Sleep and zombies, or zombies in general

[Bahgtru]

It's really new and different though. The basis was that no one liked undeath immunity because it felt forced, and like being robbed of an option. "Body reverts on death" satisfies both the game balance needs (it prevents the WW factory you described, and that I can see now) and partially the needs of those who didn't like undeath immunity: nothing to get out of death AIs, and another case of nothing to get out of lairs. Sure, we'd probably prefer to get WWs from AIs as well, but this meeting in the middle. The middle is not on the circle, where we've been so far

It also - btw - puts a weakness back onto an otherwise still last proven OP unit: removing death immunity is meaningful on a unit that "is designed to beat each common and uncommon". I would rather that death AIs were not completely helpless against my werewolves, at least until I bloodlust them.

It's also super flavourful, come on: everyone knows that WWs are mysteriously found like the original body when they die

[Nelphine]

Yeah but that's about the same thing. The number of picks is constant, the amount of nonbook picks available is very limited (max 4 retorts but the average they'll roll is closer to 2.), death books generally take away from their ability to pick nondeath books and vica versa. Basically the AI picks between 8 and 12 books, a deviation of only +/-2 books from what was assumed (10 books = 0-10 death rest nondeath). That's not large enough to be worth a more sophisticated formula, even if we could have one.

No, particularly on higher difficulties, it's reasonable to see an AI with between 8 and 15 books (I've seen both lower and higher). The ai picks up extra books from treasure as well. And that's a huge difference in strategic combat strength. (The 15 book ai with the same casting skill has 88% more strategic strength??) Decrying my example for using 4 books, when that was just for illustrative purposes, does not help your position.

Even the base difference between 8 books and 12 books on expert is a difference of 50% strategic strength. Given how carefully the AI chooses battles, after the AI reach about 100 casting skill or so, it generally means the one with more books wins.

Similarly, seeing 7 death and 3 other books, or 5 death and 5 others, is super common. But even though it's a 2 book difference, with the same total number of books, the first death wizard should only be doing 20% more death spells, not 40% more.

The problem is that you need to look at difference to each other, not difference to the expected. Difference to expected is fine for treasure hunting, but not for ai vs AI. But even then, only for strategic combat. At high casting skill (say, breaking towers), even treasure hunting is a problem for undead creation.

Edit: admittedly, the absolute totals for undead creation won't mean much. But the comparative totals for ai vs AI combat can have a lot more meaning.

[Seravy]

Actually now that I think about it. 8 or more books give a cost discount. So any Death book above the 8th, which is likely the one responsive for the unexpected fluctuations (especially 11 and up) does increase the amount of spells the AI can cast from Death. So yes, more Death books mean more spells, and more strategic strength.

It's an approximation anyway, we could say the AI is actively pursuing making undead as much as they can without risking the outcome of the battle, and make it like always 50% instead of 10% each death book, if they have any Death book and still wouldn't be wrong.

[Nelphine]

Yes, more than 8 books does mean that. But ai very rarely have more than 9 books of a given realm. I've never seen one with 12, and I don't think I've seen 11. 

The biggest fluctuation, is not from more books of one realm. It's just more books. 

6 and 6 is the most common version of 12 books. And that AI, with 200 casting skill, should not have 50% more strategic strength than the AI with 6 and 2 books and 200 casting skill, which is not an uncommon ai either.

[Seravy]

Wait, which of the books are Death in that example?

If it's 6 death vs 2 death, then being stronger (by 200%, not 50) at undead raising makes sense.
If it's 6 nondeath vs 2 nondeath, then being equal doesn't make all that much sense but again, that's not a valid AI wizard. They can be 6/4 or 7/anything but not 6/2. But let's say it's 7/0 vs 7/5, then the 7/5 not getting fewer undead is definitely not ideal, but I'm willing to overlook that under "the AI is trying to use more undead raising spells to gain advantage", since that's what a human would do in all the battles where they can get away with it.

If you mean the spellcasting strength not the undead raising one, yes, the 7/5 wizard will be stronger than the 7/0 wizard and that makes perfect sense - the 7/5 wizard has a lot more options to choose from so it has an advantage. I'm going to win more battles if I have 5 Life and 5 Chaos books too, compared to only having 5 Chaos books.

[Nelphine]

More options is not the same as a direct increase in power. You might be stronger, but not nearly twice as strong. If you can already case flame strike as chaos, there's not a lot else that death actually adds to your strength. If the ai has 8 life books and 1 of each other realm, he is not 50% stronger than a mono life with 8 books after he's already reached 200 casting skill.

Anyway, I've said my piece. I don't think the AI with more books should default to being the winner in late game AI vs AI wars, regardless of the fact it's difficult to calculate the exact value.

Generally casting skill (and therefore research) is going to be a much better judge.

[Nelphine]

And yes a wizard can be 6 2. It just means it picked an off realm instead of a retort. Although 6 1 1 may be more likely I guess.

[Seravy]

No, they can't, Mono AI is 7 to 10 books. Dual is 4-6 in each.

[Nelphine]

Huh. Didn't realize that low point cut off. Fair enough.