Wandering monsters targeting

[teelaurila]

There's a stronger wandering monster presence in the game now, since I last played some maybe 6 months ago. I understood the purpose is to make player defend his cities with other than 8 spearmen "police" garrisons. A perhaps unforeseen consequence is that AI early settler spam now often meets with monsters that then make neutral cities. I think this all makes the game a lot more interesting so that's great! (At least on expert) There seems to be one but, though: Since the monster stacks pick a new "weakest city" target every turn, it's very easy to pinball a stack between two cities for as long as it takes to muster a force to crush it. This is done by emptying the garrisons on alternating turns. As long as there is at least some strategic strength in the garrisons. This makes garrisoning a lot more pointless: you can practically always buy time. (have not yet seen a stack too fast, nor 2 stacks)

So it would seem the "dumber" targeting AI would actually be superior: Pick the weakest city when spawning, and stay on target. ?

btw what is the effect of raging monsters option now?

[Seravy]

Since the monster stacks pick a new "weakest city" target every turn, it's very easy to pinball a stack between two cities for as long as it takes to muster a force to crush it.

I'm aware of this problem but so far no one managed to suggest a better way of solving it and I don't have a good idea either. If monsters don't re-target, you can have one garrison which moves to whichever city the monsters targeted first, while all other cities are undefended.
Maybe instead of the quantity of units, their quality should be used in the formula? Then weak units like spearmen being moved in/out of cities wouldn't be enough to change the target? But then it should be relative to the monster stack's strength because 3 phantom warriors do care about the number of spearmen...while sky drakes don't.

The option doubles the total budget for the monster stacks, but without changing the cap on each individual unit. So the strongest monster that can appear remains the same, but in general there will be more and stronger monsters in each stack.

[teelaurila]

Since the monster stacks pick a new "weakest city" target every turn, it's very easy to pinball a stack between two cities for as long as it takes to muster a force to crush it.

I'm aware of this problem but so far no one managed to suggest a better way of solving it and I don't have a good idea either. If monsters don't re-target, you can have one garrison which moves to whichever city the monsters targeted first, while all other cities are undefended.

Thanks for the answers! I'm not sure I understand what you mean here, though? I presume you mean there is no way to tell when a stack spawns in the targeting routine, so that either every stack targets every turn, or a target is picked globally once? Naturally every stack would need to target something near it once.

The simplest might be worth trying: attack the nearest city, period. This means truly "random" attacks, so the player cannot game the system. It does mean cities without nearby monster sources are safe, but in practice that already is the case. (unless there is war and weak enemy stacks roaming about)
A milder version of the same is to reduce the weight on garrison relative to distance. Perhaps one could add in a rand in there?

Usually cities are 4-5 tiles apart. Since one can have 0 units in the other city, we might have 
units+2*distance+2d6 
The idea being that each element has roughly the same weight (units here vs neighbouring city, randomness). So the stack will attack a nearby empty city always if neighbours are defended, and a empty neighboring empty city will draw in the stack from a garrisoned city only about half the time. But you can never rely 100% on what the stack will do.
Assuming rands are a possibility here?

If the targeting routine can use strategic strengths, it might seem sensible for the monsters to attack the "nearest city with strength less than my strength" (in some reasonable radius). If one is not found, perhaps just move about randomly. Even that will end up attacking something eventually, and could force the player to attack the stack instead, which of course puts the human at a disadvantage.

In general, I would advocate adding randomness somehow to the way the random monsters work. The power of gaming the AI is in being able to 100% predict what happens next, and thus not needing to protect assets. Added randomness thwarts that! And in this case we can have randomness simply by these randomly placed stacks attacking the nearest city! Perhaps it can go more involved from there, but perhaps there is even no need?

[Seravy]

Assuming rands are a possibility here?

Not really. It's evaluated every turn so including random would make the monsters move...randomly, never arriving to any particular destination. Same as your pinball strategy except without the need to do anything.

The power of gaming the AI is in being able to 100% predict what happens next, and thus not needing to protect assets. Added randomness thwarts that!

I 100% agree but it can't be done as explained above. The AI does use random in pretty much everything else except overland movement.

I could exclude neutral units from the "clear target and pick another at the beginning of the turn" standard behavior, but that comes with several negative consequences :
-The player can put nearby strong units into the city and wait for the monsters to arrive, giving them the advantage of having wall, Guardian and city enchantment bonus, and having the first turn. Meanwhile currently, they have to attack the monsters, or the monsters will go to another city when they see the city is fully guarded.
-If anything blocks the path, the monsters will be stuck eternally. This already happens but currently, the monsters will be able to attack another city as soon as one that's not blocked becomes less well guarded.
-Monsters won't be able to react to a target changing control to another player or becoming destroyed.

But really the first one is the big problem, I don't want to give the player the ability to position a stack of fast units (like pegasai) in their middle city then fly them into whichever city was nearest to the monsters spawned and kill them easily. The "pinball" tactic at least keeps those fast units busy forever, and as soon as the monster stack ends a turn on a tile where it can reach two or more cities, it's checkmate.

Using strategic strength is most likely possible. It requires using far calls to the strength calculating function but I've learned how to do that safely by now.

[Seravy]

Started working on this. Question is how much priority should the defending army strength be worth.
I'm considering to go with  (1/32 of the monster stack's strength) = 1 tile distance

For example if there are 2 targets, one at 25% and one at 50% of the strength of the monster stack, it's willing to travel up to 8 extra tiles (2 cities worth of distance) for that difference. So you need about 15-25% of the monster stack as garrison in both cities to be able to safely "pinball" the monster stack, depending on the actual distances involved. That's not much, but at the very least, spearmen won't do it. As a practical example, against 6 hell hounds, you'll probably need at least 2 barbarian cavalry per city. For 6 werewolves, you'll need need about 6 of them each.

[teelaurila]

That should be a clear improvement. Also then you cannot use a few strong units vs lots of clutter to lure the stack into the stronger defenses. Though that scenario is most likely with death, which I am presently playing.

[Seravy]

What's everyone's experience with this new neutral targeting system?
I'm currently playing a Dwarf test game but I haven't seen enough to form an opinion yet. It feels as if maybe monsters were willing to go a bit too far out of the way to avoid garrisons but I do make sure I had good garrisons everywhere and dwarf units are quite powerful. The fact my starting continent is large and has a chokepoint blocked by a city isn't helping either. Any neutral selecting the city on the other side is getting stuck as expected. Maybe I should add a condition to ensure a path does exist. I didn't do that because neutrals have to be able to move through other units, but I realized that's not a problem, as looking for the path does use the same function as real movement so that special condition won't be lost.

[teelaurila]

It's definedly better in my experience. With decent garrisons the "pinball" is still possible, though. The more relevant would be just how low you can go in garrisons, which I hadn't tried.

On the other hand, it seems neutrals are at times equal or more threatening than my actual enemies. I've been playing master, though. But I don't think the neutrals should grow so strong as to single-handedly wipe mostly-magician garrisons. I suppose combat casting on the wizards side does help.

On yet another side, I've played another death game, where I get almost the same amount of summons from undead neutrals than a proper summoner would summon. Such stacks generally aren't quite as strong as summoned-as-planned, but many of the creatures also are stronger than anyone in the game can summon.

So while these neutrals are interesting, they maybe are a bit too dominant a feature. Perhaps the strongest critters could progress slower, but overall stack strength perhaps could stay?

[Seravy]

"strongest critters could progress slower"
It's scaled by difficulty. You were playing on Master so that's why they were stronger.

[teelaurila]

"strongest critters could progress slower"
It's scaled by difficulty. You were playing on Master so that's why they were stronger.

Well, perhaps that's the thing: I might suggest to re-consider the difficulty scaling of this strong neutrals feature. It feels to me that at master it no longer simply causes the need for reasonable garrisons, but is a "major enemy" well into the mid-game. A side effect is that it's also a resource for a death player, which seem to me to have become clearly the easiest build.

At the highest levels, the actual opponents come with a steep power curve. There does not seem to be a need to scale up the neutrals likewise - the player will not be able to garrison likewise faster.  So it starts to feel the difficulty comes from a "wrong" place. I am not saying neutrals shouldn't scale with difficulty, but perhaps not quite so steeply. This way AI players would also be less threatened by neutrals at high levels, which I think is better putting the challenge to the right place.