Endgame battles

[zitro1987]

Exciting to read that the 3 initial topics seem to be resolvable or not too big of a deal (I love the 5 options in one idea for example).

So the debate is now about player motivation to keep playing.
*too much convenience could translate to manual battles being pointless if optimized perfectly, player may feel like the point of playing the game is lost and abandon the game . Add pretty good auto-building (grand vizier) and game mostly plays itself. (that is the concern, right?)
*too little convenience (I'd say how it is now) - automatic battles are painful to watch, causes frustration, player will then resort to manually doing all battles, player may have time limitations (family, studies), battles get too frequent and repetitive, player is only invested in a small portion of them, so player abandons the game

In other words, the debate is about how to minimize motivation loss to keep playing. How smart should things be.

Unit positioning and fighting: I'll say be as intelligent as neutral units on fair
Spell casting: like normal difficulty AI wizards or neutral caster units on fair (basically, reasonable targeting such as confusion on great lizards, lightning bolt on heroes, reasonable buff selections)

This is not optimal, so a player will clearly choose to manually fight decisive battles and maybe automate some lairs, easy city raids, and the frequent land battles of attrition, but frustration watching the advisor play a battle wouldn't be too high as it is now.

[Anskiy]

I think Seravy is overthinking the issue. Spoiling strategies doesn't really matter by year 1420 or whatever, if the player has survived by then they likely know a thing or two about the AI. Furthermore, this is a strategy game that's intended to be played multiple times, I don't really see the harm in spoiling new players. I would be completely fine with a menu that gave the option for auto combat with spellcasting, but I would prefer strategic combat to be toggleable and more accurate. Not accounting for Flame Strike or Death Immunity seems unacceptable to me in a good strategic combat system.

[Seravy]

*too much convenience could translate to manual battles being pointless if optimized perfectly, player may feel like the point of playing the game is lost and abandon the game . Add pretty good auto-building (grand vizier) and game mostly plays itself. (that is the concern, right?)

That and it's also messing with the learning curve.
Unlike the game I abandoned due to the game playing itself, here the player isn't even involved in configuring the AI.
So it's not like you get bored because you already know how to win and set up the perfect AI that can beat even boss battles without you touching the controls.
Which means in addition, there will be a point where the difference between AI and manual play is huge in favor of the AI which means it can result in any of these :
-Laziness : always use auto battle without ever trying to learn how the game works - it wins anyway so why bother?
Obviously, this too will result in boredom and a flawed percetion that the game is much simpler and less interesting than it is.
-Overrating own skill : Use auto a lot in the first few games then try to play manually and lose hard on the same level of difficulty. Blame it on the game and rage quit because you assume you are still playing at the same level as you did when you used auto so the game's difficulty levels are wrong/inconsistent/too hard.
-A combination of both - if I can't win manually, instead of learning the game, I can press auto and win anyway.

This game definitely has a much longer and steeper learning curve than usual and being able to bypass most of that by an auto button is probably not good.

*too little convenience (I'd say how it is now) - automatic battles are painful to watch, causes frustration, player will then resort to manually doing all battles, player may have time limitations (family, studies), battles get too frequent and repetitive, player is only invested in a small portion of them, so player abandons the game

This one is correct although I still think a smart player would get around most of this by simply having strong enough armies to not get attacked often. Or just press flee in battles they don't want to play manually.
However the 9 unit per tile limit interfering with how you can effectively keep your stacks a "safe" size.
Like when your city produces a unit and has 9 in garrison, the 1 new unit will be attacked unless you can reinforce it with more units but that either leaves the city poorly defended or you need to micromanage by gathering units from 3-4 nearby cities.
In which case you are still losing time on trying to keep fresh stacks a large enough size.

My personal solution to this is to generally not bother and stop producing normal units in the endgame where this starts to be an issue. The forces I use to conquer and hold cities are mostly summoned creatures anyway. So I think this mainly affects only specific strategies that rely on normal units in the late game. Life magic or Chaos with Doom Mastery are the most likely culprits.

The current solution does target spells correctly but doesn't pick the highest rated spell if I remember correctly.
For example, it'll have equal chance to cast Prayer or Fire Bolt on a unit of bowmen or Raise Dead on your own fallen Spearmen, assuming these are the best available targets for each spell (mainly due to having nothing else dead).

...actually not entirely. It still has a higher chance to use the better spell but the "randomness" is higher.
Regular AI adds a D20 to each priority value and picks the highest.
Auto also adds a D100 to that result.

If that's too much, we can change it to a different number, I don't think any less than an extra 30 would be good but 100 might be excessive.

[Seravy]

Spoiling strategies doesn't really matter by year 1420 or whatever, if the player has survived by then they likely know a thing or two about the AI.

I'm translating this to myself as "make the auto play smarter as turn count rises". That actually does make sense because you generally want to use it in the endgame either way, and by then the deciding factor will be your economy and available spells, not so much your actual combat choices.

I mean, if I have Great Drake and Apocalypse and the enemy has only Angel and Exaltation, I'm winning and it doesn't really matter if the AI plays smart or dumb unless it's actively trying to sabotage my game by not attacking or casting any useful spells.

So maybe in the above mentioned formula, auto should add D100-(turn count/3) to the result of the roll instead of simply D100. So basically by turn 200 the AI is only adding a D30 which makes is semi-smart and by turn 300 it plays at the same efficiency it would use for itself. Or maybe make that (turns/2.5) instead?

The good part in this is, this solution can't make the game "play itself" nor interfere with player learning because you'd need to survive the early game with manual play to use it.

I think this might be our best solution?

[Anskiy]

Given everything you've said, that seems about right.

[zitro1987]

That’s not a bad idea (i prefer turns/2.5) and makes logical sense in game progression - advisor becomes smarter by experience. how can it be explained in help text? “Advisor will have increased wisdom over the passage of time, improving combat decisions” with some better written or more flavorful text

[massone]

Spoiling strategies doesn't really matter by year 1420 or whatever, if the player has survived by then they likely know a thing or two about the AI.

I'm translating this to myself as "make the auto play smarter as turn count rises". That actually does make sense because you generally want to use it in the endgame either way, and by then the deciding factor will be your economy and available spells, not so much your actual combat choices.

I mean, if I have Great Drake and Apocalypse and the enemy has only Angel and Exaltation, I'm winning and it doesn't really matter if the AI plays smart or dumb unless it's actively trying to sabotage my game by not attacking or casting any useful spells.

So maybe in the above mentioned formula, auto should add D100-(turn count/3) to the result of the roll instead of simply D100. So basically by turn 200 the AI is only adding a D30 which makes is semi-smart and by turn 300 it plays at the same efficiency it would use for itself. Or maybe make that (turns/2.5) instead?

The good part in this is, this solution can't make the game "play itself" nor interfere with player learning because you'd need to survive the early game with manual play to use it.

I think this might be our best solution?

That's a very cool idea. You might also consider scaling it to number of battles (or total battle EXP gained? To avoid counting a lot of 1 unit battles) the player has won manually instead of turn count, to be sure that the auto doesn't play better than the player themselves.

[Slingers]

That's a very cool idea. You might also consider scaling it to number of battles (or total battle EXP gained? To avoid counting a lot of 1 unit battles) the player has won manually instead of turn count, to be sure that the auto doesn't play better than the player themselves.

Only total battle EXP is nowhere to be found in the stats. So it could not be checked weather the autobattle has already reached god level. Maybe the "Auto" button could turn green.