SORCERY Realm

[Nelphine]

So if I know all 10 common spells for 4 realms, I'm getting (5+6+7+8+9+10+11+12+13+14)*4 points?

So 380?

Then I know 10 chaos uncommons, ~8 sorcery uncommons, ~5 nature uncommons, ~4 life uncommons, so that's another (15+16+...+24) + (16+...+23) + (18+19+20+21) + (18+19+20+21+22) - note I'm averaging since I don't know what spell ids I actually have.

Total of 909

Then I have ~6 rare chaos, and ~4 other rares, or another 300 or so.

Total 1209.

Then I have 1 very rare, assume it's second highest, so it's 43 more.

~1250 or so from spells.


He has 4 very rares, 18 Commons, 15 uncommons, 10 rares.

So he should have ~ 158+295+292+171 = 916

So then I have 466 skill, so another 2330, so my total is 3580. 

His total is ~90% of mine, so he has around 3200. So 2300 from skill, should still be around 460.

Assuming I've made some erroneous guesses, he could be up around 520 or so casting skill to cast spell binding instantly.


So I'd like to suggest a change to the formula. Higher tier spells are worth 4 times as much research as the previous tier, which is why I assumed very rares were worth so much. But instead a very rare might be worth 4 times a common.

And a very rare only being worth 8 skill seems really low. Those spells are huge game changers. I'd expect closer to 15 (6 very rares worth 100 seems right?).


I'd like to change it to ((round down((spellID+1)/10)*3)^2

So all Commons would be worth 9, uncommons 36, rares 81, very rares 144.

Then if needed, increase skill up to 7-10.


This is off the top of my head, so I'm certainly open to changes, but I definitely think rare and very rare are vastly underrated in the current system.

[Seravy]

I'm against changing research value. Spells of any tier are worth a lot by opening up new opportunities, so we don't want a very rare to be worth over 10 times a common. The bar does NOT represent how much resource the player spent on the spells, but how much of a threat they can be through their spell quality, quantity and diversity. This is intentional.

Knowing all spells in a realm is worth sum(0-39) = 390 plus 5*40 = 200, a total of 590 points. This is equivalent of 118 skill. That is not much. If we want research to be an equal part of the bar, then it should be around 300-350.

However, as is, the bar is a fairly good representation of casting power, while research is kept marginal in it. That provides useful information. If both were equal, guessing either would be harder, especially as wizards with more realms would usually rank higher on spells even if they are at the same tier. Also research is something the player knows anyway - either through Detect Magic, simply playing attention to what spells are being cast during combat, or trading etc - they can estimate where the AI is in their research queue especially as the spells are chosen in a specific order by the AI that never changes. There is no need for research to be something the player figures out from that bar. In fact, they better not - it's often more useful to know the individual spells the enemy has than the generic position where they are in the queue. Knowing they have Flame Strike or not is more useful than knowing "this AI is about to enter the rare tier or maybe already did.".

So I don't really see how the research being equal to skill in the bar would make the game play better - if anything it would be worse, not just for the above reason but also because research is something that doesn't change much in the endgame, so the larger weight it has, the less the AI will trigger the "player wins" feature. Skill can at least be reduced somewhat by destroying amplifying towers, research is 100% permanent.

[teelaurila]

"the bar is a fairly good representation of casting power, while research is kept marginal in it. That provides useful information... research is something the player knows anyway"

On the other side one should be able to rather easily deduce enemy casting skill by the combat spells cast in a single combat. The AI spends all, unless starved for mana, right? Of course that assumes war.

[Seravy]

Yes, the main difference is, you can know their spells without war. At least the overland ones and those that appear/not appear but should in trades, but that is enough to deduce where they are in the research order, which is already more information the bar could ever give.
Casting skill is pretty hard to guess without a war, as overland spell costs are modifier by a lot of factors and carry over between turns.

Ultimately however the bar is to show the player how dangerous the AI is. While having a large spell diversity does make the AI harder, having a higher casting skill is much more effective at that - 5 doom bolts are better than the choice of using a doom bolt, fire bolt, warp lighting, reaper slash or exorcise.

[Nelphine]

I disagree. An AI who has very rares is far more powerful than one who only has uncommons, even if they both have 400 casting skill.

Similarly, the ai who has 300 casting skill and rare spells, is often weaker than the AI who has 250 casting skill and very rare spells.

Each tier is very important.

Further, an AI with 3 realms and 100 skill vs a mono realm AI with 100 skill, having spent the same amount on research, the first is much weaker - but they will have researched far more common spells, and so they will show up as stronger on the graph.

Each tier has great diminishing returns - your first 5-9 common spells rally matter, but the next 10-30 really do nothing to add to your power. Yet you're forced to research them first, and without the higher tier spells registering as being worth more, it will look like you're stronger.


Note I don't really care about the comparison to skill. But if a common is worth 1-3 skill, there's no way a very rare is worth only 7-9 skill.

If you use my numbers I'm fine if you make skill worth 20; then a common spell is worth half a skill, and a very rare is worth 7 skill. I'd be fine with that. It's the comparison between tiers that matters to me.

[Seravy]

It's the comparison between tiers that matters to me.

Spell Power is a sum of Research and Skill.

Since we assumed the research and skill spending ratio is a constant during a typical game, we can assume the current existing research costs follow the same formula as skill itself - we don't even need to think about the formula. We assumed the player to have an average of 30 skill for common, 70 for Uncommon, 150 for rare and 250 for very rare tier. So if we want a bar that has fair proportions regardless of where the resource is invested, we'd need the ratio of tiers to be 3/7/15/25. Except, not really. Tiers are cumulative. So what we really want is :
C = 3
C+UC = 7
C+UC+R = 15
C+UC+R+VR = 25

So C=3, UC = 4, R = 8 and VR = 10.

While that's not exactly the existing ratio (3,6,9,12), it's fairly close to it and actually overrates higher tiers. Do note this is for the middle of each tier - the end did include two spells from the next tier so it's less appropriate to use, but if we used those, we'd get 4,10,20,30 which translates to a ratio of 4/6/10/10 = 2/3/5/5, even less difference than the existing.

So assuming we want the bar to be an accurate measurement of the total SP and RP spent by the player, the current distribution of tiers is near perfect. If not then I have no idea what you want it to show...

But this only really matters if we want to change the bar to have the research and skill at equal weight, otherwise the one at lower weight doesn't even matter - and currently research is that one.

[Nelphine]

That scale is for an imaginary average. When we said 30 70 150 250 we didn't need to worry about the difference between a mono realm and a triple realm player. These players will have different actual strengths in game - the type of research they do is different. The triple realm gives up individual power for a wider range of versatility. But with the same research cost, they can get 10 times as many commons as the mono realm gets rare spells. If rare spells aren't weighted heavier, and the two have the same casting skill (and the common assumption is that ai's of roughly equal size will have roughly equal skill) then the triple realm player looks much stronger, even though the rare spells are much more game changing.

Similarly, our basic scale does NOT get influenced by treasure - finding spells doesn't change what research or skill a player gets. 

Yet finding spells certainly affects the spell power bar - and most of this spells are Commons, and Commons that aren't good enough for the player to bother starting with them, which means they don't noticeably add to actual power. Again, assuming equal skill, the player who finds 8 Commons they never use, looks stronger than a player who researched 2 rare spells.

[Nelphine]

Again, I'm fine if you want to weight skill crazy high. But you still need to weight the spells by actual power compared to each other. Either pick Commons, or very rares, and weight them against skill however you want. Then weight the rest of the tiers compared to that tier.

[Nelphine]

Also currently, assuming you have a random split of spells, the current scale is 10/20/30/40. While the early ones have a bigger jump than 4/6/10/10, the ratio of very rare to common would be closer to x7 instead of the atrocious x4 we have right now. Now admittedly, that is still too low; we want 3/8/21/50 or something.

[Seravy]

then the triple realm player looks much stronger, even though the rare spells are much more game changing.

I have to disappoint you but this isn't true. Every book except the 2nd, 7th and 8th has exactly the same number of spells in it. So the triple realm player would end up with the same amount of spells, but fewer of those would be rares, so their bar would be somewhat shorter.

Of course treasure and trading can add to the total, making them end up with the larger bar, but the treasure and trades can be rare or better spells, they aren't limited to commons. And trades in particular have a price - someone else also gets to have more spells and a longer bar in exchange - while treasure has limited availability. If this is a reason to worry, we can make sure the AI trades with other AI players more often to minimize the advantage the human is getting here. I would disagree players trade for useless commons however, unless they are in the lucky situation of the trading partner being already beaten.
And while the "more = better" statement isn't always true - some spells do have overlapping functionality - this is true even within a single realm. Having all of Colossus, Great Wyrm and Behemoth doesn't make you much stronger, just like how having all of Fire Bolt, Cold Bolt and Life Drain does not.
As long as the spells are functionally different, they do contribute, often significantly, regardless of tier, but obviously higher tiers contribute more. And at this point we are talking about something entirely subjective - what's better, having Healing, Web, Wraith Form, or having Great Wyrm when already having Colossus? I would rather have the former, but if someone said they think the latter is better, I couldn't say they are wrong either.

Again, I'm fine if you want to weight skill crazy high. But you still need to weight the spells by actual power compared to each other.

No I don't. If it makes up only 25% of the bar like now, it's impossible to actually tell how much the research part really is, regardless of the system. And it's most definitely not worth hours of work in what probably is a quite limited space. (haven't yet checked)

Unless we first agree we want skill and research to be of equal weight, any discussion on how research should be distributed by tiers is a waste of time.