AI Planar Travel

[Seravy]

Let's see the pros and cons for each...

1 .
+Gets rid of a spell that's not fun to play against if the AI has it and is quite unfair (allowing one to attack another without a chance of retaliation).
-But doesn't really remove it completely as Astral Gate and Planar Travel items still remain. So it only gets rid of a third of the problem. (the AI is also bad as using the other two, and they are also unfair) Alternately we can get rid of all three if we really want to.
-While unfair, using Planar Travel is fun and works fairly nicely for playing "peaceful" games, as it's possible to colonize some of the other plane, expanding territory without a war. So removal of these spells would make peaceful strategies even less playable.
-If only Planar Travel is removed, Life without high amount of books will be a gamble with a much greater chance of missing the only planar movement spell. Having two ensured a very high chance of getting at least one.
-Obviously, this would require the most amount of work to implement.
-Reduces spell redundancy in Life, which already has the lowest spell redundancy, making the realm even stronger.
+Free slot(s) to use for new spell(s or item power).
-However, pretty hard to come up with new Life spells as the realm doesn't really have missing features anymore. (well, it's bad at summoning, but that is intentional. A new early Life summon would be useful but would make the realm much less unique)

2.
+Easy to implement
-AI stays dumb at using planar travelling mechanics

3.
-Very hard to implement
+AI gets smarter
-AI will learn to use a very unfair strategy against the player which might make the game more frustrating - this might be especially bad because it would happen to the realm which is currently the least frustrating to play against.

[Nelphine]

We stopped the ai from using shadow demons because they would just fly across and die. They weren't effective. They also prevented the ai from using the shadow demons as effective doomstacks on their home plane.

Given this, option 2 is the only option that is similar to why we stopped the shadow demons from planar travelling. I'm not sure I really care about it though, as I've never seen it, so it's rare enough that the AI doesn't usually have problems due to it. On the other hand, I've definitely seen the AI bring over a hero and some supporting units and actually cause havoc. I'd hate to lose that potential. So I'm against option 2.

Personally, astral gate and planar travel are the method life uses to avoid the human controlling all 6 toward and winning with SoM. However, astral gates are too easy to block (I regularly block them all without even exploring the other plane). Therefore planar travel is the most important spell for late game life. So I'm completely against option 1.

Additionally, I would actually prefer option 3, but only for master and lunatic. I firmly believe some ai decision making should be chosen by difficulty. However, given that it's very hard to implement and I do NOT want it for anything below master which would make it even more complicated to implement, I'd say no to option 3.


Therefore, my vote is to do nothing. I think these options are trying to fix a problem that is so rare that it isn't worth fixing.

[Seravy]

Well, I don't think option 2 is that bad. I would gladly lose one of my poorly defended cities if I get 3 pieces of top tier equipment in exchange. (In this last game, the enemy chosen one brought me an Excalibur, Capes of Pikeru and Sylph Chain. The chain is not a big deal but the cape is a +4 Def +4 Res Regeneration and Divine Protection accessory and the Excalibur is a max attack max hit phantasmal sword, both are like top 3 predefined items in their category.)

And even greater problem is, being a hero, it already receives a 1/2 total stack power penalty, and another 1/4 for being a stack of fewer than 3 units. So It can only attack things that are 1/8 of the hero's actual total force, or less. Even if it's a very good hero, it'll only be able to attack really poorly defended targets. 

But really, the worst, it's always bad for the AI. If the hero is poorly equipped, it'll just die and won't be a threat. If it's very well equipped, it might manage to do some damage, but eventually the player will find a way to hunt it down (either through gathering nearby troops or by using a combat spell that the hero isn't immune to) and then they get like 5-10k worth of artifacts that win the game by themselves.

There might be a 3% chance the hero actually is immune to everything the player can do against it, and manages to take out several cities, but...in the remaining 97% it's just helping the player, and even in that 3%, the player can prevent the hero from attacking by having at least 12.5% of the hero's military force in their garrisons, or they could leave a node open so the hero steps on it and remains there for garrison - there won't be other units to replace it so it won't ever leave the node.


Now for option 3. What if it wasn't based on difficulty, but time? Basically, it's unfair if the AI does that BEFORE towers break. But if it does so when towers are broken, especially if they are held by the human, then it's not only fair, but outright necessary. So what if we had option 3, but with the additional condition of "do not cast planar travel unless there is at least one broken tower in the world."? (or two towers, or whichever we prefer)

[Nelphine]

What I'm saying is that 99% of the time the AI doesn't send the hero by itself anyway. So, your 97% is only ~1% of total games. On the other hand, I'd wager 2-3% of the time the AI sends the hero across and has other units with it.

So sure, your option helps for that very rare scenario. But it hurts for the slightly less but still very rare scenario where the hero actually would be dangerous.

As for time: no I don't think so. On my faster strategies, ie, on master lunatic, I regularly break my first tower by 1405. Which is the same as just letting them do it all the time. On the other hand, some advanced expert games don't break them until 1415+ so you may as well just say never allow it.

So, sure you could do that if you wanted. But a barbarian ai, which breaks towers about 5 years earlier than other ai, will totally throw that off. I don't like basing someone's spell use on a different ai race choice. So I'd rather keep it by difficulty.

[Nelphine]

Also re option 2. I'm basing it on my own experience, which could certainly be wrong. So while I'm against it, I'm not terribly against it.

Option 3 I definitely want to restrict to the hardest difficulties.

[Seravy]

Actually I think we might have one more option :

4. Make Planar Travel allow the entire stack to shift plane like Wind Walking as Zitro's idea, BUT make it rare, as this effect is strictly superior to the uncommon Astral Gate, and also make it cost 250. Doing so means we need to put one of the rares into uncommon, the only reasonable candidate for this I believe is Prosperity.
+AI friendly : AI does not need to build a stack of Planar Travel units, as long as the spell does target the strongest "outside city" unit, it'll likely be part of a good stack
+Semi-decent amount of work : While planar movement still needs to be completely redone to make sure the AI only shifts larger stacks and only if the destination plane is the one that needs the troops more, no new stackbuilding procedure is needed and all plane shifting decisions can always assume every unit is capable of moving to the other plane, also AI logic for wind walking to not cast it redundantly in the same stack can be reused for this fairly easily
-The unit ability (from items or on Shadow Demons) will either need to work the same way (but I think that's not acceptable as those sources are not rare tier), or if not, it'll be somewhat confusing and inconsistent for human players, and need to be handled differently (or ignored) for the AI.
-Space in the targeting procedure for overland buffs might be an issue, need to check that
-Same as "1.", losing one of the two uncommon spells, so early peaceful settling of the other plane will be much less likely for mixed realm Life wizards.

What I'm saying is that 99% of the time the AI doesn't send the hero by itself anyway.

It does, though, as it's very unlikely the AI casts more than one Planar Travel in a turn, but does use Planar movement at the beginning of each turn. If you see multiple units, that's likely an Astral Gate, which still only drops one unit a turn, but does so every turn so the units can and will stack up fairly quickly. Unfortunately, Planar Travel units almost always come alone. (If at all - the majority of Planar Travel spells are eaten by settlers anyway, other units rarely get any)
If you actually did find a planar travel stack belonging to an AI (with the units all having the spell on them), do post the save because otherwise I find it hard to believe that can happen.
...I checked and Planar Travel is being cast before turn 100 by the AI to ensure settlers get through, past that it can only be cast through picking buffs at a fairly low chance. Unless there are like 4 Divine Orders out and the AI buffs like crazy, there won't be more than a few nonsettler units with the spell at best.

So, sure you could do that if you wanted. But a barbarian ai, which breaks towers about 5 years earlier than other ai, will totally throw that off. I don't like basing someone's spell use on a different ai race choice. So I'd rather keep it by difficulty.

Or we can just base it on turn count. Turn 200 or later, use, before, do not use. Entering the other plane is not that critical until very rare globals or spell of mastery are already being cast.

[Nelphine]

Again, playing at master and lunatic, turn 200 doesn't exist. The game is over before then. Master and lunatic, if the spell is to be relevant, it needs to be before turn 80. (To match very rares.)

And yes I agree that the spell is virtually never cast; but as I primarily play life, I regularly gift divine order to every AI. I'd say my average game has at least 3 divine orders on it. So I do see stacks with mutiple planar travel.

It's still super rare, but in my experience, it's more common than the AI sending a lone hero over to die with artifacts that actually do anything. Again, this is my experience, I'm not THAT concerned if you take option 2.


I'd avoid option 4, at least with prosperity. Prosperity is a great spell, and I consider it better than Incarnation.

[teelaurila]

I thought exaltation quite underwhelming as a rare. You get 3 more health than healing, and being able to cast it pre-emptively only shifts the AI to snipe at another unit, if this is what they are doing. It's only been of some use to me with heroes, since if you have vulnerable units with you (magicians), you generally have more than 1. But maybe I've just not got how to use it.

[Seravy]

Again, playing at master and lunatic, turn 200 doesn't exist. The game is over before then. Master and lunatic, if the spell is to be relevant, it needs to be before turn 80. (To match very rares.)

You're kidding. Higher difficulty games take much longer to finish because the enemy is stronger.
And very rares come at year 18 = turn 216. We made sure of that when redesigning the research costs.

The only turn 80 very rares possible are the ones found in treasure...

I thought exaltation quite underwhelming as a rare.

It was moved up to rare from uncommon because it can grant extra health to units even if they suffered irrecoverable damage, in other words it "heals" irrecoverable damage.

[Nelphine]

No, higher difficulty games don't take longer to win because the enemy is much stronger. Instead, you lose them much much faster, so if you don't have a strategy that lets you win before you lose, you don't win. So, if say, an advanced AI will kill you in year 25, then expert will kill you in year 21, master in year 17, lunatic in year 13. If you don't win before that, you either have already won, but the victory condition hasn't clicked in, or you lose. However, I've never seen these mythical games where both sides are even and the game drags out for years or decades.

Very rares come at year 18.. if you don't have any research boosts. If you're a specialist, then its year 16.2. If you're also a sagemaster, it's, what, year 12? If you're a lunatic AI, it's then reduced further to year 10.8? If you're on large or huge land mass size, then it gets further reduced, say to year 10?

And yes, treasure is a thing. My draconian bowmen game, I have two very rares in 1404. The AI doesn't get them as often as the human, but lunatic, you PLAN around very rares (both yours and the AI) to happen by 1408 at the latest. To do otherwise literally loses you the game.

Games on master or lunatic don't play the same as lower difficulties. You have to use completely different strategies, which drastically change when things happen.