Posts: 6,457
Threads: 134
Joined: Aug 2004
Quote:Quote:The real problem with spamming early spells is that you get spells for free. This was poor design from the start (in the original game), and it should be fixed.
This is a train of thought that I often ride on...only to get slapped in the face by reality. The AI needs the starting commons and would completely fail without them. The main reason why they don't get stuck on islands is the starting spells, for one. And honestly, the ability to pick two books for a starting common makes the game more varied and interesting - losing this feature would be sad. Commons spells have a high availability - even if you don't get them at start you'll most likely get all 10 of them with as few as 1 book - by trading, looting wizards, and finding them in lairs. So a spell being 'guaranteed' doesn't have a lasting impact. In fact, 6 books are enough to make sure you have all 10 spells.
This discussion was from the Sprites thread, where we discussed the problems with early game summoning power in the hands of a strong human player. A big part of the problem is that early summons have a prereq cost of zero: You start with the spells right away. Normal units have high prereq costs, like 250 hammers for a Fighter's Guild. That makes early summons completely dominant for the first couple of years.
A related problem is how it's easy to be certain of what spells you have available to you. We all agree that we don't want too much certainty when it comes to spells, because then you'll just use the same thing every single game. I believe Seravy used that argument previously, and I agree with it. But then why do we have absolute certainty when it comes to common spells? This is not only a problem with summons, Life is also a good example as you can get a complete unit buffing package with just 4 books: Heroism, Holy Armour and Endurance. You can create units that are all but invincible using that and it's guaranteed with having spent just 4 picks. Seravy argues that the AI needs to be certain to get a transport spell: Wraithform, Water Walking or Floating Island. Those are the top picks for the AI, and that's a very good idea.
Also from the other thread, we have a huge problem with high book strategies being useless. Being able to pick up another realm or strong retorts is almost always better than going for more than 6 books. There is simply not enough incentive to do this. I agree with Seravy that higher casting discounts and research boosts can be dangerous (we don't want to get too close to zero cost, as in the original game), so this cannot be a primary method for boosting high book strategies.
Lastly, not having any control at all over mid and late game options is discouraging. I saw Seravy express this when he was playing a Nature Summoner recently and was dismayed to find that he could not get Giant Spiders. This should not be completely up to chance. We need to find a balance where we keep the unpredictable elements but also add just a little bit more control over long-term plans.
Seravy, before I present my proposed solution do you agree with this description of the problem at hand? I want to make sure we're on the same page before we start discussing specific proposals. Summary:
1) There is too much certainty of starting common spells, and you get it far too cheaply
2) There is way too little certainty of available spells past common, what little certainty you have comes at a cost so high that it's never worth it
3) Powerful early spells can completely dominate the early game because you get them for free, without having to research them (this also devalues research as a relevant strategy, why bother when you have plenty powerful options right away?)
March 12th, 2017, 08:03
(This post was last modified: March 12th, 2017, 08:04 by Seravy.)
Posts: 10,463
Threads: 394
Joined: Aug 2015
1. I think I both agree and disagree on this one. On one hand, 9 starting commons is a lot. On the other hand, you already have a very high certainty of getting commons inherent in the system which cannot be removed.
-With 6 books, you are guaranteed to have all of them for research.
-You pick starting research out of commons, most likely all of them will be presented, so you get what you want there as well.
-Even if you have fewer than 6 books, it's very easy to trade for common spells, and the likelihood of finding them in lairs is very high, plus, you get a fairly high number of them for research so the gap to fill is small.
So this is more about "when" and not "how many" - you'll get all the common spells anyway, even if you don't start with them.
It's probably worth mentioning the starting spell system wasn't part of MoM 1.0 - it was added in a later patch, most likely for player demand. I don't remember those times very clearly, but I assume starting the game without being able to cast spells was too boring.
I feel the 3.31 patch also has an influence on this, with the addition of the starting Sawmill, the player can start building their normal units earlier, improving the summon vs normal unit balance.
One more thing : reducing the number of starting commons makes "early uncommon" strategies impossible. With commons cluttering up the research options, you'll never have that "Giant Spiders" or "Lycanthropy" or "Gargoyles" spell early enough to matter - while spiders most likely perform well even later, the other two probably doesn't.
2. I regularly play 8+ books and rarely play fewer. The certainty of having a higher number of very rare spells is a big deal, even if the specific spells aren't guaranteed without 10 book (which is a large incentive to get the 10th book).
Guaranteed uncommons might improve the game (if the amount is kept low), but guaranteed rares or very rares would go towards the opposite direction of what you intend - I would pick fewer books and go with "I have the one very rare I need to win the game, no more books are necessary" instead of wanting more books.
I believe the problem here is retorts giving too much early advantage - if the player can win the game before reaching rare spells, the additional books are indeed unnecessary. Fixing the problem with overly effective early strategies should fix this problem on its own.
3. I completely agree with this one. However the AI was taught to play in this system, it would be very hard to make it work otherwise. It puts all power into mana in the early turns, and relies on always having a summoning spell (or buffing for Life wizards).
I would like to say first that I haven't looked into the possibility of implementing a guaranteed spell system yet. It might, or might not be possible, I have no idea.
And I'd like to add this :
4. It's not only those 3 spells the AI relies on. All of their picks have been carefully selected to give them a fighting chance. A good combat spell to be able to actually do something when their outpost with 1 swordsmen gets attacked. An overland summon to make sure they have armies to retaliate or defend themselves. etc, you can read the entire discussion in the AI thread, somewhere along the first few pages.
Posts: 6,457
Threads: 134
Joined: Aug 2004
(March 12th, 2017, 08:03)Seravy Wrote: 1. I think I both agree and disagree on this one. On one hand, 9 starting commons is a lot. On the other hand, you already have a very high certainty of getting commons inherent in the system which cannot be removed.
-With 6 books, you are guaranteed to have all of them for research.
-You pick starting research out of commons, most likely all of them will be presented, so you get what you want there as well.
-Even if you have fewer than 6 books, it's very easy to trade for common spells, and the likelihood of finding them in lairs is very high, plus, you get a fairly high number of them for research so the gap to fill is small.
So this is more about "when" and not "how many" - you'll get all the common spells anyway, even if you don't start with them. But when is a HUGE deal! Or in the words of the leader of the free world: A YUGE deal! I have realized that the problem when discussing with you is that you have an extremely long-term perspective. I've looked through some of your videos, and you're one of the slowest players I've ever witnessed  It absolutely does make a difference when you get them, and there's a huge difference between being able to cast Sprites on turn 1 and having to spend 100 RP to get them. It would completely obliterate my strategy, for example (which would be a good thing). Or it would require me picking a higher number of books in order to start with it, and even more to also start with Earth Lore.
Quote:One more thing : reducing the number of starting commons makes "early uncommon" strategies impossible. With commons cluttering up the research options, you'll never have that "Giant Spiders" or "Lycanthropy" or "Gargoyles" spell early enough to matter - while spiders most likely perform well even later, the other two probably doesn't.
This is incorrect. The difference between spending 400 RP (Giant Spiders) or 600 RP (2x Common + Giant Spiders) is negligible. How soon you're able to get a power base is massively more important than having to research common spells. I'm also suggesting reducing the number of random commons you get, you won't get all 10 with just 6 books.
Quote:2. I regularly play 8+ books and rarely play fewer. The certainty of having a higher number of very rare spells is a big deal, even if the specific spells aren't guaranteed without 10 book (which is a large incentive to get the 10th book).
8+ books are weak setups, given how strong the retorts are. This is your personal preference, this is not optimal strategy. Again, you're not a good example to base this on.
Quote:Guaranteed uncommons might improve the game (if the amount is kept low), but guaranteed rares or very rares would go towards the opposite direction of what you intend - I would pick fewer books and go with "I have the one very rare I need to win the game, no more books are necessary" instead of wanting more books.
This is incorrect, as I haven't given you any numbers yet. You're very quick to jump the gun on ideas you don't like
Quote:I believe the problem here is retorts giving too much early advantage - if the player can win the game before reaching rare spells, the additional books are indeed unnecessary. Fixing the problem with overly effective early strategies should fix this problem on its own.
I agree with this to an extent (Rare and Very Rare spells should play a larger role in normal games, even with quick strategies), but that has nothing to do with the certainty aspect. It's a huge luck factor, and it's discouraging. It has nothing to do with retorts.
Quote:3. I completely agree with this one. However the AI was taught to play in this system, it would be very hard to make it work otherwise. It puts all power into mana in the early turns, and relies on always having a summoning spell (or buffing for Life wizards).
Then let's say 6 books earns you a starting common spell and 2 guaranteed common spells. It picks a summon spell or a buff spell as its starting spell and can get to work right away. It picks a transport spell and a combat spell as its 2nd and 3rd picks, and it will get those soon enough to matter. The AI works, and your concerns are addressed without having to recode the AI.
Quote:I would like to say first that I haven't looked into the possibility of implementing a guaranteed spell system yet. It might, or might not be possible, I have no idea.
Of course, it depends on whether or not it's technically possible. I trust you not to use that as an excuse, of course
Quote:And I'd like to add this :
4. It's not only those 3 spells the AI relies on. All of their picks have been carefully selected to give them a fighting chance. A good combat spell to be able to actually do something when their outpost with 1 swordsmen gets attacked. An overland summon to make sure they have armies to retaliate or defend themselves. etc, you can read the entire discussion in the AI thread, somewhere along the first few pages.
Which 3 spells? You listed a summon/buff spell and a transport spell. With 1 starting and 2 guaranteed spells, they can then pick up a combat spell as well. The summon spell is already included.
I've found the starting list in the AI thread, it looks as follows:
I can easily put 3 spells on the top 3 of each realm to give the AI a fighting chance.
Nature: It's overkill to go for both Sprites and War Bears. Substitute one of them for Web or Call Centaurs and you have a nice starting kit.
Sorcery: Confusion + Floating Island + Nagas fits the bill. They can summon, fight and transport. And Nagas are still terrible, but I'll save that discussion for later.
Chaos: No transport spell, but they have the excellent defense spell Wall of Fire instead. Also a good starting kit.
Life: Just Cause is not needed (and is overrated as an early spell unless you're exceptionally hero-crazed), bump it down a few spots and move Heavenly Light up instead. No transport spell, but Heroism + Healing + Heavenly Light makes them very tough to take on.
Death: Ghouls + Wraithform + Black Sleep is also an excellent starting kit. No problems here.
So in all 5 Realms, with just 1 starting spell and 2 guaranteed spell you've ensured that the AI can function. And most of the AIs will either have more than 6 books in a realm or 2x 6 books. They'll have plenty tools available at the start, and will quickly fill out their toolbox. Your argument has no basis in reality!
March 12th, 2017, 09:06
(This post was last modified: March 12th, 2017, 09:18 by zitro1987.)
Posts: 1,333
Threads: 23
Joined: Feb 2012
Proposal Idea #1 - Books provide more spells after 4 books, but no cost reductions until high number of books. (you may switch 1 rare to common if you like)
Benefits to gameplay:
*Books are even more based on having spells, you can have guaranteed spells with fewer books, but pay full price
*Conjurer/specialist combo drastically reduced unless picking 10 books
*Seeing 100% cost in nearly all games is more intuitive, fewer weird spell costs, more even numbers
*Obtaining a new book of a similar realm is not bad luck, you tend to enjoy decent cost bonuses (but not as extreme as original)
1 book: 2-1-1-0 - 4 spells
2 book: 4-2-2-0 - 4 more spells, can trade uncommon
3 book: 5-3-3-1 - 4 more spells, can trade rare
4 book: 6-4-4-2 - 4 more spells, can trade v rare
5 book: 8-6-5-3 - 6 more spells (cost reductions don't begin here, add 2 more spells instead)
6 book: 9-7-7-5 - 6 more spells
7 book: 10-9-8-7 - 6 more spells
8 book: 10-10-10-10 - more spells
9 book: 10-10-10-10 - Begin 6% cost 10% research reduction
10+ : Continue reductions
Proposal Idea #2 - same, but more Seravy Friendly
1 book: 3-1-0-0 - 4 spells
2 book: 5-2-1-0 - 4 more spells, can trade uncommon
3 book: 6-3-2-1 - 4 more spells, can trade rare
4 book: 7-4-3-2 - 4 more spells, can trade v rare
5 book: 8-6-5-3 - 6 more spells (cost reductions don't begin here, add 2 more spells instead)
6 book: 9-8-6-5 - 6 more spells
7 book: 10-9-8-7 - 6 more spells
8 book: 10-10-10-10 - 6 more spells
9 book: Begin 6% cost 10% research reduction
10+ : Continue reductions
In conjunction with the above - if retort stacking is to blame, limit how many picks a player can initially allocate to retorts (suggestion: 5, maybe 6 max).
Posts: 10,463
Threads: 394
Joined: Aug 2015
Quote:This is incorrect. The difference between spending 400 RP (Giant Spiders) or 600 RP (2x Common + Giant Spiders) is negligible.
I didn't mean that. I meant, with 8 slots and only 1 common to research, you have a fairly high chance for the spell to appear. (1 common and 0-3 arcane spells vs all other slots on uncommons). Each common unresearched puts this 1 spell further away, and we are not talking about 2x spells I believe, more along the lines of 3-5.
Also, the spells mentioned have a 200 RP cost, for early availability. Adding 2-3 commons (possibly including 200 cost spells - there are expensive commons!) is a huge deal, it more than doubles the total cost.
Quote:8+ books are weak setups, given how strong the retorts are. This is your personal preference, this is not optimal strategy. Again, you're not a good example to base this on.
Then the retorts are the problem which is what I was saying.
Quote:Then let's say 6 books earns you a starting common spell and 2 guaranteed common spells. It picks a summon spell or a buff spell as its starting spell and can get to work right away.
The AI isn't going to always have 6 books in a single realm. Go with the assumption the AI always has 4 books of its primary and secondary realms, the rest are random.
Also "guaranteed" isn't good enough. The AI is spending zero on research before turn 30 (I can modify that but probably shouldn't - it would impact their performance massively) and won't select those spells first unless we add ALL of them as exceptions in the research procedure.
We seem to agree the AI needs 3 starting spells. But it needs that many from 4 books not 6. Which is exactly the current system, no room for a single point of modification here.
If your suggestion is to keep 3 starting spells for 4 books, and that's where the deal ends, 5th and above books give no spells, sure, there is nothing wrong with it... except one thing. As a human player, if I get to pick 3 spells in the realm, I can execute most of the existing early strategies. Sprites+Web+Earth Lore, all you need.
Yes, I have to give up on some utility spells but those weren't part of the primary strategy anyway. They are unnecessary for it and I still don't need to spend on research - but have fewer spells to use making the game more boring. And to make matters worse, if the primary strategy fails for some unforeseen reason (chaos wizard fire storms the sprites early in this example) then I lost the game instead of being able to adapt to the situation and play differently from a less optimal position.
Posts: 10,463
Threads: 394
Joined: Aug 2015
(March 12th, 2017, 09:06)zitro1987 Wrote: In conjunction with the above - if retort stacking is to blame, limit how many picks a player can initially allocate to retorts (suggestion: 5, maybe 6 max).
This already exists : a player can't have more than 6 retorts at a time. (Both applied for the start and treasure found).
Removing the cost and research bonus from books 5-7 might be a good idea. Increasing it on books 8+ might be as well. Changing the distribution of spells isn't - If I get all spells with 8 books, why would I EVER want more than 8 books? A 6% cost reduction? Every retort is better than that but to make it obvious, look at Specialist.
Posts: 10,463
Threads: 394
Joined: Aug 2015
Oh, one thing. I'm 90% certain even without looking that it's either starting spells, or guaranteed spells for each rarity, and not both. There is one interface for picking spells, afterall. This should only matter for commons though.
March 12th, 2017, 10:06
(This post was last modified: March 12th, 2017, 10:17 by zitro1987.)
Posts: 1,333
Threads: 23
Joined: Feb 2012
Then I'll modify the proposal: how about 8 books being the maximum one can select in one realm? The cost reductions of 6%/10% (could be improved to 8%/12%) could be left only for node/lair awards, not that we're going to really miss this.
let's face it - more spells -> more fun!
spell reductions -> not as much fun
1 book: 3-1-0-0 - 4 spells
2 book: 5-2-1-0 - 4 more spells, can trade uncommon
3 book: 6-3-2-1 - 4 more spells, can trade rare
4 book: 7-4-3-2 - 4 more spells, can trade v rare
5 book: 8-6-5-3 - 6 more spells (cost reductions don't begin here, add 2 more spells instead)
6 book: 9-8-6-5 - 6 more spells
7 book: 10-9-8-7 - 6 more spells
8 book (max option): 10-10-10-10 - 6 more spells
Additional books (via awards) - 8%cost/12%research bonus
This should finally [hopefully] balance books vs retorts ... where books are entirely based on initial power, spells.
Proposal #2 - harder to get spell awards / trading, but even more researchable spells:
1 book: 2-1-1-0 - 4 spells (can trade common)
2 book: 5-2-1-0 - 4 more spells (can trade uncommon)
3 book: 6-4-2-2 - 6 more spells
4 book: 7-5-3-3 - 4 more spells, (can trade rare)
5 book: 9-6-5-4 - 6 more spells
6 book: 10-7-6-5 - 4 more spells (can trade v rare)
7 book: 10-9-8-7 - 6 more spells
8 book (max option): 10-10-10-10 - 6 more spells
Additional books (via awards) - 8%cost/12%research bonus
Posts: 10,463
Threads: 394
Joined: Aug 2015
(March 12th, 2017, 10:06)zitro1987 Wrote: Then I'll modify the proposal: how about 8 books being the maximum one can select in one realm? The cost reductions of 6%/10% (could be improved to 8%/12%) could be left only for node/lair awards, not that we're going to really miss this.
1 book: 3-1-0-0 - 4 spells
2 book: 5-2-1-0 - 4 more spells, can trade uncommon
3 book: 6-3-2-1 - 4 more spells, can trade rare
4 book: 7-4-3-2 - 4 more spells, can trade v rare
5 book: 8-6-5-3 - 6 more spells (cost reductions don't begin here, add 2 more spells instead)
6 book: 9-8-6-5 - 6 more spells
7 book: 10-9-8-7 - 6 more spells
8 book (max option): 10-10-10-10 - 6 more spells
Additional books (via awards) - 8%cost/12%research bonus
This should finally [hopefully] balance books vs retorts ... where books are entirely based on initial power, spells.
I don't get it. All this achieves is having more picks left for retorts, I though we wanted the opposite?
(even with 11 picks, it leaves 3 picks for retorts while the current version leaves 2)
Posts: 6,457
Threads: 134
Joined: Aug 2004
(March 12th, 2017, 09:26)Seravy Wrote: Quote:This is incorrect. The difference between spending 400 RP (Giant Spiders) or 600 RP (2x Common + Giant Spiders) is negligible.
I didn't mean that. I meant, with 8 slots and only 1 common to research, you have a fairly high chance for the spell to appear. (1 common and 0-3 arcane spells vs all other slots on uncommons). Each common unresearched puts this 1 spell further away, and we are not talking about 2x spells I believe, more along the lines of 3-5.
Also, the spells mentioned have a 200 RP cost, for early availability. Adding 2-3 commons (possibly including 200 cost spells - there are expensive commons!) is a huge deal, it more than doubles the total cost. I can assure you that you will be able to pick a number of books that allows uncommon spells to appear early on. In fact, this system could well lead to them appearing sooner. Would that satisfy you?
Quote:Quote:8+ books are weak setups, given how strong the retorts are. This is your personal preference, this is not optimal strategy. Again, you're not a good example to base this on.
Then the retorts are the problem which is what I was saying.
No, you were saying that going with a high book setup was worth it for the guaranteed spells. It most certainly is not! This is not because of retorts, it's because you get hardly any advantage from adding extra books. And the advantages you do get are mostly random. This is not satisfactory.
Quote:Quote:Then let's say 6 books earns you a starting common spell and 2 guaranteed common spells. It picks a summon spell or a buff spell as its starting spell and can get to work right away.
The AI isn't going to always have 6 books in a single realm. Go with the assumption the AI always has 4 books of its primary and secondary realms, the rest are random.
I just ran 3 test games, books for AI wizards were distributed as follows:
So 6 is a good baseline to use. Even in the single case of 4+5, he should have enough combined to get him started.
Quote:We seem to agree the AI needs 3 starting spells. But it needs that many from 4 books not 6. Which is exactly the current system, no room for a single point of modification here.
No, I disagree. The AI needs 3 guaranteed spells. It doesn't need transportation right away (unless you insist on putting them on tiny islands). It also doesn't need a combat spell right away.
Quote:Also "guaranteed" isn't good enough. The AI is spending zero on research before turn 30 (I can modify that but probably shouldn't - it would impact their performance massively) and won't select those spells first unless we add ALL of them as exceptions in the research procedure.
Guaranteed absolutely is good enough. I just had a look at the difficulty settings, and an Impossible AI wizard with 10 books gets 37 power from turn 1. He also gets a 60% discount on casting cost and 22 gold income. He can cast 1 Sprite per turn if he wants to! He can easily divert 10% of that into research from turn 1, and 20% from turn 2. He'd probably be better off for it, he doesn't know what to do with all those starting units or buffs. Also, keep in mind that the human player is no longer able to pull off sneaky rushes on his fortress as we're closing that loophole. You're doing him a favour by not asking him to put all his resources in useless assets so early.
I still maintain that I can easily come up with a distribution algorithm that satisfies your concerns. The AI will function well (probably better, if it's bleeding resources at present), and the human player will be presented with a wider variety of choices than now. He'll also have a bit more control over his destiny, and research will play a much more important role in his game than now as he cannot rely on having a starting toolkit handed to him. So, are you interested?
|
