Co-op rules wanted. Inquire within.

I haven't played enough to even think about testing something like this, but I was thinking about this too. First, I think you have to assume that the players will make the most advantagous movement for the villain, otherwise everything will go down the drain. That being said, having a truly random deck of actions, can leave very little room for the heroes to have a stragey beyond "there is still card X in the deck."

You could also create a very detailed tree that deals with different decision. For example, "If there are two hero targets withing X spaces of the villain, do Y." The biggest problem with this is that it is way to easy for the heroes to predict what to do.

The medium I was looking at, once again pure theory here, is to have a deck of cards that have conditions on them. Some specific condition, and some generic condition. This can give a litle more randomness, while still allowing for some strategy for the heroes, once they know the attack patterns. For example, a card could say, "If there are two or more heroes within X spaces, do Y. Otherwise, move towards the hero with the lowest hp and do Z."

Obviously, this would need to have a lot of testing, but I think I would begin with a system like that.

I was thinking of starting in the other direction, hence the cards. When you have something that is very predictable, you are running into some unfun issues, especially, when the heroes have no issues of randomness (they can use any card at any time). Let's say you did have a steady stream of Thugs, I think you would run into 3 major issues: 1) Since you know what they will do, there will almost always be an optimal move. 2) Since the scenario is always the same, that optimal move will always be the optimal move, removing any sort of replayability. 3) Since you know exactly how everything will work before hand, you can almost plan out the entire game before making your first move. The only possible issue is if you rolled to many misses, but even then, it wouldn't be hard to compensate.

To me at least, the question is more about how to get randomness into the game. You could simply have a chart and roll a die for the villains action, or have a deck of cards, but I don't think complete randomness would work in Tectics case. In SotM, you have almost complete randomness, and you will have some pointless turns (e.g. Terramforming on the first turn). However, to a certain extant, this is balanced by the randomness of the heroes. The villain doesn't have to play perfectly every time, since the heroes will always have different cards at there disposasl. In Tactics, there hero can always play the best card in every situation, and this is what I think is the inherent problem in Tactics Co-op. Pure predicatability will not be fun, and will turn the game into a puzzle. Pure randomness needs to be able to stand up against the non-randomness of the heroes.

I've been working on a solo/co-op scenario that involved hoards of mobile minions rushing to the objective markers while the players fight them off, but the issues i have been having is thst in modt casss there is a blantent optimal configuration for powers, usually the best multi target options that each hero has. the gane is balanced around dealing with fewer, more significant targets rather than hoards it seams. the problem of course is constructing a effective but streamlined boss ai.

Then I guess the question is where should we start? trying to figure out meta-rules and ideas, and create multiple scenarios from that? Or start with a scenario, fix that one up, and then apply what we learned to other scenarios?

My initial work on AI baddies:

1.  Ditch LoS.  Writing AI Scripts isn't going to go well if the heroes can bait the villain AI into easy traps.  LoS needs written into AI or avoided.

2.  Little heavily specialized minions (speedy devils, tanky tanks, ranged nukers, buffers, ETC.) acting on very simple scripts.

3.  Bigger baddies that act on more complex scripts and cause major havoc for heroes.

I've been able to keep them functioning pretty decently in some scenarios:

1.  Bad guys taking objectives:  an easy progression from entrance to point 1, point 1 to point 2, etc.

2.  defending against the same kind of progression by player characters.

3.  hostage rescue situations (the players do the rescuing)

Further thoughts:

I think the flat maps will allow for some really cool infiltration style scenarios vs. AI.

The bigger baddies could have different styles:

This unit provides cover for allies within radius 2.

This unit creates hazard spaces for enemies in a radius of 2.  [appropriate attack]

GO:  This unit moves 3 toward the next objective.

This unit counts as an objective for minions.

Or:

GO:  This unit teleports to the closest point in line of sight to attain range 5 line of sight on a hero target.  Then makes this attack  (good ranged attack designed around hitting at range 5, like reach 2, 1&2 automiss.)

If no hexes qualify this unit moves 2 towards the next objective.

The first one would gather a crowd of minions around it and move toward the objective, granting cover and attacking heroes that invade its space, while the second is going to attack heroes whenever it can, even moving across the map.

The second one takes a good bit of work, figuring out where line of sight options exist.  Limiting it to range of 5 helps, but maybe 4 hexes away would work better, that would give fewer hexes to check for eligibility.

Stats would need to show the unit name, class (minion or boss) health and defense.  you'd also need to have a set order of activating AI targets so it would be easy to keep track.

One possible method would be a split card, have a set number of minion types with cards for them with basic stats, and then the scenario would need to have the other half of the card that would tell its script for that scenario.

You would need directions that in the case of two equally viable spaces the AI moves to the one closest to the same level it is already on (progression) or that its target is on (attack)

Basic minion scripts could go:

Sprint towards the next objective, then make this attack against the nearest enemy target (Attack!)

or

Sprint towards the next objective, then any unit within 2 hexes regains 1hp.

or

Sprint towards the next objective, any damage that would be dealt to ally targets within radius 2 is dealt to this target instead.

I'm currently busy with GenCon stuff, so I can't contribute much at the moment.

That said, whoever wants to work on this might take a gander at the Heroes Wanted rulebook for inspiration. It's a tactical hex-based game with an automated villain.

Herp derp. If I'd done some looking first, I would have discovered they aren't up yet. Sorry. It just got released, so I imagine some enterprising soul will put them up soon.

I don't have too much time to look into some of the (I assume) fantastic ideas that other poeple have had, but if you want some resources to look at, look into how the Gears of War board game or the D&D games like Legend of Drzzt or Wrath of Ashardalon work.  Heres a couple images:

Also, I noticed that in all of these games there is a bad guy turn for every good guy turn.  That might be one way to help balance out the fact that the villains will not make as effective choices as if they had a player controlling them.  Have a villain turn in between every hero turn regardless of how many of each there are.

So, I've had this idea to make bot scripts for Individual characters in a way that wouldn't be entirely predictable.  The idea is their power cards are divided into three catagories: Paradigm - Cards that define how the bot acts while in play,Support - Power card that typically provide more modifiers and boosts to the character and usually only have minimal effect on the bot's scripts and One-shots, which normally aren't used, but have conditions where the bot will activate them instead of its normal behavior.  Technaically, all oneshots are always in play for the bot, but are typically not used unless their criteria is met.  Otherwise, the character keeps both a paradigm and a support power card in play at all time.  Whenever the character's script tells them to replace either type of card, they shuffle all cards of that type together and draw one at random to replace the one removed.  These scripted characters ignore the limts of how many power cards may be played per turn and if they would replace a card with one they already removed, its treated as though it was never taken out of play.

So, for an example, I wrote up a quick Ra script, since he is fairly straightforward in his gameplay and have cards that easily fit into all three catagories.  Also, there is refrences to Scenario paradigm, which is a scenario specific general script which provides the fundimental actions for the bot to attempt to win the scenario.  I've listed a basic rush the Scenario Marker Scenario paradgim for example as well.

Scenario Paradigm - Rush the Marker

For this Scenario, the Scenario objective is the scenario markers

Movement

If the character starts their turn adjacent to a Scenario marker, they attempt to skip their movement phase

Whenever the character performs a Scenario movement, they move as close as possible to a scenario marker while using a single action.

Ra

General

Ra use either Sprint or Move actions to move, whichever gets him closest to his destination.
Ra uses +1 tokens, Aim tokens or Dodge tokens at the rate of one per roll whenever a valid dice roll occurs.

Paradigm - Flame Spike

Movement

If possible, Ra moves so that he has line of sight to the nearest hero and is within range 3
Otherwise, Ra performs the scenario Movement
If his Movement phase would be skipped, either due to the Scenario Paradigm or having no other valid action, if possible, he instead uses Fire blast on the closest valid defender target.
Otherwise, he takes an Aim action.

Action

Ra uses Fire blast on the closest valid defender target, if Possible.
Otherwise, he performs an additional movement phase, if he would skip this additional movement phase, he also skips his action phase.
If his action phase would be skipped either due to no other valid action or the Scenario Paradigm, he instead Dodges.

Paradigm Shift

If Ra did not use Fire Blast during the previous turn, Change his Paradigm and Support cards at the start of his turn.

Paradigm - Flaming Tornado

Ra places the tornado as close as possible to the hero who is closest to a scenario objective

Movement

If doing so would also bring him closer to a scenario objective, ra moves towards his Flaming Tornado
Otherwise, Ra performs the scenario Movemtn.
If he would skip his movement phase due to either the Scenario paradigm or having no other valid action, he takes an Aim action.

Action

Ra uses Fire blast on the closest valid defender target, if Possible.
Otherwise, he takes an additional movement phase, if he skips this additional movement phase, he also attempts to skip this action phase.
If Ra would skip the action phase either due to a scenario paradigm or having no valid actions, he instead takes an Aim action, unless he has already taken his aim action, in which case, he takes a dodge action instead.

Paradigm shift

If Ra moved away from the Flaming tornado in the previous turn and no hero took damage from the tornado after that turn ended, he changes his Paradigm card at the start of his turn.

Paradigm - Inferno

Movement

If possible, Ra moves such that he can target more than one target with Inferno
Otherwise, if possible, Ra moves within at least 3 range and line of sight of a defender
Otherwise, he performs a scenario movement
If Ra would skip the movment phase either due to the Scenario paradigm or having no valid actions, he instead takes an Aim action.

Actions

If possible, Ra uses Inferno, targeting the most amount of targets as possible, favoring closer targets with a tie.
Otherwise, he takes an additional movement phase, if this movement phase would be skipped, he also attempts to skip this action phase
If Ra would skip the action phase either due to a scenario paradigm or having no valid actions, he takes a dodge action.

Paradigm Shift

If Ra has not attacked more than one foe since the start of his previous turn, he changes Paradigm.

One-shot - Drawn to the flame

If at the start of the turn, Ra can move such that he can affect at least 3 targets with the push, he moves and uses this action instead of his normal paradigm

Support - Living Pyre

If he is damaged by a ranged attack after his previous turn, he changes his Support card at the start of his turn.

Support - Solar Flare

If he has at least 3 -1 defense tokens, he changes his current Support card at the start of his turn

Support - Staff of Ra

If Ra changes his Paradigm card and he has at least 1 point of damage, he changes his support card as well.

Feel free to check this out.

I actually was thinking of a completely different approach yesterday. Unfortuantely, I don't have a specific example, but here is the general idea:

The villain will start out with a very basic script based on the scenario (move towards X, attack, Y, etc.).

Now, the heroes turn is going to change just a bit. When they roll for movement, instead of taking a die based on the arrow, they get to choose which die they want. One die they place for their movement, the other they give to the villain. This works for msot heroes easily since many of them only roll 2 dice, and the others can be modified a bit.

The villain will have a chart with the number 1 - 6. Every time they get a die, you out it on the corrosponding number. Which spots have dice and how many, will change the villains script a bit.

The lower numbers can be more powerful, since that means the heroes can move more, while the higher one weaker. This can give the players some interesting choices on their turn, which die do they keep, and which die do they give the villain. This can also keep it somewhat unpredictable, since you don't know the villains possibility until you roll after your turn. Also, each player adds a die on their turn, so you won't know the full villain turn until the end.

Since each player would add a die, it is really easy to scale for the number of players.

Since all you need is a single sheet for the villains, it can be really easy to print out and assemble (unlike printing a deck of cards).

Here are some possibilties for the different spots:

Movement: At the beginning of the villain turn, roll all the dice here for movement, and pick the highest roll.

Cards/powers: If X dice are here, put Y into play and remove all the dice.

Charged Up: Roll all the dice here for attack/defense (maybe for specific powers).

Script: If there are X dice here, the villain targets the hero with the lowest health (with X attack, possibly using the dice), or does something else.

Obviously, this needs to be created for each villain and scenario, but I found the appraoch interesting. It isn't even close to the same game as competitive play, but that might not be a bad thing.