Sprites are still overpowered

[Nelphine]

What about giving nodes innate spellcasting, like they had their own AI running the place? Wouldn't help lairs, but would make nodes far harder to conquer?

[GermanJoey]

Passing a billion turns for every strong lair doesn't sound like a lot of fun either, and it's not great when unfun tactics are the most effective.  ...  Is it possible to have short-term flying effects, so that a unit starts combat with flying (or has a one-shot-per-battle ability that gives it flying) but the effect wears off after some number of turns, after which it can be attacked normally?

If you turn combat animations off, its not too bad, except for Great Wyrms and Unicorns. (which still have their merging/teleportation animations shown)

[Catwalk]

Thank you for a well thought out and constructive response Seravy. I agree with most of it.

Incidentally, early War Bears are a gamble because you lose them so easily. Sprites are a perfectly calculated gambit that give you guaranteed returns in every single game. I've played 6 games now and it hasn't failed a single time. This is not a rush strategy, this is just good strategy.

I think it's the opposite. War Bears give a consistent, predictable result, and losses are usually within the expected range. Sprites on the other hand are entirely random - the presence of good targets, the treasure obtained, and losses too - if the stack of sprites is attacked by an enemy stack, the stack dies - I had a fair share of losing stacks of 2-6 sprites to an enemy ship, or water walking hero. And if any enemy learns firestorm, they wipe a stack of 9 sprites in a single turn without needing to attack - I have had this happen in ~1405 in my first test game.

There is certainly a luck factor when using Sprites. However, this luck factor determines how awesome your start is, not whether or not it's awesome. The save that you tested above was one of the most succesful ones (huge target pool), I've also had some less succesful ones where my start is still really good and gives me a considerable advantage.

Personally, I think it's a problem that rush strategies are so overpowered. You should not be able to win Impossible this easily. Also, I actually don't think they're the most interesting to play.

Good, this is exactly what I think, too. So we have a common goal here.

Precisely, I rather like long games where I feel it's a struggle and the outcome hangs in the balance until past the mid-game. That's the balance I want the game to provide me with when I play optimally on Impossible. It should certainly be viable to be aggressive early on (especially with nodes and neutrals), but it should give you modest rewards that don't overpower everything else.

And with all due respect, I don't have the impression that you're a strong player.

The game changes on almost a daily basis so...yes, that's probably true. As I can only fix problems I'm aware of, it means the strategies I use to win get less effective over time, as the AI learns to counter them, or the balance is improved. Meanwhile strategies I'm not aware of might stay unchanged.

I'm glad you acknowledge this, and the game does indeed change regularly. I also just realized that the new Earth Lore effects are a big part of why this works so well, it was a bit more uncertain back when lair contents weren't shown. Still doable, but slower and a lot more luck dependent. Note that it's excellent that Earth Lore works this way (big thumbs up on your work on this), we just need to balance it properly.

They're not even all that skill dependent, you just need to cast Earth Lore and look at where to go. Have you tried this strategy in depth?

Well, I'm in the process of playing one game when I'm using sprites but I can't keep up with your speed. For me a game takes several days to complete, while you play 2 hours the drop it and start another. Earth Lore wasn't doing what it does now 2 weeks ago. Lairs used to have much more stronger monsters about 2 months ago. So this whole problem seems to be brand new - I did play sprites many times in the past, and while they didn't suck, they weren't enough for an easy win on Extreme.

...all this war bears testing got me sidetracked, I don't even remember where I was in the sprites game anymore. I better load it and check... Oh boy. Compared to that war bears game I'm doing horrible...1405 december, no wizards banished, and while I do have 2 nodes, I have less power overall even with those -  only 45. Fewer cities, less gold income, at war with 3 other wizards...meh.
So as far as the games I played is concerned, it's 1-0 for bears, I might as well drop this game with sprites, they didn't even get close. (Not a surprise though, this game did not have spellweaver and alchemy, the bears one did. Again we are back to retorts interfering too much.)
The previous games where I used the no longer existing retort combination do not count - we fixed it.

Yeah, we definitely need as similar conditions as possible in order to do a fair test.

I'm not sure you understand how it works, with your comments above about nodes and lairs not being important. They're a huuuuge source of early resources, and that snowballs your development into a dominant position.

This is not my experience but I haven't done enough games to judge it - and most of those games were interfered with by retorts.

I hope you'll take my word for it, I'm not exaggerating when I'm saying that it's been enough to put me in a winning position for 6 test games straight.

An idea : we want to test sprites, so let's eliminate every other factor. I'm going to use RVL and PWR on turn 1, summon 9 sprites, and give them all 15 movement then save. I will then send them into every lair and node I can reach with that many moves, clear the node, if successful, make a note of the treasure, reload and do the next. Do this, like, 10 times. Then we'll have an approximate idea of how much treasure this means in the early game on average. If the number is above X gold, then I'll admit sprites are overpowered - but we'll have to agree on the value of X.
...this will be ultra boring to do, though. I know better! Let's automate the process. We know which monsters can be killed by sprites. So I can write a tool that reads the map, counts the treasure, and outputs it into a file. It'll probably be done in less than 2 hours and saves another 100 hours of testing. Meanwhile suggest a value for X.

This is a great idea, and I agree that it can give us mass quantities of data. There are a few problems with the approach, though. A big part of the appeal of Sprites is that you can get whittle down defenses over several turns. Even if it takes 4-5 turns to get to the orange, keeping at it with 2 or 3 Sprites is well worth it. Waiting for a huge stack is very far from optimal. Would you be able to get me a good data set of the map read? If so, I suggest that I write a tool in Excel which analyzes how good your results will be. For each monster lair I can approximate a good formula saying how fast it can be cleared with X Sprites, and that can be compared with the distance for an ultimate overall treasure result. That way we get a balanced output which will closely reflect actual game results. I actually did run 10 test games and manually scanned lair contents just to see how many I would be able to do. That's close to what you're suggesting, only with a lot less data (I didn't see what treasure I got) and slower. I was very pleased with the results of my analysis though, and that's what led to me exploring the strategy further.

I disagree completely, you can fix all those concerns easily. Since you say that you often ride that train of thought, are you interested in changing the system? There should be a tangible reward to having a high number of books in a realm,

How exactly am I supposed to easily fix the AI not getting a spell it needs? Sorry but I don't see this happening and I have my hands full of the million other problems we face already. (Sprites, starting conditions, retort stacking, and the bugs on my list that need fixing for starters)

If I hand you the solution on a silver plate, will you promise to look carefully at it with an open mind? I can 100% address those concerns above.

It's a big part of what makes high book strategies useless.

I believe the best answer to this is to get rid of all the unintended rush strategies - we seem to agree they are a problem, and based on our experience so far, they are rely on stacking retorts for an early advantage somehow, which results in low book counts. I don't think low book count late game strategies are going to be a major problem - the lack of rare and very rare spells makes these weak.

Keep in mind that being active early on should not be mistaken for rush strategies. Going for neutral cities, lairs and enemy cities early on is not necessary a rush strategy. It's only a rush strategy if you go all out and secure the game with it. Modest campaigns early on should be encouraged, we don't want it to turn into a builder game where stuff only happens later. I think this is also what you want.

Back to the sprite testing tool. I believe this list of creatures should be included as "sprites compatible" :
Fire Elemental
Hell Hounds
Chaos Spawn
Skeletons
Werewolves
Zombies
Unicorns
Guardian Spirit
War Bears
Cockatrices (these are hard in large numbers but doable with web)
Earth Elemental
Great Wyrm
Phantom Beast
Phantom Warrior
Nagas
Sprites (need resist elements and probably will have losses anyway but can do - I don't think I ever did, though. I rather throw away my bowmen for this. Do we want to include it anyway?)

I'll check this list more closely tomorrow. There is both A and B compatible, where A compatible is always 100% safe and B compatible is safe with a large stack of Sprites and using 2x Web. For example, Cockatrices are easily B compatible and you can take up to 4-5 of them easily with a stack of 9 and Web. EDIT: And I see that you already wrote that in the list

[Catwalk]

Passing a billion turns for every strong lair doesn't sound like a lot of fun either, and it's not great when unfun tactics are the most effective.  ...  Is it possible to have short-term flying effects, so that a unit starts combat with flying (or has a one-shot-per-battle ability that gives it flying) but the effect wears off after some number of turns, after which it can be attacked normally?

If you turn combat animations off, its not too bad, except for Great Wyrms and Unicorns. (which still have their merging/teleportation animations shown)

Precisely, I used to hate this (and wanted to eliminate it), but that's actually a very doable and balanced strategy now. Seravy did a good job with that, especially with setting the turn count to 25 instead of 50. It should definitely stay in the game, it balances a crude fleeing mechanism which will never be sufficient on its own.

[RefSteel]

Cool; I'm glad waiting out the turn timer is rarely tedious here! (Sort of like in Orion where it's only really a problem with Repulsor Beam animations....)

Note I wasn't suggesting that flying be replaced by temporary-flying; I just thought it might be a neat lesser ability/spell and perhaps a way to reduce the potency of Sprites. Even so, coding and balancing such a thing probably makes it not worth the effort, especially as it would be difficult or impossible to teach the AI to use it well, and I can see why it's better to let Sprites in particular just stay in the air all the time.

[Catwalk]

For Sprites in particular, it's far better to remove Flying altogether. There's no other reason than flavour to let them keep it, and huge balance issues associated with them having it. No other common unit has free access to so many lairs from turn 1. They can take out the dreaded Great Wyrm with ease, a very rare creature that just about every other unit in the game has huge problems with.

[Catwalk]

This is my list of Sprite approved targets. () means they're not entirely safe, but doable. Keep in mind that Web can work wonders against a small number of tough opponents. For each, I will analyze how many Sprites it will take to take them out safely. This may get complicated, but I'm confident I can come up with something good.


Phantom Warriors
Nagas
Phantom Beast
(Storm Giant) 4 ammo
(Djinn) 4 ammo

War Bears
(Sprites)
Great Lizard
(Giant Spiders)
(Cockatrices)
(Gorgons)
Earth Elemental
Great Wyrm
(Behemoth)

Hell Hounds
Fire Elemental
(Gargoyles)
(Chimera)
(Fire Giant)
(Efreet)
(Doom Bat)
(Chaos Spawn)
(Great Drake)

Skeletons
(Ghouls)
Zombies
Werewolves
(Shadow Demons)
(Wraiths)
(Demon)
(Death Knights)

Guardian Spirit
Unicorns
(Angel)

[Seravy]

If I hand you the solution on a silver plate, will you promise to look carefully at it with an open mind? I can 100% address those concerns above.

I will definitely look at it but can't promise I'll like the solution, if one even exists.

Phantom Warriors
Nagas
Phantom Beast
(Storm Giant) 4 ammo
(Djinn) 4 ammo

War Bears
(Sprites)
Great Lizard
(Giant Spiders)
(Cockatrices)
(Gorgons)
Earth Elemental
Great Wyrm
(Behemoth)

Hell Hounds
Fire Elemental
(Gargoyles)
(Chimera)
(Fire Giant)
(Efreet)
(Doom Bat)
(Chaos Spawn)
(Great Drake)

Skeletons
(Ghouls)
Zombies
Werewolves
(Shadow Demons)
(Wraiths)
(Demon)
(Death Knights)

Guardian Spirit
Unicorns
(Angel)

I see a lot of things on the list I consider unsafe - you can at best do it if there is no more than 1 (or in some cases 2) of those monsters and you have 9 sprites. If we include these we should limit the quantity of the monsters - my current list come with unlimited quantities.

[Catwalk]

That's all I'm asking, that you keep an open mind. You state that you're supportive on principle, but that there are AI problems with implementing a solution. I will make a new thread to discuss it in depth.

As for the Sprite list, anything that can potentially harm Sprites is put in parenthesis. I'll make sure to handle those cases accurately, and I will explain my method. Data analysis is what I do for a living, I'm confident that I can quickly develop an analysis tool that gives results that are very close to reality. If there's 1 Death Knight, for example, it's perfectly realistic to take it out with only 1 Sprite lost. If there are 2, it's largely impossible. In my data entry I'll list a max amount for unsafe creatures, as well as a minimum number of Sprites required. It will also estimate casualties.

Note that my analysis assumes generous Web usage. For non-Nature nodes I'll make separate outputs depending on whether or not you pick Astrologer.

[Seravy]

Meanwhile I was able to come up with a new sprites unit I like, in the case we determine change is necessary, but I'm worried. It'll probably be still overpowered albeit in a completely new way, and the implementation might prove impossible as well.

The idea is to implement a new attack flag which does "when the attack is performed, apply the effect of Fairy Dust to the opponent", the same way Immolation applies a Fireball spell effect. This would effectively turn Sprites into units with ranged immolation, something we don't have right now. I believe there is an unused attack flag.

The potential problem is finding space to add the effect in the applyattack procedure, and of course it might just be overpwered as a whole - while mostly ineffective on nodes and lairs, it's an attack that wipes out a normal unit in a single shot afterall.