Endgame battles

[Anskiy]

A complex system that weights the importance of freeing the tile, the AI's own risk taken based on enemy combat spells, the potential MP wasted on each side and who that benefits more based on income and available mana crystals, and an analysis of expected human player MP spending behaviour for that battle would be necessary to be able to tell apart battles the AI can safely skip from those that they should be doing. This goes way beyond what's reasonably possible to do.

I don't think it has to be that complicated, you could make the AI have a "charge at important point" state and have it override the "don't fight small stacks" behavior. But I don't know how the AI is coded underneath, so it's upto you and how you implemented it.

[zitro1987]

But one could argue the current status quo of inconsequential battle micromanagement has room for improvement, and is comparably much more time-consuming than other 4Xs

Maybe the solution is related to ideas already shared, but we haven't thought out the solutions well yet ... doesn't mean we should shut them down completely.

Reevaluating giving the choice of instant battle results. It has been argued that giving the option has extreme abuse potential and I can see why. So let's try something different:
*Seravy already taught Master of Magic to understand battle odds through one of the score modifying options that lets you know your odds against lairs. When you have no chance of winning ('predictor that you army will be wiped out by the monsters') and when you are expected to win with little to no losses.
*Taking that into consideration, for player vs AI battles, based on army stats, we could implement the same thing. If a battle encounter EXTREMELY favors the winning or losing side, player is given the option of seeing instant results or playing the battle manually.
*It would be a huge timesaver to get instant results for these and would be very appreciated if we get that kind of option to save time, it would feel like 'you will easily win this battle, but want to control it anyways?' or 'you will lose badly, do you want to see your army get wiped out?'. Sometimes I'd like to see, but mostly controlling the battle pads game length with little impact.
*Cheating is drastically reduced because when you have clearly losing odds, instant results will not look great for the player. And when you have clearly winning odds with minimal or no losses, instant results will look very favorable just like with a manual battle
*Therefore, proposing the option only for these extreme ends will limit a lot of these issues. We will unfortunately continue being forced to have some low-stakes manual battles like 2 armies of halberdiers battling it out, but I can't think of a better solution.

Then the proposal of modifying AI to lower incentive of attacking 1-unit stacks ... that one is harder because we have to think about what that 1-unit stack is doing - blocking pathfinding, scouting in opponent's territory, etc.

[Suppanut]

I think there should be option for overland map battle withdrawal too. Basically if odd is too different, you have option to fight manually or attempt to withdraw before fight with same rate of unit lost as retreat manually in battle mode. Even with result would be the same as first turn battle retreat, it still much more useful than auto-resolve battle which clearly lose (as ai would only attack in battle that they would win in auto-resolve as Seravy said earlier) unless autoresolve could create considerable damage for ai even if they win.

[zitro1987]

I think there should be option for overland map battle withdrawal too. Basically if odd is too different, you have option to fight manually or attempt to withdraw before fight with same rate of unit lost as retreat manually in battle mode. Even with result would be the same as first turn battle retreat, it still much more useful than auto-resolve battle which clearly lose (as ai would only attack in battle that they would win in auto-resolve as Seravy said earlier) unless autoresolve could create considerable damage for ai even if they win.

I think this is a poor substitute, you can just hit manual and hit the flee button. Letting players autoresolve only the extremely lopsided battles (where losses to one side are very minimal) is a huge timesaver as there is more to it (more than 1 battle tur, unit positioning, spells, etc)

[Seravy]

*Seravy already taught Master of Magic to understand battle odds through one of the score modifying options that lets you know your odds against lairs. When you have no chance of winning ('predictor that you army will be wiped out by the monsters') and when you are expected to win with little to no losses.
*Taking that into consideration, for player vs AI battles, based on army stats, we could implement the same thing. If a battle encounter EXTREMELY favors the winning or losing side, player is given the option of seeing instant results or playing the battle manually.
*It would be a huge timesaver to get instant results for these and would be very appreciated if we get that kind of option to save time, it would feel like 'you will easily win this battle, but want to control it anyways?' or 'you will lose badly, do you want to see your army get wiped out?'. Sometimes I'd like to see, but mostly controlling the battle pads game length with little impact.
*Cheating is drastically reduced because when you have clearly losing odds, instant results will not look great for the player. And when you have clearly winning odds with minimal or no losses, instant results will look very favorable just like with a manual battle
*Therefore, proposing the option only for these extreme ends will limit a lot of these issues. We will unfortunately continue being forced to have some low-stakes manual battles like 2 armies of halberdiers battling it out, but I can't think of a better solution.

Unfortunately, using the expected results to determine the choice isn't helping here. The player WILL want to only use this feature when they are expected to win using it in the first place.

Example scenarios :

1.
Strategic combat : Players army destroys 1 enemy swordsmen, minimal damage taken.
Real combat : AI casts Flame Strike on turn one, wiping out half the player's army.

2.
Strategic : Player's 1 Night Stalker wipes out the enemy army of low resistance zombies as Death Immunity isn't applied here and instant death damage is very high.
Real : Player loses the battle and cannot even damage the zombies.

In both of these cases it's a huge advantage to swap to strategic combat EXACTLY BECAUSE it allows the player to get an overwhelming victory they otherwise cannot get.

We'd need to measure the outcome of the real combat for your suggestion to work but that's impossible. If we could have a perfect simulation of real combat, the problem wouldn't exist in the first place.

I think there should be option for overland map battle withdrawal too. Basically if odd is too different, you have option to fight manually or attempt to withdraw before fight with same rate of unit lost as retreat manually in battle mode. Even with result would be the same as first turn battle retreat, it still much more useful than auto-resolve battle which clearly lose (as ai would only attack in battle that they would win in auto-resolve as Seravy said earlier) unless autoresolve could create considerable damage for ai even if they win.

Wouldn't this be slower and more confusing than simply pressing the Flee button on the first turn? You are the defender so the AI isn't even taking up time by moving their units or anything. You can do it immediately.

[zitro1987]

The perfect simulation of battle was suggested a few times but was rejected : having ‘auto’ in battle screen play with casting spells and an AI that matches opponent (not intentionally playing dumb as dirt). Rejection reasons (if i remember) included
*the cheating rationale again because auto combat decisions might be better than player, so gotta keep it dumb (i never liked this argument, there is so much wrong with it and the current auto intentionally playing dumb is so frustrating!)
*That it didn’t feel right because you are the wizard so you cant be controlled (realism isn’t a Master of Magic trait)
*fear of wasting mana (this one is logical, but can be resolved with a prompt option or a new auto button)

If everything is rejected (especially with cheating rationale) it gives us the impression of being forced to play in a way specific to developer(s) that forces pkaying through a ton of inconsequential battles. It is not a good feeling

[drake178]

Rejection reasons (if i remember) included...

You missed the single most important one: it's computationally expensive. Essentially, it's mutually exclusive with the design idea of the AI pre-playing all of its battles to gauge their outcome.

[zitro1987]

I meant within the battle screen not strategic (unless im misunderstanding). If hitting auto means seeing normal intelligence managing my battle (not intentionally dumb) i would feel better hitting auto. A good comparison is Hitting auto in age of wonders series seeing the battle unfold.

The issue is that we dont have the modern 4x battle conveniences yet with a very high frequency of battles

[Seravy]

Making the Auto button cast spells is trivial since it already does so if a hero unit is moving.



The problems with that were exactly what you listed :

-It would use the same logic heroes use which is dumber than the AI exactly because we don't want to give spoilers to a player about what's the winning strategy. Spoilers are always bad if they are too easy to access, especially as part of a feature that seems to be just for convenience when in reality it does a lot more than that. As far as I remember this is currently set to "use random spell on best target".

-Your military advisor can't cast spells. (Flavor reason, can be solved by changing the existing text that refers to the advisor leading the battle)

-We'd have THREE "yes/no" windows pop up every time you press the button which is a bit too much for an otherwise convenience feature. (confirmation, one turn or entire battle, use spells or not)



It might be worth adding it despite all of these problems but I'd like to hear much more opinions about that.



But let's try to look at this in deeper detail.

Spoilers don't matter because the AI will show the same strategies to the player.

(but that information isn't relevant for you for the current battle so you don't pay too much attention to it)
(also spoilers only matter on the first few playthroughs while you're still learning the game, in which case you actually won't be able to use those AI strategies and by the time you do play the right spellbooks, you are a good enough player to figure it out yourself. Yet you could use spoiled strategies if it applied to your own side.)

I did delete games before because they had poor AI controlling my party members (RPG)

(but in another game because the AI was playing well and I was unnecessary to win the battles. All I had to do was  walk towards the next monster and watch it die)



So overall there is a fine balance here, as too little or too much convenience, too good or too weak AI plays can both result in players disliking the game.

The consequences aren't equal though.

If the convenience feature is too good and the player feels unnecessary (the game is playing itself), they will stop playing the game and uninstall. 

If the convenience feature is too weak then, as long as it's not mandatory (in the above example RPG it was mandatory, you always had a party of 3 with two AI controlled player characters, one of which was a healer who was programmed to go and hit monsters with melee attacks for 1 damage) the player can simply avoid using it and keep playing.



So from a designer perspective, weaker AI autoplay is healthier to the game, if the perfect balance is unreachable between the two.
Fortunately, combat is only half of the game here, there is still economy and overland moves. It is the more fun half though so I do see a risk of lower enjoyment if the AI plays the battles.

I don't know about Age of Wonders but assuming it was one of those generic 4X games where unit strength determines the outcome and choice of moves and spells are secondary, that unfortunately doesn't apply to spells in CoM.
Basically, if you have 3x Great Drake vs an enemy of 7x Swordsmen, no matter how poorly or well you position your units, you win either way in most other "modern 4x". So AI strength in the autoplay feature has little relevance, if it plays poorly or well don't make much difference unless the armies are close to equal but then the player won't activate it usually. If the armies are uneven the expected side will always win if you use autoplay. The autoplay won't spoil you much by moving the drake to the right position, or casting some decent spells.
Not so much in CoM.
Add spellcasting and your 3 Great Drakes might lose to the 7 Swordsmen. The enemy only has to cast Entangle, Call Lightning, and start to move away from your now too slow drakes. But this of course also happens if you are on the side of the swordsmen.
In don't think it's good if autoplay can often have the opposite of the expected result - winning battles where you activate it to lose quicker, and losing battles where you activate it to win quicker.

Anyway, Auto spellcasting was a feature I seriously considered at that time and I'd say I was 45% for and 55% against it. It definitely wasn't a trivial decision at all. We might want to discuss it further and possible add the feature anyway.

[Seravy]

A new idea to fix the "three yes/no windows" issue would be to use a menu instead.

1. Entire battle, use spells
2. Entire battle, no spells
3. One turn, use spells
4. One turn, no spells
5. Cancel

So we can ignore that being an issue.