Experimental version

[Seravy]

Found a major bug.
at $DE75B I replaced 6 with F when I was supposed to do it at $DE75E instead.
Don't use EXP 2, it's going to corrupt your data. I'm going to delete it right now, hopefully I'll be done with EXP 3 later today.

[Catwalk]

Their offense was increased a huge amount on anything with more than 1 figure, so that isn't terribly surprising.

It's not surprising, but it's slightly surprising that I can still take out Great Wyrms. On my first attempt I did 53 damage with a full stack, well within the safety margin. So they should be every bit as effective at early node hunting as before. Sprites need to lose flying or their ranged attack, otherwise they'll remain game breaking unless you turn them into a weak unit.

[Catwalk]

Found a major bug.
at $DE75B I replaced 6 with F when I was supposed to do it at $DE75E instead.
Don't use EXP 2, it's going to corrupt your data. I'm going to delete it right now, hopefully I'll be done with EXP 3 later today.

I'll keep testing with it a bit, I did a fresh install anyway. You're doing a great job at keeping these bug free.

[Catwalk]

I'm also able to take out Stone Giants in nodes (14 shields and 25 health).

If I take 10 nature books + Specialist + Conjurer I'm getting a 50% reduction on nature summons. Shouldn't it be 25 + 15 + 16 = 56%? Or did you somehow make the 16% multiplicative and the 40% additive?

[Seravy]

I'm also able to take out Stone Giants in nodes (14 shields and 25 health).

If I take 10 nature books + Specialist + Conjurer I'm getting a 50% reduction on nature summons. Shouldn't it be 25 + 15 + 16 = 56%? Or did you somehow make the 16% multiplicative and the 40% additive?

Not sure where the 16 is coming from?
The old system was 12 (2 for each book after the 4th), the new one is 10 (5 for the last two)

[Nelphine]

I think research is 16, maybe he got that mixed up?

[Seravy]

Update!

Experimental 3
-Fixed EXP2 bug : AI Stalling tactic stopped working
-Fixed EXP2 data corruption and crash bug
-New AI research selection procedure - The AI will now select their next research better.
-Just Cause research cost lowered from 300 to 200.
-Stream of Life casting cost lowered from 300 to 200.
-Gaia's Blessing casting cost lowered from 300 to 250.
-The AI no longer has doubled summoning priority when the circle is not at their fortress. Base priority of summoning increased from 100 to 120.
-Fixed : Uranus' Blessing unlocks “disjunction priority” category for AI selection as though it was Disjunction True while Spell Binding does not unlock it.
-Increased the AI's base priority to cast buffs for each tier : (25,60,200)->(60,120,250)
-AI power distribution : If the AI has no common summoning spell, it'll spend on research between turn 10 and 30.
-AI starting spell picks - Heavenly Light is now the AI's 3rd Life spell pick and Healing is the 4th instead of the other way around.
-AI starting spell picks - Just Cause will not be picked if the AI has no common summoning spell - in this case they'll pick Holy Armor instead.
-Fixed - the AI didn't pick their starting guaranteed uncommons when having exactly 8 books.

As this is an emergency patch to fix the crash bug, I still haven't had the chance to play an actual game to test any  new features for balance and enjoyment.

[Catwalk]

Yeah, I misread the 16% research bonus. My bad.

I agree with the notion that picking a low number of books and a high number of retorts is still the way to go. You get way too many common spells at low amounts of books.

[Seravy]

You get way too many common spells at low amounts of books.

Way too many? It's 3 for the first book and 1 for each additional, I don't see how it can be any lower than that unless you want the first book to contain no spells at all.

...maybe you mean starting commons.

I suppose with the new research procedure in EXP 3, I can in theory drop starting commons entirely, make them "guaranteed" instead, and force the AI to push their research up and get them anyway....but why do that?

The game would be back to "set initial research and production, hit next turn 10-20 times then start playing". Boring and achieves nothing - we have had enough of that in vanilla MoM (20 turns for a builder's hall, yay!).
If anything, this would make starting commons STRONGER. Everyone will pick (or research if we remove starting commons entirely) a summon first...and those summons will obliterate all the early outposts even easier - nothing to stop them. Normal units won't - summons are stronger and the city is producing settlers anyway - and combat spells are not researched to do it.

Your argument of "starting spells = less books more retorts better" sounds flawed. Picking 12 books, 6 in each of two realms yields the most starting spells and the highest power base and starting skill to use them. Yet you say more retorts are better.
If starting spells are the problem, the choice that provides 10 of them would be the best not the one with 5. So if what you say is true the source can only be retorts being better than starting spells, and all other benefits from a book, combined.

[Nelphine]

No, the problem is that cities are your base ability (as I stated elsewhere), c. Some retorts multiply this ability, r; some spells multiply this ability, s; some spells add to this ability, m; some spells multiply that spell ability, s; and some retorts multiply that spell ability, r.

Total ability = c*r*s + m*r*s.

You can pick retorts and starting spells that all multiply the same thing, which gives you the highest total ability at the earliest point in the game. Taking more books never increases that, because the more spells you have, the more likely you are to have spells that don't multiply with each other. (for instance, having a summon and holy armor. Holy armor multiplies your city power, summon adds to your city power.)

In the long run, this can be strong, but when you need to maximize your strength due to ai cheating at high difficulties, you can't afford to wait for the long run. A strategy that allows all 12 picks to be multiples of each other, instead of additive, is a better strategy. Since most of a spellbooks power is random (even if in the long run its much more powerful than a retort), you can't guarantee that multiplication early on, which means you can't choose it.

Thus, retorts + starting commons (even guaranteed commons) will always be better than more spellbooks for the highest difficulties, due to the amount of cheating the ai does. Exactly where the breakpoint is depends on what exact strategy you are going for.