SORCERY Realm

[zitro1987]

((Skill-X)/Y spells (rounded up), where Y should be 150-200 and X should be 0-150 most likely.

(Skill - 150) / 200 with a cap at 3 if playing with this formula. A strict cap is needed as it can be easy to get to 4 or even 5 with a denominator of just 200 (with the right spells/retorts and AI bonuses, especially if myrran).

I do prefer my 1+X/400 (seems easier to explain)

[Seravy]

Wait am I missing something? Spell binding by itself should no longer have any effect on the opponent casting their spell right? It's just that you'll have spell blast and could stop the opponent from casting it? (Which you could do regardless of you having or not having Spell Binding?)

Exactly. And while you are casting Spell Binding, you can't use your Spell Blast.
...though one other detail I didn't realize, this also applies to casting your stolen spell. So every time you want to cast your stolen Armageddon, you are giving your opponent a chance to cast theirs.
Old Spell Binding didn't do this - it gave them the window of opportunity, BUT once finished, it took the spell away anyway (you just had to pick that spell instead of your original choice first).
So maybe the high cost on Spell Binding itself might not even be that important. You have to pay for that stolen Armageddon, too. Albeit both having a high cost means the initial casting gives the opponent(s) an even larger window and risks multiple spells going through in exchange for gaining one. Might even be a bad deal for the Sorcery wizard. So we need to be careful not to "overnerf" things, too, there might be potential for that.

That means, with both sorcery skill spells, I'd say +125 or so is a basic minimum (aether binding alone is ~50? Skill by very rare phase).

I guess I can agree to that, but Uranus' Blessing isn't exactly a cheap spell. Casting it on every city is a nice long term plan every Sorcery wizard should have, but assuming it already happened by the time you get your Spell Binding might be too idealistic. (Albeit with the current research system I guess that is much more realistic than before, so maybe I'm overthinking this)

Still I prefer to err on the side of more spells than fewer - players might be playing on below Fair land size, or have other reasons for a low casting skill. So I'm not willing to have the first threshold any lower than 200, but anything between 200 and 300 is probably acceptable.

Do note that Sorcery is supposed to encourage peaceful strategies, so the player only having 15 cities is something you should consider. Which means only 230 skill in 1420, see my game. (I wasn't Sorcery though - if I was, I would have had over 300 skill. Less than 350 though, and 1420 is kinda late, too.)

I think the third spell should be determined by where you want master AI who push to fall when they research the spell.

The AI uses the same costs for both RP and SP as the human. Thus, the time when they hit Spell Binding will have the same casting skill as a human. It will happen X years earlier than the human due to the resource bonus, but that's about it, unless the AI distributed their power very differently than an average human. In my game the AI had ~400 skill when they got SB, so your 500-550 is an overestimation. They had that much 4 years after getting SB.

(Skill - 150) / 200 with a cap at 3 if playing with this formula.

So, 0-349 : 1 spell, 350-549 : 2 spells, 550+ 3 spells?

I'd probably prefer the cutoff points at 300 and 500 (for the human)
However if we set it up this way, the proposed priority system would not work - not enough cutoff points to have those categories.

[Nelphine]

Master is going to have more resources, which means more cities, which means more resources. More resources also means more nodes, which means more resources. More resources also means more heroes which means more resources. More resources means more spells which means more resources.

Even assuming only a 5% gain over expert per category, most of that goes to skill increase, not faster research, at least for a human (even though you have more skill, you have more resources you need your overland skill to enhance, so you don't need to research anything faster or you can't use the spells you already have as per your Uranus blessing comment), so 5% per category means a little over 100 extra skill by the time very rare comes around. Which is in that 500 to 550 range.

[zitro1987]

Yeah, having 300-500-700-900-110 cutoffs might be more intuitive and balanced for a powerful spell (if capping at 5, thus maintaining your priority system)

I think Nelphine's idea of human gaining skill is a bit idealistic or reserved for the best of players.

[Nelphine]

And I know I'm faster than most humans, so I qualify as deep in the 'pushing faster' category, but I typically have 200 casting skill by the time I research my first rare (not including spellweaver). As barbarians. I think I've got 3 amp towers in my current game and I'm not sorcery.

So if someone takes some of my techniques for building skill, and applies them to a dark elf sorcery strategy (not all will work obviously, that's fine), someone pushing skill will easily be in the 550+ range by the time they get Spell Binding.

Do you want humans able to start with all 3 spells? (I personally don't. No not everyone is going to push that hard, but the spell is a nice opportunity to grow power and reward pushing that hard.)

[Seravy]

Even assuming only a 5% gain over expert per category, most of that goes to skill increase, not faster research, at least for a human

Yeah, "for a human" is the keyword.

The AI decides on power distribution based on objective, and by looking at whether they are behind in either category compared to the human.
Since we are talking about high difficulty and assumed the AI is ahead (in skill), they won't be prioritizing skill for that reason.
So it's only personality, and the mana/skill ratio. But the mana/skill ratio for humans and AI is identical during peacetime. (AI might spend more mana during war, leaving less to spend on skill)

I don't think anything in the AI power distribution exists that makes the AI weight skill over research consistently, but do tell if I missed something.

Do you want humans able to start with all 3 spells? (I personally don't. No not everyone is going to push that hard, but the spell is a nice opportunity to grow power and reward pushing that hard.)

Let's not balance the game on the assumption that everyone plays the game as well as you do.
I don't have more than 200-400 skill (varies greatly per game) at the time I have my first few very rares, and I almost never go beyond 400-500 unless the map size is Huge.
If I could, I've already won the game.

For AI purposes the ideal cutoffs would be 400 and 600, but any formula that allows the AI to delay their middle choice(es) until 500 skill should be good.

300-500-700 would...put their last cutoffpoint 100 too high. I don't think a 700 skill AI is realistic to happen unless they are a runaway Myrran (which we want to fix), the difficulty is Lunatic (which might even be made easier so it's not much of a concern right now), the land size is Huge (which is inherently broken), or the game is way past 1430 (in which case the human likely deserves to lose), but I might be underestimating the AI here, idk.


It also means the AI will never delay their choice until 500 - as they naturally get a choice exactly then. That eliminates the possibility to have priority tiers where the AI uses delaying - instead they never have that pick to spend earlier in the first place (which is equally good I guess?)

Maybe we should first decide how many picks the AI should be getting for various casting skills.
I do think the AI should spend at least 2 of their picks after a "cutoff" point where there is a high chance of better spells already being researched, so around 500 and 600 skill respectively. They should get at least one, maybe two picks immediately when they get the spell (400), and they likely shouldn't get more unless they hit extremely high (700) skill.

Meaning we want :

Skill Range, Human spells, AI spells
000-299, 1, 1-1
300-399, 2, 1-2
400-499, 2, 1-2
500-599, 3, 2-3
600-699, 3, 3-4
700-any, 4, 4-5

The first AI option is bad - it makes the AI use the spell really poorly until 600 skill.
The second option is good but results in the AI requiring an additional pick at 600+ skill.
However, the cutoffs must be 500 and 600 I believe - if earlier, the AI risks using up all picks before good spells appear, OR has to hold back using the spell for too long, resulting in a disadvantage. If later, well, the AI won't really get that far in skill under normal game conditions.

...Also AI with Spellweaver will completely break the system by wasting their picks early, unless we force it not to by pretending it does not have the additional skill from the retort.

...unless someone has a better idea on how to make the AI decide when to use up their picks on spells.

[Seravy]

Alternate idea : what if we drop the casting skill thing entirely and instead make it work like

"If you used this spell, it requires 2 years to recharge for each previous use."

So, if you research this in 1420, then you'd steal a spell in 1420, 1422, 1426, 1432, 1440, and...well, game is probably over by this time so no one would ever get the next in 1450.
Basically you'd get spells on year 0,2,6,12,20,30.

This would be AI friendly - the AI could be sure they'll get another shot at it to use it for gaining a better spell eventually, and we could base the priority on the number of uses (for example do not take priority<100 if number of uses>2 and a better spell is unresearched somewhere)

(or we can use a different formula for the turns if we don't like the 2X years)

It also eliminates the unwanted interaction with Spellweaver, skill boosting spells, and there is no possibility of influence from AI resources and Objective.

[Nelphine]

I wouldn't want to use 2 years. On higher difficulties 2 years is most of the game. It would mean most AI would never get a third cast. And some wouldn't get a second.

Maybe 8 months? No, for some people it might take that long just to cast it and the new spell. Blegh. Variance is too high.

How bout research base? For every 4 (average cost) very rares (or equivalent) you've researched you get one use, to a maximum of 3 or 4? (I should really look up the costs of the lower tiers.. ideally all 10 rares plus Spell Binding itself would be about 4 or 5 very rares? )

Treasures wouldn't count, but research into SoM would?

[Seravy]

I don't want to base it on research. Research is already something you want to do most when playing Sorcery.

If you can win the game in two years, you didn't really need Spell Binding anyway.

[Seravy]

If we make it 18 months (1.5 years), then it would be
0, 1.5, 4.5, 9, 15, 22.5
maybe that's better?

The other option is 1 year which would result in
0,1,3,6,10,15,21