Fix to TacAI Decoys

[halfzware]

I got involved in a chat yesterday about dom2 in the irc room. The subject of decoys came up - someone suggested a post about it, so here it is.

Decoys are an abuse of the tac targetting ai. Setting a screen of 1 unit squads seriously reduces the ability for archers and mages to cause the amount of damage they should, as well as seriously delaying melee units (this effect is much more pronounced in the small scale combat regularly seen in early multiplayer fights/blitz games - where the 3-4 turns of protection from archers and mages can have a profound effect).

My suggestion to a fix for this is pretty simple (but would probably be too large a change for patch -- maybe dom3 though):

All unit types have a minimum unit size. For example, militia have a minimum unit size of 5. If militia are placed in squads smaller than 5, they'll rout automatically. Of course more powerful units should have smaller minimum sizes.

But setting a minimum size would force you to at the very least invest some cost when employing decoys - while at the same time giving ranged weapons a better chance of routing or killing those decoys effectively (a mage casting falling fire on a group of 5 has a much greater chance of hitting a couple of them - and therefore routing them).

Anyway - this might not be the perfect solution - for some things it might not be thematic to enforce minimum sizes (vinemen spring to mind) - but you can make a case that most weaker summons need to be used in groups for mages to properly control them (weak undead for instance).

Maybe some food for thought at least.

[Truper]

I love this idea! (I was the one who insisted halfzware post it).

[Saber Cherry]

Agreed, min squad size sounds pretty good. Although dynamic regrouping might be nice too. For example, if two groups of 5 light infantry advance and one group loses a soldier, the understrength squad could merge (on the battlefield) with the other one, and subsequently all 9 soldiers would be a single squad.

If I were a brave, proud, nationalistic soldier and 90% of my squad was annihilated, leaving me alone... I'd probably try to join another squad during the battle, rather than routing or fighting alone.

As I understand it, your solution is to force-rout tiny squads once combat occurs. However... allowing the player to form single-member "squads" of very weak units in the first place is rather silly. No soldier would take their leader seriously if he said "Bob, you stand alone at the front, while everyone else is in a dense wedge formation 200m behind you." The only real problem I see with preventing this on the user-side is that when you have only 4 militias in a province (perhaps they retreated there) you can't get them out easily, since you can't form a squad of 4 weak units. And adding more rules to allow formation of small squads in situation X but not situation Y would cause problems and confusion.

The main problem I see with your proposed version of squad-size control, even if dynamic regrouping were added, is that it might be problematic for tiny armies. For example, what if you buy 3 points of province defense, enough to prevent an enemy scout from singlehandedly stealing a province? Your PD will generally rout... forcing their leader to rout... and giving an automatic loss. Also, as you mentioned, auto-routing cannot solve the problem of mindless AI decoys, since they can't rout. Yet enforcing squad size limits on the user-side brings other problems, as I mentioned.

So... it seems like more thinking needs to be done to address the corner-cases, as there is no obvious, universal answer that is clean and simple. But I think it's a very good approach to eliminating the tactical decoy problem (and adding some realism to battlefield deployment), if those niggling details can be addressed. This forum is full of smart people; I'm sure someone will have an idea.

[halfzware]

Dynamic squad regrouping I'm not so sure about - as I think the code for this could be tricky. I would like to see some kind of rally system though, for instance if you have guys on their way off the screen but the entire enemy army routs it'd be nice for your guys to stop running off.

The only real problem I see with preventing this on the user-side is that when you have only 4 militias in a province (perhaps they retreated there) you can't get them out easily, since you can't form a squad of 4 weak units. And adding more rules to allow formation of small squads in situation X but not situation Y would cause problems and confusion.

I think preventing you from forming the groups in the first place is not a good solution.

Perhaps a simple addition to the ui would be best - like a moral number visible on the unit formation screen. If a squad has below "squad minimum" they will have 0 squad moral. Once it goes to minimum they'll have 10 moral, and so on using some multiplier like ( base moral * ( squad size - minimum squad size ) ) / minimum squad size.

But yeah - basically being able to put together units however you want should be possible - what happens when you're on the field with the fire flying is another thing.

[Arralen]

I think you got it all the wrong way. The best fix might be the one that sounds most complicated at first: Fixing the targeting algorithm.

If a squad's target is set to anything else than "Large Monster", squads with only 1 single unit in it should be ignored, unless there is no other target at all.

Mages obviously must be handled a bit different. But they already choose their spells dependent on how much damage they do to the target. So it's simply a matter of giving them better chances to actually do some damage at all: Atm, bolt spells are mostly useless against single targets, as they usually hit the right square only 30% of the time. And then can still fall on an unoccupied slot, or be blocked by a shield. If low-level bolt spells would have way higher accuracy, the IMHO the mages would use them automatically to fry any decoys in range. And as the spells have low path requirements, fatigue gain would be small.

Btw., IMHO this fix has the biggest chances to get implemented: One of the devs said once it's much more fun to fiddle with the AI than to alter the Ui or code new battle mechanics

[The_Tauren13]

The auto routing of small squads wouldnt fix the problem. Archers would still target the single troops set in front even if they immediately retreat...

[Arralen]

Not quite true:

Targeting happens at the very beginning of each combat turn, as well as Re-targeting when the original target routs.

So in the first turn, the archers would target the decoys.
In the second turn, the decoys would be routing and the archers would switch to another, non-routed target, prefering those not in direct (base-to-base) contact with friendly troops.

[halfzware]

If a squad's target is set to anything else than "Large Monster", squads with only 1 single unit in it should be ignored, unless there is no other target at all.

This is more problematic than it seems. The obvious for instance would be single pretenders, or other SCs. But even smaller for instances would be other strong single units. A blessed hydra for instance. If you don't like that example, how about a blessed jotun, a black hunter, or even a knight. These single units do need to be targetted and differentiation would require tagging all the units appropriate for individual targetting - this would be essentially the same work as setting minimum unit sizes only the result would be less desirable.

Mages obviously must be handled a bit different. But they already choose their spells dependent on how much damage they do to the target. So it's simply a matter of giving them better chances to actually do some damage at all: Atm, bolt spells are mostly useless against single targets, as they usually hit the right square only 30% of the time. And then can still fall on an unoccupied slot, or be blocked by a shield. If low-level bolt spells would have way higher accuracy, the IMHO the mages would use them automatically to fry any decoys in range. And as the spells have low path requirements, fatigue gain would be small.

This again opens itself to potential abuse - even if bolt spells were more accurate a group of two militas could still eat up 2 turns of spell casting in order to rout/kill them.

The routing system isn't perfect - even handling minimum groupings of 5 militia can be abused. The point though is increasing the costs involved when deploying 'gamey' tactics. The hidden value in minimum unit sizes isn't just in its initial cost - 5 milita is only 15 gold in CB. The real value is that constant renewal of your decoy groups is more difficult and more costly (each time you use a decoy, they'll probably die or rout - meaning you have to constantly renew them - in groups of 5 instead of individual soldiers). This is especially important in early game fights where the tactic is most abusive and where every unit counts.

[Ironhawk]

Nah, minimum group size is not the way to go. It might seem good on the face of it, but it breaks down. Think about mixing up unit types - how could you possibly even guess what the min group size for it would be?

Arralen is right that the correct fix is to repair the targetting ai. Archers need to look at how much of thier volley in each turn is "wasted". So like 20 archers firing at 1 militia results in massive waste. But 20 archers firing at 50 militia will mean that probably every single arrow will hit and perhaps kill a unit.

This said tho: whether or not any changes are made to the targetting AI, there will always be loopholes if the algorithm is static.

[halfzware]

Nah, minimum group size is not the way to go. It might seem good on the face of it, but it breaks down. Think about mixing up unit types - how could you possibly even guess what the min group size for it would be?

for each unit in squad
add element unit size to squad moral total
divide squad moral total by number of units

Also known as averaging, if averaging isn't to your liking there are other tools that can be used simply. Standard deviation, etc. Code is very good at handling this kind of thing - and methods for aggregating numbers are many and well known.

Arralen is right that the correct fix is to repair the targetting ai. Archers need to look at how much of thier volley in each turn is "wasted". So like 20 archers firing at 1 militia results in massive waste. But 20 archers firing at 50 militia will mean that probably every single arrow will hit and perhaps kill a unit.

I agree that having a targetting AI that behaves as a human controlling it would want it to behave is the best answer. But as you say, the most likely scenario is that people will find methods to abuse the targetting logic here also. There will always be loopholes, which is why its better to prevent worst case abuse at the onset. Minimum unit size is a simple - I dare say elegant way to achieve this.