but EitB is balanced around practically only MP and they always (or in 99% of the cases) play on Quick. Again Qqqqqq, no disrespect, you did a hell of a job, it's just apples and oranges.
Loki is much more deadly on Epic and Marathon. Building a settler usually takes between 40 and 70 turns on Marathon. Loki can scout entire map, take over 10 other cities and come back to steal the one you're going to build with that second settler.
EDIT: Or maybe give Loki the Hero promotion and have his unique abilities unlocked as he gains XP, like Combat I/Combat II... and have the ability to steal cities the latest promotion when he gains over 80 XP? That way, AI can build a few cities and start producing culture in them so Loki can hurt them but not cripple them. He could steal steal a city or two, but if AI looses a 5th or 6th city, it can cope, and by that time, AI usually has an army so it can go to war to take them back.
Or move Loki further in the tech tree, so that he can't appear until everyone has already built a few cities?