PDA

View Full Version : Artificial stupidity


Arker
July 30th, 2006, 11:54 PM
The worst thing about this game for me is the so-called AI. I'm really steamed right now. Scripted an atlantian king to cast a handful of spells then attack. He has the champions trident, but also a chest wound, so I'd really rather not have him attack, but the thing is, a few turns back when he was on 'stay behind troops' he decided to cast the cold-radiation spell and destroyed my mages with it. Dumb &(@%... grr... so I figured having him attack would prevent that. Nope. He casts his spells, and throws in that cold radation trick again all on his own, then goes to attack, killing another mage (the just arrived replacement from the capital for the two he killed before) and several spearmen in the process.

There's got to be something to do about this, it's just way too stupid. I've come up with a lot of little tricks to try and minimise it, but I haven't found any way to keep it from happening entirely. Very frustrating. Anyone else?

Any word on whether Dom III will pull the same crap? I think I'm going to wait to be convinced it doesn't before I buy it. I'm so sick of this.

Wish
July 31st, 2006, 12:13 AM
you could... um... jut move him away from the mages for his starting position?

Arker
July 31st, 2006, 12:35 AM
I do. He moves over to them and kills them anyway.

Arker
July 31st, 2006, 01:38 AM
Started a new game with Jotunheim. At least breath of winter doesn't cause friendly-fire deaths with the giants. Still just hilariously stupid though.

Niefel giant prophet. Scripted:

Divine bless
Holy avenger
Fanaticism
Attack one turn
Fanaticism
Attack

What's he do?

Divine bless
Holy avenger
BREATH OF WINTER
Attack one turn
Fanaticism
Attack

I'm not making this up. Heck, I couldn't make this up. Blows my mind. This guy has that effect already, why would he waste fatigue casting it? *ROFL*

I've gone through the replay over and over, and thought and thought, and the only thing I can come up with is this. He hits the first fanaticism, no one has been hit yet, so there's no morale loss to fix, so he decides to cast another spell instead. That much of it makes sense, I guess. But why he picks a spell that does nothing but increase his fatigue is a mystery. Why his Atlantean predecessor would consistently pick that *same* spell, and proceed to decimate his own mages with it, is similarly a mystery. Playing man last week I had a Jade Sorceress with the same fixation. Whenever she got a chance, she'd fire off that Breath of Winter and then get lonely and start snuggling up with the lizards. This was particularly funny when she chose to do this while fighting Jotun, so not only was there no chance of it ever hitting anyone except her own side, it wouldn't hurt the enemy even if she did get close to them somehow.

The spell-casting AI in this game is seriously whack.

Ygorl
July 31st, 2006, 01:57 AM
You're right about why he didn't cast Fanaticism. In a case like that, the AI picks a spell out of everything he might cast. The AI tends to favor spells that protect the caster, followed by spells that summon dudes, unless there's a nice target available for a direct-damage spell.

It might be true that Niefel Giants have "chill" (which causes fatigue and sometimes chilling of neighbors) but not "Breath of Winter" (which also causes actual damage). I suspect this is the case, actually? Anyway, the AI likes the spell, and isn't aware that it can sometimes hurt your own guys. It's maybe the biggest spellcasting AI frustration in the game.

The way the AI works, by the way, (if I'm remembering right) is that it first looks at the scripted spell - if it would have no effect, or if the script is such that the caster's spell is not specified, it practice-casts a bunch of spells (several times each?) and picks the one that has the best effect to actually cast. Not sure how "best" is defined.

shovah
July 31st, 2006, 09:10 AM
I had the same problem with atlantis, the answer is either spread your mages out and set them to their spells then just cast (so they never move) or script them all to BoW.

Saxon
July 31st, 2006, 09:29 AM
I recommend spreading your mages out, no matter what. One group has a higher chance of attracting arrow fire and of being hit by stray shots. Also, if they fatigue out and your troops run, the enemy has to cover more ground to kill all the mages. Sometimes one or two will recover from fatigue and be able to flee.

I agree, you see some less than ideal spell choices at times, but you will also get some very pleasant surprises. I never script the spells that hold an enemy in place, yet one of my casters did just that to a flying pretender god. End result, dead pretender.

It was once described as giving orders and hoping they are followed. As a manager of real life humans, I can tell you that the results you see are very lifelike. Not that a game about magic, dragons and talking lizards should try to create “realism,” but in this case they have!

shovah
July 31st, 2006, 10:16 AM
and if you were a mage about to be charged by a group of knights even though you could kill some with a spell wouldnt you cast a self protection spell?

Oversway
July 31st, 2006, 11:11 AM
The AI is pretty BoW crazy, though. Once you've researched it, you can pretty much bet that any mage with 2w (or 1w and a gem) will cast it if they are scripted to cast some other spell that would have no effect (like fanaticism on a turn where troops do not have a morale check).

Graeme Dice
July 31st, 2006, 11:55 AM
Arker said:
I do. He moves over to them and kills them anyway.



First off, he will only cast breath of winter if you give him an untyped "cast" order, or if you didn't put him close enough to the front lines so that his scripted spells would reach any enemies, or if you asked him to cast spells that require gems. and there's only a trivial number of opponents. Secondly, if you tell him to attack, then he will only move into range of your other mages if there is an enemy in that direction.

Graeme Dice
July 31st, 2006, 11:56 AM
Arker said:
I'm not making this up. Heck, I couldn't make this up. Blows my mind. This guy has that effect already, why would he waste fatigue casting it?



No, he doesn't have breath of winter by default. He has a chill aura that causes fatigue and can freeze opponents. Breath of winter is a completely separate effect.

NTJedi
July 31st, 2006, 01:58 PM
Arker said:
I do. He moves over to them and kills them anyway.



Move him closer to the front and move your mages closer to the back.
OR
Place him on one side of the battlefield and place your mages on the exact opposite side of the battlefield.

Arralen
July 31st, 2006, 06:46 PM
A mage does not move over to other mages ... in fact he doesn't even know about them. He moved over behind the rearmost squad, as he was ordered to do!

Never, I mean NEVER set a mage to "stay behind troops" !!

A commander set to "stay behind troops" will move to the rear of the rearmost non-fleeing, non-bodyguard squad and stay there until the squad flees - but nothing more. No matter if he's a mage or a priest or has spellcasting ability from an item, he wouldn't cast any spells.(or would attack an enemy unit in melee, unless attacked himself)

But you want your mages to cast spells. Therefore their script should read:
"cast [spell XY]"
"cast [spell XY]"
"cast [spell XY]" OR "attack -" if you want him to move up to the front line
"cast [spell XY]"
"cast spells"

Be careful to choose your spells XY in such a way that the mage will be able to cast them.

If you want your mage to attack after buffing up, make sure there's a way for him to reach enemy forces. "attack nearest" might be a bad idea after 4 rounds of spellcasting - you troops will be engaged already as "hold and attack" has them on the move on the 3rd turn. With troops engaged and squads fanning out he won't be able to find his way around them and get stuck in their rear.
Therefore, use "attack nearest" only on turn 3 but not later, or place your troops and mage/pretender in such a way that there's a clear path for him towards the enemy. If he needs 3 or 4 turns to buff up, better set him to the side and use "attack rearmost" or "attack archers"... . Just make sure he's strong enough or assign him some bodyguards.

shovah
July 31st, 2006, 08:10 PM
Atlantian kings are great thugs, add on some decent equipment and they can all do nice buffs: all can do quickness+BoW, earth can do summon earthpower+invunerability, nature gets regen, fire gets fireshield, air gets mistform and mirror image, astral gets quite alot,death causes extra fear and i guess blood could cast blood vengance although ive never seen a blood king)

Wish
July 31st, 2006, 08:50 PM
I have seen blood kings.

shovah
July 31st, 2006, 09:14 PM
I thought they existed, just a sort of myth for me.

For equipping kings give non earth kings decent armour (i suggest either dragonscale for 18 prot and only a small defence reduction) or marble armour for alot of prot), a wraithsword(or hellsword if you must), preferably a horror helm or some other helm (maybe starshine), lucky pendant and then any extras you want like reinvig boots/regen ring or something. In non cb give then a bloodthorn and a charcoal/vine/accursed shield, this will give them fireshield/very easy to hit enemies/super high defence.

The best thugs imo are earth and astral kings, all they need is a wraithsword/blood thorn+shield, some reasonable armour for the astral and maybe a horror helm and they can run through small/medium sized armies. When fully tooled they can do great things (quickness+BoW+30 prot/astral and luck along with good equipment is amazing) and make great (and recruitable!) thugs until you get golems (i suggest you empower an earth king to summon them) and tartarians (i always have a d6 pretender where possible then just give him a skull staff and let him mass produce them while he/a vine king GoR's them)

Arker
July 31st, 2006, 09:25 PM
They definitely DO cast on 'stay behind troops.'

Arralen
August 1st, 2006, 04:11 AM
No.

capnq
August 1st, 2006, 07:19 AM
I script mages to (cast specific spell)x5, stay behind troops all the time. They don't stop casting spells until they run out of fatigue.

Arralen
August 1st, 2006, 02:17 PM
Just tested - I stand corrected.

This must have been changed somewhere along the development path - I vaguely remember testing this some (long) time ago and mages behind troops moved but didn't cast then.

However, if you don't want your water mage to move up to the nearest, rearmost squad, simply set him to "cast spells", and he'll stay put.

quantum_mechani
August 1st, 2006, 03:51 PM
shovah said:
I thought they existed, just a sort of myth for me.


There is exactly one blood king (unless you count all the various demon lords), a corrupted earth king.

It is pretty easy to avoid breath of winter casualties if you spread out your mages and set them to cast spells at the end of their script, which are good ideas to do anyway.

shovah
August 1st, 2006, 07:33 PM
I meant an king of the deep (atlantis) with a blood random, i know about father illearth.

Wish
August 2nd, 2006, 02:41 AM
I have had such a king of the deep, but then I've tooled around with very long atlantean games.

quantum_mechani
August 2nd, 2006, 03:11 AM
shovah said:
I meant an king of the deep (atlantis) with a blood random, i know about father illearth.

My mistake- what I get for skimming the thread. http://forum.shrapnelgames.com/images/smilies/happy.gif

Valandil
August 2nd, 2006, 08:26 PM
"I thought they existed, just a sort of myth for me. "

are you saying that:

a) randoms aren't very random, i.e. that blood is less likely than others, or

b) you havn't built very many kings, or

c) you have been (un)lucky

I suspect c, but helps to check.

Ironhawk
August 2nd, 2006, 10:48 PM
Arralen said:

Just tested - I stand corrected.

This must have been changed somewhere along the development path - I vaguely remember testing this some (long) time ago and mages behind troops moved but didn't cast then.

However, if you don't want your water mage to move up to the nearest, rearmost squad, simply set him to "cast spells", and he'll stay put.




Whoa, they actually do cast? I'm with you I would have bet 100% that they would move and NOT cast. If they do follow the troops and cast tho... that could provide a niche for your weak mages. If they can only manage the close range (5-10) evoc spells you could set them to follow and they could get up close to the enemy and blast them. Hmm...

Arker
August 2nd, 2006, 11:00 PM
Exactly. Except for their tendency to senselessly cast BoW and then kill each other with it, this is a very handy tactical tool.

Also it seems that when you put them on 'stay behind troops' they will fire ranged weapons on and off, keeping their fatigue down and their spellcasting going longer, but on 'cast spells' they ignore their ranged weapons and just cast until they pass out.

Arker
August 3rd, 2006, 11:57 PM
Testing reveals that switching the final order from 'stay behind troops' to 'cast spells' or, for that matter, to null, does not alter the behaviour. I still get mages running across the battlefield to get close to other mages, archers, or illithids and casting BoW. There is some variation, though, sometimes cast BoW THEN run to where it can inflict the most casualties on their own side, sometimes run to position first, THEN cast BoW.

*sigh*

I don't know which is dumber, the mage doing that in the first place, or the other units inside the area of effect simply standing there until they die from it, but either way, there's a stupidity overload going on for sure.

Wish
August 4th, 2006, 02:50 AM
your solution is to not research enchantment, I guess.

Arralen
August 4th, 2006, 04:34 AM
There are certain limitations to the game engine atm - your perceived 'stupidity overload' is the result of, sorry to says that, your ignorance towards them:

1) Commanders do what they are ordered to - literally. They do not 'think about' what might be your intentions. The only freedom of choice is what spell to use when ordered to 'cast'. In that case, the AI will mostly go for buff (target:self) spells first. If there are targets in range, the battle engine will try out several different ranged spells and choose the one which deals the most damage. That does not mean it will do maxed damage, or any damage at all, because the random factors are newly calculated when the spell is actually cast.

2) Units do not 'know' about battlefield hazards. They move into poison clouds as well as into a breath of winter. Likewise, units do not 'know' about their own auras/fields.


Ok, another quick test:

A mage who is ordered to "cast spells", but can cast absolutely no spell defaults to "stay behind troops" it seems. (E.g. Blood-1 mage without slaves and low research - only spell would be 'bleed' which requires a slave).

"sometimes cast BoW THEN run to where it can inflict the most casualties"

Obviously, the only spell the mage could cast at that time was BoW. As he was ordered to cast, he did just that. Afterwards, he was ordered (by the game engine) to "stay behind troops" - and he did just that.

"sometimes run to position first, THEN cast BoW"

Behaviour like that I couldn't recreate unless I ordered the mage to move via "attack" order. If ordered to cast a specific spell 4x and then cast spells, the mage always stays put until running completely out of spells. In that case, he casts "BoW" first and moves on the next turn.
Exception might be heroically quick or quickened mages. The additional action points screw up the order of orders (if that makes sense).

"to where it can inflict the most casualties on their own side"

As unit do not know about auras, fields, units around them (other than who is 'rearmost', 'nearest', 'archer' etc), they can't deliberatly run where they can cause the most casualties. They go for the rearmost unit. Simply make sure your rearmost unit is 100% cold-resistant and keeps a healthy distance from the rest of your troops and you'll be fine. If you bunch up all your vulnerable mages, either by setup or use of "stay behind troops", the 'stupidity overload' is surely not on the games side http://forum.shrapnelgames.com/images/smilies/tongue.gif

Admittedly, there are some possibilities for improvement with the battle engine as well as with the game engine in general http://forum.shrapnelgames.com/images/smilies/wink.gif

Arker
August 4th, 2006, 07:05 PM
Arralen said:
There are certain limitations to the game engine atm - your perceived 'stupidity overload' is the result of, sorry to says that, your ignorance towards them:



Obviously it has limitations, and obviously all game engines have limitations. But, sorry, your second clause is exactly reversed. The perception of stupidity here is a result of learning about those limitations, not of being ignorant of them.


1) Commanders do what they are ordered to - literally. They do not 'think about' what might be your intentions. The only freedom of choice is what spell to use when ordered to 'cast'. In that case, the AI will mostly go for buff (target:self) spells first. If there are targets in range, the battle engine will try out several different ranged spells and choose the one which deals the most damage. That does not mean it will do maxed damage, or any damage at all, because the random factors are newly calculated when the spell is actually cast.



Actually, that's the point, they DON'T just do what they're ordered to do. As you point out, there are several things they will do without being ordered to do. AI is difficult programming, I realise as well as anyone, and practically speaking it's impossible to get it perfect. I don't expect that.

I did expect, however, an AI that with some basic, fairly simple logic checks like 'don't cast this spell if it's more likely to harm your own troops than the enemy' or 'don't move close to friendly troops when you have an area of damage effect following you around.' That seems pretty elementary.


2) Units do not 'know' about battlefield hazards. They move into poison clouds as well as into a breath of winter. Likewise, units do not 'know' about their own auras/fields.



That's a pretty severe limitation, and quite an annoying one from a gameplay perspective, but it wouldn't actually be necessary to fix that to stop what I'm talking about.


Ok, another quick test:

A mage who is ordered to "cast spells", but can cast absolutely no spell defaults to "stay behind troops" it seems. (E.g. Blood-1 mage without slaves and low research - only spell would be 'bleed' which requires a slave).



Yep. It will do that even if not given that command.


"sometimes cast BoW THEN run to where it can inflict the most casualties"

Obviously, the only spell the mage could cast at that time was BoW. As he was ordered to cast, he did just that. Afterwards, he was ordered (by the game engine) to "stay behind troops" - and he did just that.



Sounds right. And easily fixed, by anyone that has access to the code, one would think.


"sometimes run to position first, THEN cast BoW"

Behaviour like that I couldn't recreate unless I ordered the mage to move via "attack" order. If ordered to cast a specific spell 4x and then cast spells, the mage always stays put until running completely out of spells. In that case, he casts "BoW" first and moves on the next turn.
Exception might be heroically quick or quickened mages. The additional action points screw up the order of orders (if that makes sense).



It does, actually, as the latest observation of that behaviour was out of a severely quickened unit. (A random R'lyeh hero, Traitor King? or something like that, L4 water mage with heroic quickness and the spell quickness as well. Definitely NOT ordered to attack or stay behind troops. I have the save game stashed to look at again if there's any point in it.)


"to where it can inflict the most casualties on their own side"

As unit do not know about auras, fields, units around them (other than who is 'rearmost', 'nearest', 'archer' etc), they can't deliberatly run where they can cause the most casualties. They go for the rearmost unit. Simply make sure your rearmost unit is 100% cold-resistant and keeps a healthy distance from the rest of your troops and you'll be fine. If you bunch up all your vulnerable mages, either by setup or use of "stay behind troops", the 'stupidity overload' is surely not on the games side http://forum.shrapnelgames.com/images/smilies/tongue.gif



Your rearmost unit, if you have archers, is pretty much always going to be those archers. In this case, as I recall, I had my Illithids (archers in the broad sense) on the right hand side, firing. The melee troops were spread across right-middle and slightly forward, and the assorted mages spread across the back row tending toward the left, with the exception of the hero with the BoW fixation, who was on the far right side and forward, just behind some melees, to keep him as far as possible from other troops when his casting orders ran out, on the basis of previous assertions on this thread. The melees move forward, the illithids fire, the mages cast... for several rounds it's all working well. Then, I'm guessing, what happens is that in the final rounds of the battle, as the spell ques empty, the battle has been won, the enemies are all in flight... and at this point there's no more targest for the spell casters. So they all run and bunch up behind the illithids. The earlier posters on this thread had said this was because I was using the 'stay behind troops' command, remember, so I was testing this and not using the command, and it happens anyway.

I can't believe I'm the only one that finds this beyond annoying. It's absolutely senseless. And fixing it wouldn't require any radical expansion of the AI engine so far as I can see. Simply disabling that forced 'stay behind troops' command you hypothesise (and it matches my observations) would pretty much do the trick. There's no need for mages to be moving around and casting spells when all enemy forces have broken anyway (and that's always or very nearly always when this happens - AFTER all enemies have broken.)

Valandil
August 4th, 2006, 08:55 PM
actually, I've never had a problem with this, and I play ry'leh. Order your mages to ALL cast BoW, and then cast, and, since mind burn is absurdly long-ranged, they should never move.

shovah
August 5th, 2006, 09:41 AM
who says he will only be using mind burn? seems a waste of such expensive mages.

Folket
August 5th, 2006, 10:06 AM
He means that the mages will be able to cast mind burn when the enemy is runing away becuase of the range, so your mages will not default to stay behind troops as they do when they can't cast any useful spell.

Arralen
August 5th, 2006, 04:49 PM
.. make that "can't cast any spell". They will cast even the most unuseful spell before they start moving.

Btw. - to correct this whole issue is obviously not trivial, or the devs would have done that long before.

There's no 'do nothing order' - even commanders set to "stay behind troops" wiggle about a bit each turn. So a mage who cannot cast any spell (no matter who silly it may be) can't simply stand still for one turn - he has to move and defaults to "stay behind rearmost" because of this.

Folket
August 6th, 2006, 08:51 AM
Are you sure a fire mage being out of range will cast fireflies after the opponent?

Arralen
August 6th, 2006, 09:38 AM
Nope. He can't cast any spell, because he has no target in range to cast fireflies at. Dunno where I should have said otherwise.

He will cast at a single inf unit at max range despite the to-hit chance must be around 0.1%, though.

Likewise, he'll cast any other spell that has a target that is in range - preferably buffs spells, because those have range "self" or "1".

Only if he runs totally out of castable spells he'll start to move.

Agrajag
August 6th, 2006, 04:27 PM
Here's what I want to know, how come everytime someone remarks that an AI is stupid he says Artificial Stupidity in mock of Artificial Intelligence. Why not Genuine Stupidity which is a much more fitting title. (IE the AI is genuinely stupid)

Wick
August 6th, 2006, 07:09 PM
Because computers are stupid but for idiocy on a grand scale you need a human. ;-)

Gandalf Parker
August 7th, 2006, 10:30 AM
Actually neither term should be used. Artificial Intelligence isnt all that hard and has been done often. Its a common topic in the AI work groups. In those groups they joke about AI always meaning almost implemented because anytime an AI project is completed it stops being referred to as AI. Like diagnosis software, or autopilot, or streetlight management. All were AI projects until they were done.

What the players want is artificially human. As just one example... artificial intelligence would use a straight line. It takes a artificially human would sneak around behind or do a flanking attack. Or better yet to randomly select between them, sometimes when its even a bad idea but might be a surprise.

What programmers, especially game programmers have learned is that the closer you get to AH the more random gets worked in. So its often a shorter project that instead of starting with intelligent then gradually add more randoms, AH is often best achieved by starting with it being random then making it gradually more intelligent.
Just FYI trivia

Graeme Dice
August 8th, 2006, 02:29 PM
Arker said:
Actually, that's the point, they DON'T just do what they're ordered to do.



No, they do exactly what they are ordered to do, which is exactly as has been described to you. It doesn't necessarily match with what you think you ordered them to do, but that's a problem with your understanding of the logic used for decision making.


I did expect, however, an AI that with some basic, fairly simple logic checks like 'don't cast this spell if it's more likely to harm your own troops than the enemy' or 'don't move close to friendly troops when you have an area of damage effect following you around.' That seems pretty elementary.



If it's simple, then you'll be able to tell us all how to determine both cases. How does the AI know whether breath of winter is more likely to hurt your troops than the enemies?


Yep. It will do that even if not given that command.



No, it will only move forwards if there are no other possible spells to cast, or if you didn't give your commanders any orders at all. Stay behind troops is the default order.


Sounds right. And easily fixed, by anyone that has access to the code, one would think.



If it's easily fixed, then you'll be able to describe exactly what the proper behaviour would be in each possible game situation.


(A random R'lyeh hero, Traitor King? or something like that, L4 water mage with heroic quickness and the spell quickness as well. Definitely NOT ordered to attack or stay behind troops.



I already pointed out in my first post to this thread tat heroic quickness screws things up. Perhaps you should have realized that when somebody refers to heroic quickness, they are explaining that your units with heroic quickness fall under that category.

Ballbarian
August 8th, 2006, 03:15 PM
I've mentioned this before, but a little check box next to researched spells in the spell list would solve most of these types of issues. If the check box is unchecked, then God has outlawed this spell (removed from castable spells list for this nation as far as the mage's ai is concerned) and any mage caught casting that spell will be stoned to death. http://forum.shrapnelgames.com/images/smilies/happy.gif

Cainehill
August 8th, 2006, 03:26 PM
Ballbarian said:
I've mentioned this before, but a little check box next to researched spells in the spell list would solve most of these types of issues. If the check box is unchecked, then God has outlawed this spell (removed from castable spells list for this nation as far as the mage's ai is concerned) and any mage caught casting that spell will be stoned to death. http://forum.shrapnelgames.com/images/smilies/happy.gif



But of course, only _after_ the offending sage's BoW has killed all the expensive mages. http://forum.shrapnelgames.com/images/smilies/wink.gif

Arker
September 1st, 2006, 07:29 AM
Graeme Dice said:

No, they do exactly what they are ordered to do, which is exactly as has been described to you. It doesn't necessarily match with what you think you ordered them to do, but that's a problem with your understanding of the logic used for decision making.



I don't see how you could say that with a straight face, knowing that there are several final orders that can be given, yet as discussed in this thread the AI always uses 'stay behind troops' even when it's not the order given?


If it's simple, then you'll be able to tell us all how to determine both cases. How does the AI know whether breath of winter is more likely to hurt your troops than the enemies?



Using an expert system approach, you'd have a few rules like these:

1. Is the enemy Jotunheim? If yes, forget BoW, it won't do you any good.

2. Are you Jotunheim? If no, it's very likely going to decimate your own lines. Don't cast.

That wouldn't be the most sophisticated set of rules, but it would still be much more sophisticated than what it does now. Depending on the amount of information the programmers want to make available to the AI, more sophisticated chains of tests could be devised of course, but even the crudity above would work.

An even simpler fix would be simply to take this out of the list of spells the AI will cast unbidden. This is the type of spell that, if you're going to cast it at all, you should probably be casting early in the battle and following with a command to engage in melee anyway. Which, I gather from another post, is actually what they're doing with Dom3.

Graeme Dice
September 1st, 2006, 01:24 PM
Arker said:
I don't see how you could say that with a straight face, knowing that there are several final orders that can be given, yet as discussed in this thread the AI always uses 'stay behind troops' even when it's not the order given?



I'm telling you that because you are wrong when you claim that the AI always uses stay behind troops. You are also wrong when you claim that the AI sometimes chooses to use stay behind troops when you haven't selected it. I'm also not sure why you continue to ignore the fact that the behavious has been completely explained to you. I suppose it's because you can't be bothered to realize that when somebody tells you that heroic quickness can cause buggy behaviour when stacked with spell quickness, and you have a unit that has both heroic quickness and spell quickness, that it's obviously something else that must be causing the strange behaviour and you should continue to claim that some other bug is causing the problem.


2. Are you Jotunheim? If no, it's very likely going to decimate your own lines. Don't cast.



So you propose that such a list be made for every single spell? Let's look at the other situations where it would be useful. Caelum, for example, or the undead Ermor themes, or miasma C'Tis.

Arker
September 1st, 2006, 10:31 PM
Graeme Dice said:
I'm telling you that because you are wrong when you claim that the AI always uses stay behind troops. You are also wrong when you claim that the AI sometimes chooses to use stay behind troops when you haven't selected it.



Actually, that wasn't my claim. Several people, YOU INCLUDED, claimed that, and I concluded it fit what I was seeing and decided to believe you.


I'm also not sure why you continue to ignore the fact that the behavious has been completely explained to you. I suppose it's because you can't be bothered to realize that when somebody tells you that heroic quickness can cause buggy behaviour when stacked with spell quickness, and you have a unit that has both heroic quickness and spell quickness, that it's obviously something else that must be causing the strange behaviour and you should continue to claim that some other bug is causing the problem.



You know, I ignored your last crack about this for a reason. You're out of your freakin skull. Learn to read.

The poster that mentioned heroic quickness got a reply back from me saying that YES the particular character in the particular incidence I mentioned had heroic quickness, and I AGREED that explained that particular bit of strangeness I had reported. And you want to rant and rave for two posts now about me denying what I actually confirmed? You're just making yourself look like an idiot.


So you propose that such a list be made for every single spell? Let's look at the other situations where it would be useful. Caelum, for example, or the undead Ermor themes, or miasma C'Tis.



First, what other spells are causing problems? This is the only one that seems to give me much grief, although I imagine there are one or two others that would have the potential, I don't know what they are offhand. As I said already, if the programmers were motivated they could make some very sophisticated logic for this, but that's not necessary, a very simple fix (such as the one I understand has been implemented for Dom3) is also possible, so one shouldn't use the difficulty of implementing a very sophisticated fix as an excuse not to implement any fix at all. The simple one should still be a huge improvement.

I don't play Caelum that much, but it's my understanding that human players normally use lots of non-national troops, so it would probably be as big a problem for them as for anyone else. C'tis, again, I only played once, but I remember it being a *mixture* of troops, so again it's not safe to cast. Even Jotunheim, in 2 of 3 themes, is likely to have troops that are vulnerable to it, in fact.

Ashen Empire and Soul Gate could probably get away with it though. So sure, s/Jotunheim/"Jotunheim, AE, or SG" that's fine. Or even simpler, just don't cast the thing unbidden. A human player is going to script it when he wants it, casting it otherwise is almost never going to be any advantage, and very often will be a problem. And for AI players, the same thing really applies, they aren't smart enough to use it, it's far more likely to harm them than to help them.

Gandalf Parker
September 1st, 2006, 11:34 PM
If the programmers were motivated? You might have meant that casually but Ive seen rants on that which grated me.
There is one programmer and he has kept up a steady stream of improvments.

Also he has said in the past that he would look at flowcharts if someone wanted to psuedo-code something. That doesnt seem uninterested.

Arker
September 1st, 2006, 11:51 PM
One? I had the impression it was two, but I assume you're in a much better position to know than I.

Either way, the game is a monumental accomplishment, and if you read anything I've written as a slight to that work or those who did it, I've been misunderstood.

All I meant by that is that, while I'd be happy to write out pseudocode for a much more sophisticated chain of logic there, it seems pointless to do so. It's my impression it would require considerably more work than is realistically going to be put into it to actually implement (as it would require the AI to have access to a lot of information it apparently does not currently have access to, and probably logic it wasn't designed to cope with at all.) Particularly considering it's not really necessary, as the much simpler solution should have almost exactly the same effect.

Gandalf Parker
September 2nd, 2006, 11:21 AM
Arker said:
One? I had the impression it was two, but I assume you're in a much better position to know than I.



In a way there are. There are two developers. Johan is a programmer by trade and he makes it all work. Kristoffer teaches religion and mythology. He comes up with the nations, units, spells, equipment. He designs the stuff that Johan has to make work. They probably overlap abit.

Those are my own impressions of what Ive read and I sincerely hope Im not insulting either of them.

Graeme Dice
September 2nd, 2006, 11:57 AM
Arker said:
Actually, that wasn't my claim. Several people, YOU INCLUDED, claimed that, and I concluded it fit what I was seeing and decided to believe you.



If you decided to believe us, then you wouldn't have just told me that you didn't believe us, and that there was still unexplained behaviour.

"I don't see how you could say that with a straight face, knowing that there are several final orders that can be given, yet as discussed in this thread the AI always uses 'stay behind troops' even when it's not the order given?"

You are simply wrong when you claim that the AI uses stay behind troops when not ordered to. If you think that's what the people who understand the game are telling you, then you need to go back and re-read their posts.


The poster that mentioned heroic quickness got a reply back from me saying that YES the particular character in the particular incidence I mentioned had heroic quickness, and I AGREED that explained that particular bit of strangeness I had reported. And you want to rant and rave for two posts now about me denying what I actually confirmed? You're just making yourself look like an idiot.



If you actually agreed that heroic quickness caused the AI glitch (which it obviously did), then you wouldn't have just told me that the AI randomly picks different orders from what you tell it to do. If you want to be believed, then perhaps you should provide a battle replay where a mage that doesn't have heroic quickness disobeys your final order.


First, what other spells are causing problems? This is the only one that seems to give me much grief, although I imagine there are one or two others that would have the potential, I don't know what they are offhand.



Then perhaps you need to play the game more so that you understand the various rules. A spell as common as blade wind can cause more damage to your troops than breath of winter.


C'tis, again, I only played once, but I remember it being a *mixture* of troops, so again it's not safe to cast.



It might be a mixture of troops, it might also be nothing but marhsmasters and undead. It's definetly safe to cast in the undead only case, which is by far the more common of the two.


Even Jotunheim, in 2 of 3 themes, is likely to have troops that are vulnerable to it, in fact.



Well, except for the fact that nobody would use those troops.

capnq
September 2nd, 2006, 12:46 PM
Graeme Dice said:

Arker said:

Even Jotunheim, in 2 of 3 themes, is likely to have troops that are vulnerable to it, in fact.



Well, except for the fact that nobody would use those troops.



I doubt that there is a type of troop in the game that no one will ever try to use.

Strider
September 2nd, 2006, 02:01 PM
Cut out the personal attacks, people. Think it through before posting further comments.

Arker
September 2nd, 2006, 02:25 PM
Graeme Dice said:
If you decided to believe us, then you wouldn't have just told me that you didn't believe us, and that there was still unexplained behaviour.



Like I said, learn to read. I didn't say the former, and the unexplained behaviour was explained long ago. It's still stupid but it's not unexplained.


"I don't see how you could say that with a straight face, knowing that there are several final orders that can be given, yet as discussed in this thread the AI always uses 'stay behind troops' even when it's not the order given?"

You are simply wrong when you claim that the AI uses stay behind troops when not ordered to. If you think that's what the people who understand the game are telling you, then you need to go back and re-read their posts.



Both you and Arralen have explained in detail exactly the conditions under which the AI ignores other orders and goes to 'stay behind troops' as the "default order" whether it's given or not. And your explanations, as I've said already, match my observations. Perhaps you should do some re-reading, or chug a pot of coffee, or something?



If you actually agreed that heroic quickness caused the AI glitch (which it obviously did), then you wouldn't have just told me that the AI randomly picks different orders from what you tell it to do. If you want to be believed, then perhaps you should provide a battle replay where a mage that doesn't have heroic quickness disobeys your final order.



The heroic quickness glitch explains the one instance I observed where the commander, with final order 'cast spells,' moved behind the rearmost troops and *then* cast BoW. Without it, he would cast, then move. The end result is the same. The basic problem exists with or without quickness. If you think Quickness explains more than that you certainly haven't explained what. Why on earth you're on about me needing to provide a replay to show behaviour you already explained I don't know.

The AI suffers from a chronic need to 'do something' every turn, and if there's nothing useful for it to do it will cast useless or harmful spells, and when it can't even do that it will 'stay behind troops' - exactly as you've explained, exactly as I've observed many times, exactly as anyone that's played this game very much will have seen.


A spell as common as blade wind can cause more damage to your troops than breath of winter.



Blade wind will also, in the course of a game, do tons of damage to the enemy. It's cast at the enemy, and occasionally hits friendly troops by accident. That's part of the design of the game. I don't think anyone minds that, although of course IF you go into battle with a horde of low AC troops, against heavily armoured opponents, it could become a very stupid spell to cast. And yes, it would be nice if the AI was smart enough to know that. And, in fact, the AI seems to be *almost* that smart, in that it seems to pick other, more suitable spells instead of blade wind when facing heavily armoured troops - the only obvious improvement there would be if it were smart enough to recognise the cases where it's better off casting nothing at all. In the course of a game blade wind will do a lot more damage to the enemy than your own troops, despite the occasional friendly fire casualty.

This bears only the slightest resemblance to the BoW situation, which is likely to cause friendly fire casualties AND very unlikely to cause any damage to the enemy, when cast by the AI.

BigJMoney
September 2nd, 2006, 10:05 PM
I keep noticing you say that there isn't the need to add all the extra logic checking about battlefield dangers because it can be done much easier, but I've missed what that easier solution is. I'm not sure if it was in an early post somewhere or not, but tell again, if I've missed it, what the simpler solution is that you suggested because this thread just doesn't read correctly without it, lol. I like someone's suggestion to be able to ban spells from a world list or also a specific caster's list.

Quick question from your OP: does the trident do anything other than what its stats list? It doesn't make the user immortal or anything, does it? I've noticed that most weapons I give my SCs are better than the trident, so I never figured out why I even care to compete in that silly tournament.

Finally, I'll add that I sympathize. I know if it were me, I'd be just as mad and I think you have every justification. I can't believe others aren't being more empathetic themselves. Here you are trying to work your strategy and there seems to be no way out of this stupid trap -- not even a jury-rigged workaround. I believe one person actually suggested you not research an entire branch of magic. :-/ I'm glad you didn't respond to that.

=$= Big J Money =$=

Frostmourne27
September 2nd, 2006, 11:53 PM
With regards to his solution, I believe it was basically check if nation is cold immune/mostly cold immune (undead ermors, caelum, jotun etc) and make the spell uncastable, except when a human scripts it. The trident IIRC give 50% quickness (as per water 9 bless) so its not too bad, especially on casters. I get particularly annoyed when comp ermor wins tournament with a dusk elder, it makes them into army generating power horses. Nothing an SC wont beat, but powerful nonetheless. Also, I second Strider's request for a bit of calm here, lets not get like so many other cheesy gaming forums out there.

BigJMoney
September 3rd, 2006, 12:16 AM
Good point. What's done is done, and the community here is important. It doesn't matter whether we agree with his points or not, although I maintain my symapthy with the frustration.

I think it's time for me to do some forum searches on 50% quickness. I don't understand how something can be 50% quickened. /:-| I always thought it was a status; double or nothing.

=$=

Arker
September 3rd, 2006, 12:31 AM
BigJMoney said:
I keep noticing you say that there isn't the need to add all the extra logic checking about battlefield dangers because it can be done much easier, but I've missed what that easier solution is. I'm not sure if it was in an early post somewhere or not, but tell again, if I've missed it, what the simpler solution is that you suggested because this thread just doesn't read correctly without it, lol. I like someone's suggestion to be able to ban spells from a world list or also a specific caster's list.



Well of course if you define the topic at it's broadest - "the AI is rather dumb' - there isn't any simple solution, no. Making the AI smarter in general would be complicated, I'm sure.

But if you look at it from a narrower point of view, chop the topic up a little, there are simple solutions. The biggest problem, for me, is BoW. The simple solution is simply to remove that spell from the list of spells the AI will cast unbidden. Which I gather is exactly what has been done in Dom3.

Another example I've run into is nature mages insistently casting 'protection' on all your troops, when you either are, or are fighting against, Abysia. Of course, Abysian troops radiate heat, and protection raises AC but lowers fire resistance... VERY poor trade when facing fire radiating troops, even WORSE trade when your own troops are radiating. Obviously susceptible to the same fix - just removing it from the list of spells the AI will cast on its own immediately ends the problem.

Another player had the AI cast 'vortex of returning' for him. Hilarious... if it's not your army that just got recalled from a battle they should have won to the other side of the map http://forum.shrapnelgames.com/images/smilies/wink.gif Again, the simplest solution is to remove the spell from the AIs list.

It can be objected that this gives the human player another advantage in the game, but I don't think it does, because the AI is just as bad at determining when to cast these spells for the computer players as for the human ones. The best I understand, it really *doesn't* have any logic for determining these cases at all, just a random number generator and a preference for casting spells that do direct damage. That preference keeps spells like this from being cast very often, but it doesn't do anything to cause them to get cast when they'll actually help, or not cast when they wont, it just keeps them from being cast at all most of the time.

In fact, this is why I think yanking these spells entirely is probably a better idea than introducing a user interface to ban certain spells. That, I'm afraid, WOULD favour the human over the AI.

For some spells the case for removal is better than others, though. BoW obviously is more likely to harm the AI than to help it, but that is probably not true of protection. Heat-radiating units are much more rare than non-cold-immune units. Still, it's just as bloody annoying to the human player when it's cast at the wrong time.

So, yes, a more sophisticated AI would be wonderful, but realistically, when simply yanking the spells off the cast list has the same effect in most cases, where is the motivation to go to a lot more work for almost the same result? That's what I meant by mentioning motivation. It's perfectly reasonable if you can make a good enough fix in 1 hour or a *slightly* better fix in 1,000, you're going to have trouble finding motivation to go the extra mile, particularly when there are other things you could be working on instead. At any rate, simple solution NOW, more complex solution later, when/if it can be done, seems like the sensible thing to me.

Another fairly simple thing that could be done would be to recode what the AI does with a mage when it has no targets in range. Apparently what it does now is go to 'stay behind troops.' This is not a very smart thing to do at all, as it results in a bunch of leaders piling up right behind the rearmost unit of troops for no real reason. Since it usually happens late in battle, it's not as big a problem as it could be, but it's still definitely in the category of 'not-smart' and REALLY becomes a problem when one of those mages has BoW. I would suggest a mage set on 'cast spells' should, when no enemies are in range, 'advance' until some enemies ARE in range, and then resume casting, instead. It would also be good if troops NOT seeking melee (i.e. on 'fire' or 'cast spells' orders in particular) would advance directly forward, maintaining their position in relation to the top and bottom of the battle map. Currently, for instance, archers set on 'fire' will advance towards the nearest enemy when they cannot fire, so they move forward and left or forward and right, resulting in more troop clumping.

This *might* solve the problem with BoW for human players without removing it from the cast list, as it would make some of the responses I got early in this thread actually make sense - the big problem isn't so much just that the mages cast BoW, but that they typically follow up on that by all clumping together right behind the archers. I'm not sure, but honestly I think it would be a good idea either way. I'm sure it's less simple to do than removing some spells from the cast list too, but it shouldn't be nearly as complicated as a truly sophisticated AI spell-choice code.

Oh, whilst on the subject of battlefield movement, just in case the programmer-guy happens to read this, *please* make light cavalry worthwhile! This would only take a bit of added AI battlefield movement logic. As it is, each unit seems to only move in basically one direction - forward. Unless, of course, it routs. So light cavalry (and this applies to archers of all sorts too, but it's most damaging to light cavalry tactically) moves right up to range, and starts firing, but they never pull back, so they wind up in melee very quickly. Real life light-cavalry armies did a LOT of backward movement, this is what made them effective on the battlefield. They would move up into range, fire, then pull back to avoid melee.

The logic would go something like this:

1. Are we in range of target?
--->A. Yes. Is target within their movement radius of melee range?
------->I. Yes: Is our missile range greater than their movement range?
----------->a. Yes: Retreat to our maximum missile range, or the maximum range we can achieve while still retaining movement points to fire one volley, whichever is less, then fire.
----------->b. No: Stand and fire.
------->II. No: Stand and fire.
--->B. No. Advance our maximum movement, or to our maximum missile range, whichever is least. If movement points are left, fire.


Quick question from your OP: does the trident do anything other than what its stats list? It doesn't make the user immortal or anything, does it? I've noticed that most weapons I give my SCs are better than the trident, so I never figured out why I even care to compete in that silly tournament.



It's actually a fairly good weapon, I think. What are you making that's better? Remember it gives extra attacks. But no, I don't often find it worth entering the tournament for.


Finally, I'll add that I sympathize. I know if it were me, I'd be just as mad and I think you have every justification. I can't believe others aren't being more empathetic themselves. Here you are trying to work your strategy and there seems to be no way out of this stupid trap -- not even a jury-rigged workaround. I believe one person actually suggested you not research an entire branch of magic. :-/ I'm glad you didn't respond to that.



Thank you.

Frostmourne27
September 3rd, 2006, 01:26 AM
Somewhat OT, but 50% quickness (AFAIK) gives you 150% movepoints, and two actions each round. I don't THINK you get stat bonuses. It's very similar to what you get from heroic quickness. The quickness (or quickening) spells however, are a little bit different, you get double your move points, and two actions each round, as well as stat boosts. You also get 100 xp from winning the tournament.

Arker
September 3rd, 2006, 02:18 AM
Frostmourne27 said:
Somewhat OT, but 50% quickness (AFAIK) gives you 150% movepoints, and two actions each round. I don't THINK you get stat bonuses. It's very similar to what you get from heroic quickness. The quickness (or quickening) spells however, are a little bit different, you get double your move points, and two actions each round, as well as stat boosts. You also get 100 xp from winning the tournament.



IIRC it also raises your defence score. And, again IIRC, the trident has *three* attacks per round, base.

I usually skip the tournament anyway. It takes a high value commander that could usually be doing something else out of play, with a high risk of death. You can't change your script each fight, you have to use one for all the arena matches, which is pretty hugely annoying - for instance you script a bunch of casting to deal with Ermors champion, and as a result your champion stands there casting useless spells while another nations Pretender stomps him. Or you kill the enemy Pretender, thereby starting a war with a neighbor you really didn't want to be at war with, and then Ermors champion kills yours next round anyway http://forum.shrapnelgames.com/images/smilies/wink.gif

Eh, depends on the game too. I tend to play against AI on large maps with 12 or more enemies, entering the contest is a longshot there. In a smaller game, 3, 4 enemies, it can be a really good thing to do though. You might forge better weapons end-game, but that trident is *mighty* powerful on round 3...

Frostmourne27
September 3rd, 2006, 02:58 AM
I dunno about the tridents combat stats, i find there are better weapons for melee, but that its nice for casters. As for quickness and defence, i think you're right, but im not sure. Tournament i nice for death mages and other powerful mage nations. I wouldn't usually send a pretender of SC tho. As a side not, if you have no arms, you don't get the trident, and you aren't stuck doing all the subsequent tournaments, which is can be a bonus.

Daynarr
September 3rd, 2006, 04:40 AM
BigJMoney said:
I think it's time for me to do some forum searches on 50% quickness. I don't understand how something can be 50% quickened. /:-| I always thought it was a status; double or nothing.




Water 9 blessing gives 50% quickness. Blessed units get 50% movement points and attack twice every other round.

However Trident gives 2 attacks every round.

Arker
September 3rd, 2006, 01:52 PM
It may seem a little wierd to reply to myself, but I thought over what I've posted and have a bit to add. Editing the post this long after it's posted would be even wierder.


Arker said:
In fact, this is why I think yanking these spells entirely is probably a better idea than introducing a user interface to ban certain spells. That, I'm afraid, WOULD favour the human over the AI.



I think this stands up, but with some qualification. Simply adding the interface for humans and nothing else would be unbalancing - human players would milk this and the computer players would have no counter. BUT, if the computer players were given the same ability, and the logic to handle it, that would be very different. Trouble is the logic would probably be a LOT more work than the human interface. Bleah.


For some spells the case for removal is better than others, though. BoW obviously is more likely to harm the AI than to help it, but that is probably not true of protection. Heat-radiating units are much more rare than non-cold-immune units. Still, it's just as bloody annoying to the human player when it's cast at the wrong time.



After some thought, I'm not sure the case here actually IS weaker than for BoW. It's true that fire-radiating creatures are relatively rare, but fire magic in general certainly isn't. Protection *might* still be an advantage assuming random opponents, *but* it's very exploitable. I know if I see an army with nature mages in it, I make sure I've got as much fire magic as possible scripted when I attack. So even if it helps the computer players, on balance, when fighting each other, I'd bet it's a net loss for them against human players, which is really what counts.


1. Are we in range of target?
--->A. Yes. Is target within their movement radius of melee range?
------->I. Yes: Is our missile range greater than their movement range?
----------->a. Yes: Retreat to our maximum missile range, or the maximum range we can achieve while still retaining movement points to fire one volley, whichever is less, then fire.
----------->b. No: Stand and fire.
------->II. No: Stand and fire.
--->B. No. Advance our maximum movement, or to our maximum missile range, whichever is least. If movement points are left, fire.



This could even be simplified a bit and still work pretty much as it should. At its simplest, you'd remove all reference to the targets movement abilities, and simply try to stay at maximum range. It could also be made more sophisticated, for instance using an estimated optimum range instead of maximum, aiming to stay as close as possible without being drawn into melee. At any rate, I do think this line of thought is absolutely the key to making light cavalry playable, and it would improve the gameplay and the usefulness of all ranged-attack units. (And of course the AI should NOT be perfectly effective at this - it should definitely screw up sometimes, part of the charm of the game - but as it is it screws up every time, and the units are essentially useless beyond the ability of any modder to fix. This is a real shame, as these units are some of the more interesting in the game, for instance mounted Vanir, T'ien Ch'i cavalry, Centaurs... units that should really be very useful and fun, but aren't, because the combat AI just can't use them properly.)

KissBlade
September 3rd, 2006, 02:10 PM
Offtopic but, heh ... might I point out, while the TC cavalry has some parts to be desired, the Vanir and Centaurs are actually incredible units if you know what to do with them.

Arker
September 3rd, 2006, 03:04 PM
So what's your strategy then?

KissBlade
September 3rd, 2006, 03:25 PM
Duel bless them. Vans are incredible with f9/w9 (or even just w9 actually) and the centaurs are easily blessed to e9/n9 thanks to medusa.

Sindai
September 3rd, 2006, 03:52 PM
Ideally, every single battlefield spell would have a "desirability" function that observes factors about the battlefield situation and tries to gauge the usefulness of each spell. Then just cast the most useful spell.

This would take a lot of coding and testing but is probably not totally unfeasible, mostly because a lot of spells have basically the same behavior. For example, there's scads of projectile spells that all function in basically the same way. There's a lot of spells that hit the whole battlefield. There's a lot of personal protection spells. There's a lot of summoning spells. And so on and so forth.

Of course it's far too late to add this to Dom2 or even 3, but it's a nice pipedream.

Gandalf Parker
September 3rd, 2006, 04:50 PM
Thats actually in there already. I just dont understand it. Here is a combat log. It gives me a headache trying to follow what its thinking but it looks good. I know that "com " means commander so this seems to be the "thinking" for one commander named Bellare until the spell as chosen.
Gandalf Parker

Mrlreport (left): good0 broken0 autobreak0 turn0
com Bellare cast spell (favspell Summon Earthpower) (mayusegems 1)
est. choices 27
comp_castspell: eval Fire Flies result -1
comp_castspell: eval Air Shield result -1
spellscore, Freezing Touch score -9999 (boost 102 scorat 0)
Eval: Freezing Touch score 0 (fat 10)
comp_castspell: eval Freezing Touch result 0
best Flying Shards this far, 14 10 (4 pnts)
spellscore, Flying Shards score 13 (boost 83 scorat 0)
Eval: Flying Shards score 12 (fat 15)
comp_castspell: eval Flying Shards result 12
best spell so far Flying Shards (score12)
comp_castspell: eval Twist Fate result -1
comp_castspell: eval Hand of Dust result -1
spellscore, Sleep Touch score -9999 (boost 101 scorat 0)
Eval: Sleep Touch score 0 (fat 10)
comp_castspell: eval Sleep Touch result 0
comp_castspell: eval Bleed result -1
best Banishment this far, 3 6 (0 pnts)
spellscore, Banishment score 0 (boost 119 scorat 0)
Eval: Banishment score 0 (fat 0)
comp_castspell: eval Banishment result 0
best Blessing this far, 3 6 (35 pnts)
spellscore, Blessing score 35 (boost 114 scorat 0)
Eval: Blessing score 38 (fat 0)
comp_castspell: eval Blessing result 38
best spell so far Blessing (score38)
comp_castspell: eval Sermon of Courage result -1
comp_castspell: eval Smite Demon result -1
comp_castspell: eval Holy Avenger result -1
comp_castspell: eval Divine Blessing result -1
comp_castspell: eval Smite result -1
comp_castspell: eval Fanaticism result -1
comp_castspell: eval Word of Power result -1
comp_castspell: eval Burning Hands result -1
comp_castspell: eval Fire Darts result -1
comp_castspell: eval Flame Bolt result -1
comp_castspell: eval Shocking Grasp result -1
best Slime this far, 14 10 (5 pnts)
spellscore, Slime score 5 (boost 105 scorat 0)
Eval: Slime score 4 (fat 20)
comp_castspell: eval Slime result 4
looser spell Slime (score 4)
comp_castspell: eval Cold Bolt result -1
comp_castspell: eval Geyser result -2
comp_castspell: eval Acid Spray result -2
comp_castspell: eval Star Fires result -1
comp_castspell: eval Fire Resistance result -1
comp_castspell: eval Charge Body result -1
comp_castspell: eval Aim result -1
comp_castspell: eval Resist Lightning result -1
comp_castspell: eval False Fetters result -1
spellscore, Cold Resistance score 1 (boost 108 scorat 0)
Eval: Cold Resistance score 1 (fat 5)
comp_castspell: eval Cold Resistance result 1
looser spell Cold Resistance (score 1)
spellscore, Resist Fire score 1 (boost 81 scorat 0)
Eval: Resist Fire score 1 (fat 10)
comp_castspell: eval Resist Fire result 1
looser spell Resist Fire (score 1)
spellscore, Fists of Iron score -9999 (boost 84 scorat 0)
Eval: Fists of Iron score 0 (fat 15)
comp_castspell: eval Fists of Iron result 0
best Earth Grip this far, 14 10 (8 pnts)
spellscore, Earth Grip score 7 (boost 102 scorat 0)
Eval: Earth Grip score 7 (fat 5)
comp_castspell: eval Earth Grip result 7
looser spell Earth Grip (score 7)
best Earth Might this far, 3 6 (27 pnts)
spellscore, Earth Might score 29 (boost 100 scorat 0)
Eval: Earth Might score 26 (fat 20)
comp_castspell: eval Earth Might result 26
looser spell Earth Might (score 26)
comp_castspell: eval Hand of Death result -1
spellscore, Eagle Eyes score 135 (boost 114 scorat 0)
Eval: Eagle Eyes score 142 (fat 5)
comp_castspell: eval Eagle Eyes result 142
best spell so far Eagle Eyes (score142)
spellscore, Poison Touch score -9999 (boost 89 scorat 0)
Eval: Poison Touch score 0 (fat 5)
comp_castspell: eval Poison Touch result 0
spellscore, Resist Poison score 0 (boost 116 scorat 0)
Eval: Resist Poison score 0 (fat 5)
comp_castspell: eval Resist Poison result 0
spellscore, Barkskin score 293 (boost 116 scorat 0)
Eval: Barkskin score 308 (fat 5)
comp_castspell: eval Barkskin result 308
best spell so far Barkskin (score308)
comp_castspell: eval Personal Luck result -1
comp_castspell: eval Combustion result -1
comp_castspell: eval Phantasmal Warrior result -1
comp_castspell: eval Mirror Image result -1
spellscore, Quicken self score 878 (boost 98 scorat 0)
Eval: Quicken self score 798 (fat 20)
comp_castspell: eval Quicken self result 798
best spell so far Quicken self (score798)
spellscore, Stoneskin score 523 (boost 102 scorat 0)
Eval: Stoneskin score 550 (fat 5)
comp_castspell: eval Stoneskin result 550
looser spell Stoneskin (score 550)
best Armor of Achilles this far, 14 10 (5 pnts)
spellscore, Armor of Achilles score 1 (boost 86 scorat 0)
Eval: Armor of Achilles score 1 (fat 20)
comp_castspell: eval Armor of Achilles result 1
looser spell Armor of Achilles (score 1)
best Earth Meld this far, 3 6 (-35 pnts)
best Earth Meld this far, 14 10 (17 pnts)
spellscore, Earth Meld score 21 (boost 105 scorat 0)
Eval: Earth Meld score 12 (fat 80)
comp_castspell: eval Earth Meld result 12
looser spell Earth Meld (score 12)
comp_castspell: eval Immolation result -1
comp_castspell: eval Mistform result -1
comp_castspell: eval Ghost Wolves result -1
comp_castspell: eval Numbness result -1
spellscore, Ironskin score 1003 (boost 98 scorat 0)
Eval: Ironskin score 1003 (fat 10)
comp_castspell: eval Ironskin result 1003
best spell so far Ironskin (score1003)
best Protection this far, 3 6 (256 pnts)
spellscore, Protection score 293 (boost 116 scorat 0)
Eval: Protection score 266 (fat 20)
comp_castspell: eval Protection result 266
looser spell Protection (score 266)
comp_castspell: eval Body Ethereal result -1
comp_castspell: eval Spirit Curse result -1
best Tangle Vines this far, 14 10 (9 pnts)
spellscore, Tangle Vines score 11 (boost 116 scorat 0)
Eval: Tangle Vines score 10 (fat 20)
comp_castspell: eval Tangle Vines result 10
looser spell Tangle Vines (score 10)
comp_castspell: eval Summon Storm Power result -1
comp_castspell: eval Summon Water Power result -6
comp_castspell: eval Phoenix Power result -1
comp_castspell: eval Summon Lesser Fire Elemental result -1
comp_castspell: eval Summon Lesser Air Elemental result -1
Not enough vis for Summon Lesser Water Elemental
comp_castspell: eval Summon Lesser Water Elemental result -3
spellscore, Summon Earthpower score 203 (boost 83 scorat 0)
Eval: Summon Earthpower score 184 (fat 20)
comp_castspell: eval Summon Earthpower result 184
best spell so far Summon Earthpower (score100184)
Not enough vis for Summon Lesser Earth Elemental
comp_castspell: eval Summon Lesser Earth Elemental result -3
comp_castspell: eval Power of the Spheres result -1
comp_castspell: eval Desiccation result -1
comp_castspell: eval Farstrike result -2
comp_castspell: eval Blink result -1
comp_castspell: eval Returning result -1
comp_castspell: eval Communion Master result -1
comp_castspell: eval Communion Slave result -1
comp_castspell: eval Horror Mark result -1
comp_castspell: eval Dust to Dust result -1
comp_castspell: eval Decay result -1
comp_castspell: eval Frighten result -1
comp_castspell: eval Seven Year Fever result -2
comp_castspell: eval Curse result -2
comp_castspell: eval Bonds of Fire result -1
comp_castspell: eval Mind Burn result -1
comp_castspell: eval Berserkers result -1
comp_castspell: eval Sleep result -1
comp_castspell: eval Rage result -1
comp_castspell: eval Sailors' Death result -1
best Iron Will this far, 3 6 (13 pnts)
spellscore, Iron Will score 15 (boost 97 scorat 0)
Eval: Iron Will score 15 (fat 10)
comp_castspell: eval Iron Will result 15
looser spell Iron Will (score 15)
comp_castspell: eval Panic result -1
comp_castspell: eval Arcane Bolt result -1
best Mossbody this far, 3 6 (186 pnts)
spellscore, Mossbody score 160 (boost 85 scorat 0)
Eval: Mossbody score 145 (fat 20)
comp_castspell: eval Mossbody result 145
looser spell Mossbody (score 145)
castspell: cnr113 spl438 (Summon Earthpower) vis0 x3 y6 spldmg4096
vis 0 xvis 0
blastsqr: unr14799 x3 y6 aoe0 dmg4096 eff23 spc8404992 as10192 al9
affectvic vic14799 hv0
hitunit 14799 14799 dmg4096 spec8404992 ba2
battle_incheck

Arker
September 3rd, 2006, 05:36 PM
Sindai said:
Ideally, every single battlefield spell would have a "desirability" function that observes factors about the battlefield situation and tries to gauge the usefulness of each spell. Then just cast the most useful spell.



Actually I *don't* think this would be a good thing, at least not exactly as you explain it. Part of the charm of Dominions2 is the way combat is never completely controllable, things never go perfectly. There's a saying 'no battle plan ever survives contact with the enemy' and while it's exagerrated for effect, there's a lot of truth there. I wouldn't want the AI to always make the optimal move.

I like it being somewhat random. People in battle are somewhat random. They make decisions based on imperfect information, without the time to sit down and weigh all the factors and think it all through carefully, and that means a lot of the time they don't make the best choices.

But that said, yes, what would be good would be for it to have at least some grasp of these factors. Casting BoW then bunching up with a bunch of friendlies with no cold resistance is taking it way beyond realism into the land of the absurd, for instance. And that's what's so frustrating about it for me, I guess, it's not that I'm taking friendly fire (I take that all the time, naturally I don't like it, but in a wierd way it adds to the charm of the game... in cases where it makes sense.)


This would take a lot of coding and testing but is probably not totally unfeasible, mostly because a lot of spells have basically the same behavior. For example, there's scads of projectile spells that all function in basically the same way. There's a lot of spells that hit the whole battlefield. There's a lot of personal protection spells. There's a lot of summoning spells. And so on and so forth.



It's my understanding it *does* pretty much what you said, *in regards to those spells.* That is, direct damage spells, it tends to favour those, and it does seem to be calculating which spell is the best choice at that moment, to damage the enemy. For instance, Ulm Master Smiths will cast blade wind against light infantry, but switch to magma bolts for knights, which is exactly what they should do. I have a little problem with that part of the routine - for instance, when the knights are in position to hit my infantry next turn, and there's a big group of slingers waaaay back where they won't be able to do anything for a couple more turns (as if they could do anything to my troops anyway) I'd really like it if at least *part of the time* my smiths were smart enough to realise that the knights are what they need to be focusing on, instead of racking up meaningless kills on those slingers and letting the knights hit my line unscathed... even though it might be worse in terms of winning or losing, I'd rather see it cast the wrong spell but at the right target, i.e. hit the knights with blade wind. Stupid, but at least it would seem more tactically aware. If they'd hit em with the earth sink or whatever it's called, which does no damage at all, but locks em in place for a bit and lowers their defence... now *that* would be sweet AI.

Anyhow that's a comparitively minor point. They do seem to calculate what spell will do the most immediate damage to enemy troops and cast that.

The big problem is with the spells that don't fit that mold. The buffs and utily spells basically, tangle vines, aim, protection, blink (THAT one can really cause problems too) anything that doesn't do direct damage... the AI doesn't seem, from my observation, to have any underlying logic at all to when they cast those, they just seem to pull one out of the hat and throw it out like 'what the heck, I have absolutely no clue.' Throwing out fire clouds directly in the path of their own troops when the enemy is running away, for another example to add to the ones we've run into the ground http://forum.shrapnelgames.com/images/smilies/wink.gif I've seen that several times. Most of the time it does silly stuff it's not directly destructive though, and so you don't notice it unless you're paying close attention... casting strength buffs on other mages when those knights are charging the heavy infantry line for instance... watch closely for it and you'll see, but it's not even in the same league with BoW for annoyance, and as long as it doesn't happen too often it actually makes the thing seem more 'human,' no?

I don't want it to be perfect, but I do want it to simulate someone with something approximating a clue. And it really *does* do that most of the time. Just needs some work on certain spells...

Arker
September 3rd, 2006, 05:48 PM
KissBlade said:
Duel bless them. Vans are incredible with f9/w9 (or even just w9 actually) and the centaurs are easily blessed to e9/n9 thanks to medusa.



That's exactly what I thought you would say http://forum.shrapnelgames.com/images/smilies/wink.gif

What you're doing there, though, is giving up a huge number of creation points to supercharge them magically, and what you get out of it is really some quirky heavy cavalry.

I've played both of them that way, and experimented with the bless strategies, and my honest opinion is that while you can make them work that way, you'll still get your rear handed to you up against someone that spent those points on scales instead, and uses independent heavy cav/knight recruits.

But for the sake of argument, let's say you at least have a fair chance of winning that way. It's still beside the point to me - I'd like to be able to play it in role, that is to say, play them as light cavalry, using their speed advantage to bring those ranged weapons to bear effectively, not converting them into some oddball heavy-cav units.

And of course it's totally inapplicable to non-holy light cavalry, T'ien Ch'i, Arco... and actually not applicable to the Centaurs I was talking about either. You know, the Centaurs, not the White Centaurs, just plain old Centaurs, the ones that should logically by description and theme be the feared and fabled archery section of Pangaeas army, but in reality are totally useless http://forum.shrapnelgames.com/images/smilies/wink.gif

Arker
September 3rd, 2006, 05:57 PM
Thanks for posting this!

The 'favspell' entry is interesting, this might confirm my early hunch that each caster seemed to get a little fixated on their favourite spell... an impression I discarded as silly anthropomorphic imaging http://forum.shrapnelgames.com/images/smilies/wink.gif so maybe I was wrong on that.

I'd guess the -1, -2, -3 values are special disqualifications. I'd guess -1 is out of range to start with... and then we can discard that hypothesis almost immediately I think... I never use fire flies so I'm guessing, but doesn't it have at least as much range as flying shards? So it's something else... the really wierd thing is that it seems like it settled on mossbody(?) as the best choice, then cast summon earth power anyway... geez... gonna take some time to figure out how to parse this thing properly. Anyone already figured it out by chance? Might really clarify a lot...

Gandalf Parker
September 3rd, 2006, 08:14 PM
I probably should have mentioned in case someone is wondering. That was an excerpt from a Dom3 log.

Cainehill
September 4th, 2006, 01:27 AM
Arker said:

KissBlade said:
Duel bless them. Vans are incredible with f9/w9 (or even just w9 actually) and the centaurs are easily blessed to e9/n9 thanks to medusa.



That's exactly what I thought you would say http://forum.shrapnelgames.com/images/smilies/wink.gif

What you're doing there, though, is giving up a huge number of creation points to supercharge them magically, and what you get out of it is really some quirky heavy cavalry.

I've played both of them that way, and experimented with the bless strategies, and my honest opinion is that while you can make them work that way, you'll still get your rear handed to you up against someone that spent those points on scales instead, and uses independent heavy cav/knight recruits.

But for the sake of argument, let's say you at least have a fair chance of winning that way. It's still beside the point to me - I'd like to be able to play it in role, that is to say, play them as light cavalry, using their speed advantage to bring those ranged weapons to bear effectively, not converting them into some oddball heavy-cav units.

And of course it's totally inapplicable to non-holy light cavalry, T'ien Ch'i, Arco... and actually not applicable to the Centaurs I was talking about either. You know, the Centaurs, not the White Centaurs, just plain old Centaurs, the ones that should logically by description and theme be the feared and fabled archery section of Pangaeas army, but in reality are totally useless http://forum.shrapnelgames.com/images/smilies/wink.gif



_Sometimes_ the dual bless strategy gets smashed - sometimes it doesn't. That's one of the things that makes Dominions such a great game - no strategy wins all the time.

But : Vans and centaurs (all kinds) actually both have big advantages over other cavalry, even without an effective bless (though I can't think of a reason to skip the bless with them). Vans have mirror image, _and_ glamour : very stealthy, and completely avoiding some early damage. Then all the centaurs : decent HPs, and recuperating : they recover from wounds if they survive the battle.

Even if non-sacred, both would still be viable units, unlike a lot of the other cavalry/chariot units.

Arker
September 4th, 2006, 06:50 AM
Cainehill said:
But : Vans and centaurs (all kinds) actually both have big advantages over other cavalry, even without an effective bless (though I can't think of a reason to skip the bless with them). Vans have mirror image, _and_ glamour : very stealthy, and completely avoiding some early damage. Then all the centaurs : decent HPs, and recuperating : they recover from wounds if they survive the battle.

Even if non-sacred, both would still be viable units, unlike a lot of the other cavalry/chariot units.



I guess anything could be called viable if you want to play it badly enough, but one thing is for sure - you can't get them to use basic light cavalry tactics on the battlefield, it just won't happen.

According to the figures from Sunrays, you can recruit fifty longbowmen, sixty crossbowmen or woodsmen, or seventy four archers or tribal archers for the price of twenty centaur. They're good archers, but they aren't worth the money just for that. (Nor would it make any sense if they were, really. Light cavalry should be a good deal more expensive than foot archers.)

But the only other way the AI knows to play them is as heavy cavalry. As heavy cavalry, they're cheap, but woefully underpowered - they just aren't (and shouldn't be) capable of pulling that role off. The centaur cataphracts, on the other hand, fill that role quite well. Spend the same gold on a mix of centaur cataphracts and independent archers as you would on centaurs and you'll be a lot better off. Sure, the archers won't have recuperation, but they're cheap enough to replace, and most battle wounds don't really reduce their effectiveness anyway.

The Centaur warriors are even worse in this respect. You can get 80 independent light infantry for the same price as 20 centaur warriors, and it's a much better deal, as long as they don't know how to use their movement points.

Vanir have some great advantages, sure, like the centaur warriors their stats would be wonderful if the AI could just pull of basic light cavalry tactics with them. But since it doesn't they wind up being played as heavy cavalry, and they just don't have the punch to pull that role off against a serious opponent. Sure, they can chase down routers, but any unit can do that.

For the price of sixty Vanir, you can recruit seventy independent knights. The Van has thirteen hp, compared to twelve, but thirteen protection, compared to TWENTY. The knights have higher morale, more str and MUCH heavier weapons. Try it and see, the Van get chewed every time. Glamour is nice, but it gets knocked off real quickly when they engage in melee with heavy infantry, let alone knights.

Which is exactly as it should be, to that point. They're light cavalry, after all. Ultra-elite light cavalry, but light cavalry nonetheless. Light cav isn't meant for melee, and it isn't meant for shock charges. It's meant to wheel and dance just outside the enemies reach, while calling them rude names and showering them with volley after volley of projectiles. It's meant to screen the main army, obstructing the enemy but never engaging, keeping them off balance, harrassing them like a swarm of hornets, and in the end, running them down after they break formation and start to flee.

And the Vanir have the stats to do that job wonderfully. If only the AI was programmed to do it.

KissBlade
September 4th, 2006, 01:55 PM
Eh ... to say nothing about how 6 dual blessed centaurs will trounce 80 light infantry easily ...

The cost of Van to Knights are simply incorrect. Knights cost SIGNIFICANT resources and often that will stunt you way further than gold. Furthermore Vanir are sacred and really sacred units are meant to be blessed in some shape and form. Make no mistake, knights are probably the strongest recruitable basic troops but once their lances are exhausted, they're really not /that/ great. Meanwhile Vans can give SC's a run for their money easily and then say nothing of the extra magic res which is very very important. I really think you're not taking full advantage of either units here, and mind you they really don't need to be dual blessed. F9 is one of the cheapest blesses and considering that both units are stealthy, they are among the elites of raiding (Which dominates a major portion of most wars between dom II players). Someday you'll need to get on and duel one of us to really see this in application and you'll agree that Van and White Centaurs are just incredible units compared to indies. =)

Graeme Dice
September 4th, 2006, 08:43 PM
Arker said:
Like I said, learn to read. I didn't say the former, and the unexplained behaviour was explained long ago. It's still stupid but it's not unexplained.



Are you trying to pretend that you never wrote:
"I don't see how you could say that with a straight face, knowing that there are several final orders that can be given, yet as discussed in this thread the AI always uses 'stay behind troops' even when it's not the order given?"


And your explanations, as I've said already, match my observations. Perhaps you should do some re-reading, or chug a pot of coffee, or something?



Perhaps you should make up your mind whether you agree with us, or whether you actually agree with your statement that I've quoted above.


The heroic quickness glitch explains the one instance I observed where the commander, with final order 'cast spells,' moved behind the rearmost troops and *then* cast BoW. Without it, he would cast, then move.



Then he either had absolutely no spells to cast that would reach any enemy targets (as was already explained to you), or you are mistaken, and you left him on stay behind troops or with no final order.


Why on earth you're on about me needing to provide a replay to show behaviour you already explained I don't know.



I want a replay because you are claiming behaviour that does not happen in the actual game.


The AI suffers from a chronic need to 'do something' every turn, and if there's nothing useful for it to do it will cast useless or harmful spells, and when it can't even do that it will 'stay behind troops' - exactly as you've explained, exactly as I've observed many times, exactly as anyone that's played this game very much will have seen.



Why don't you make up your bloody mind whether you think that it's a bug or not when your mages move in various situations. The previous paragraph has you complaining that it's a bug when your mages change to stay behind troops for a turn. Now you're trying to tell me that it's not a bug.

Graeme Dice
September 4th, 2006, 08:49 PM
Arker said:
I've played both of them that way, and experimented with the bless strategies, and my honest opinion is that while you can make them work that way, you'll still get your rear handed to you up against someone that spent those points on scales instead, and uses independent heavy cav/knight recruits.



I really want to know why people think that a F9W9 Moloch with Order 3, Sloth 3, Cold 1, Death 1, Misfortune 3, balanced magic, and dominion 5 is going to be killed easily by a heavy scales pretender. You make a minor sacrifice of 6% gold for the sloth 3, 2% for the death 1 scale, and will research slightly slower, and get F9W9 blessed Vanii in return.

Graeme Dice
September 4th, 2006, 08:52 PM
Arker said:
The Centaur warriors are even worse in this respect. You can get 80 independent light infantry for the same price as 20 centaur warriors, and it's a much better deal, as long as they don't know how to use their movement points.



No, it's not a better deal. 20 centaur warriors will trash 80 light infantry with perhaps a half dozen losses. 20 centaur warriors will also trash 40-60 hoplites without severe losses

Wick
September 5th, 2006, 11:29 PM
In the combat simulator using 195 units in equalizer mode and with 200 Heavy Infantry (ind, 20 res) as the baseline 77 centaur warriors, 141 Hoplites, and 286 Carduces are equal utility.

The combat simulator assumes size = 0 and has a number of other quirks, nor is my sample random, so add salt to taste.

Gandalf Parker
September 6th, 2006, 12:31 AM
You know, using map commands you can add armies to yourself. You can make armies of your choice then play them with scripting and everything. Even do it over and over.

Wick
September 6th, 2006, 10:26 PM
I've done that too, but when trying to find the equal value levels for a lot of units it's just way too much work. At least, unless there's a way to run the game and get the battle results by script...

Gandalf Parker
September 6th, 2006, 10:47 PM
Hmmmmm actually might. Im not sure what info gets put into the log by the debug command. So we have start-game, and end-numbers (maybe). But the combat would require keypresses. Maybe with a macro...

Gandalf Parker
September 6th, 2006, 10:48 PM
You could also increase your statistics by creating many neighboring provinces and setting it up to have the same combat in all of them.

Wick
September 7th, 2006, 10:12 PM
My default test map is an 8x8 grid and I usually use four battles, alternating attacker and defender.

I thought about using the debug log and it looks possible but hard. I don't have any idea about running the battles though so I'm letting it slide. How would you do a macro?

Gandalf Parker
September 7th, 2006, 11:36 PM
a third party macro software that will record mouse movements, clicks, and keypresses, then play them back. For repeating the same thing over and over it can be handy

JPSeraph
September 8th, 2006, 04:59 AM
How does the AI scale in this game? Is Easy simply resource deprived or is it also less intelligent than Normal, for instance.

And I gather that the two Hard difficulties mostly benefit from improved resources rather than improved strategy in comparison to Normal or is this incorrect?

I like to play SP versus an AI that is intelligent but not one loaded with resources as my playing style is somewhat ponderous.

So far I've found that Normal difficulty is maybe a little faster than I'd like, but if the AI on Easy isn't going to avail itself of the same strategies as Normal, then I'd rather stick with Normal.

Endoperez
September 8th, 2006, 07:31 AM
The AI doesn't change its tactics AFAIK. However, if the games are too fast, you should try fiddling with the game settings a bit. Up the strength of independents to 6 or so, or play in a bigger map, or with fewer enemy nations. The first one, especially, can have a huge impact.

Gandalf Parker
September 8th, 2006, 11:39 AM
Changed AI setting changes the amount of points that it can spend on its pretender if I remember right.

JPSeraph
September 8th, 2006, 05:27 PM
Ok, so AI setting affects design points and either starting gold(?) or a #supplymult, #goldmult, #resourcemult, etc for individual nations?

As long as the AI isn't "dumber" on normal or easy, then I don't really mind. I just like SP games to start off very slowly before developing into all-out slaughter.

Endoperez
September 9th, 2006, 04:49 AM
Higher AI levels get more gold, supplies, resources and gems, IIRC. Higher levels of AI also get more design points. Normal AI has everything the same as humans, except possibly e.g. supplies to make up for the fact they easily starve their armies or other things they need because they aren't smart enough.

Daynarr
September 9th, 2006, 07:23 AM
I don't think they get more gems on higher levels. It may appear that way later in game since they will have more money to buy mages and thus search for sites more. But you can see in early turns its normal gem income.

I think they get more design points on higher levels as well.

Twan
September 9th, 2006, 07:34 AM
One thing is sure, impossible AIs are really fast... to cast the most destructive global enchantements.

In my last game against a pack of impossible AIs they have casted Burden of Time, Foul Air and Perpetual Storm before turn 30 and Wrath of the Sea some turns after, resulting in a nightmare situation... especially for them (the funniest being Jotunheim casting Burden of Time when they have access to the best mortal troops, and Man casting Perpetual Storm forcing their superior longbowmen to fight in melee).

As I play Machaka and don't have any astral mage, I'm condemned for the moment to watch AIs spoiling the world but, as they can't adapt their strategies for their own spells, I'm finally in a better position than most.

Graeme Dice
September 9th, 2006, 04:20 PM
Twan said:
In my last game against a pack of impossible AIs they have casted Burden of Time, Foul Air and Perpetual Storm before turn 30 and Wrath of the Sea some turns after, resulting in a nightmare situation... especially for them (the funniest being Jotunheim casting Burden of Time when they have access to the best mortal troops, and Man casting Perpetual Storm forcing their superior longbowmen to fight in melee).



I'd think that perpetual storm would actually benefit Man (or Pythium) greatly, as it removes the need to carry around a staff of storms to enable wrathful skies or Summon air power. Caelum wouldn't be able to fly around anymore, so they might not want it cast.

Twan
September 9th, 2006, 05:21 PM
Yes for an human it's not a bad choice, but the Man AI prefer to use hordes of archers unable to fire in the storm than Wrathful Sky (the Man AI mages seem to be mostly used for 3-4 Call of the Winds per turn and as many Swarm per fight, choices especially funny when birds can't fly).

Arker
September 11th, 2006, 11:22 AM
Graeme Dice said:
Then he either had absolutely no spells to cast that would reach any enemy targets



Exactly what I've been saying!



Why on earth you're on about me needing to provide a replay to show behaviour you already explained I don't know.



I want a replay because you are claiming behaviour that does not happen in the actual game.



And yet, immediately above, you are quoted recognising once again that this behaviour *does* happen. In the very same post. It's enough to make me seriously question your sincerity.


No, it's not a better deal. 20 centaur warriors will trash 80 light infantry with perhaps a half dozen losses. 20 centaur warriors will also trash 40-60 hoplites without severe losses



True, if you tried to make an entire army out of just the one unit. But that would be rather silly.

Use a heavy infantry core, then the archers as support. And yes, eighty light infantry will be more effective in that role than 20 centaur warriors.

Centaur warriors are good for making a fast, sneaking army to scout and take out poorly defended provinces. And I usually play CB, where they're much less expensive. Ludicrously cheap, actually. The point was just that, with a small improvement to the battlefield AI, they *and many other units* could be usable in a more realistic, thematically appropriate, and fun, manner. And modmakers wouldn't need to make them ludicrously cheap to get people to use them.

KissBlade
September 11th, 2006, 11:46 AM
Arker said:


True, if you tried to make an entire army out of just the one unit. But that would be rather silly.

Use a heavy infantry core, then the archers as support. And yes, eighty light infantry will be more effective in that role than 20 centaur warriors.




Eh ... no it really wouldn't. Assuming we're talkign about the typical e9/n9 white centaurs, I don't think you've ever seen what they're capable of. They literally don't die to anything short of heavy cav/knights.

Nerfix
September 11th, 2006, 12:01 PM
Try them out and then make your judgement.

Graeme Dice
September 11th, 2006, 01:30 PM
Arker said:
And yet, immediately above, you are quoted recognising once again that this behaviour *does* happen. In the very same post. It's enough to make me seriously question your sincerity.



I told you that units do exactly what they are ordered to, which is completely correct except in the case of heroic quickness, and that any deviation is actually you not understanding how the orders work. You responded by telling me that I shouldn't say that with a straight face because the AI "always uses 'stay behind troops' even when it's not the order given". Since the AI does not "always uses 'stay behind troops' even when it's not the order given" but instead only uses stay behind troops when no other order is possible, I told you that you were wrong as that's certainly not a bug. You seem to be confused about the difference between erroneous or buggy behaviour, which is what happens when a unit has heroic quickness, and fuly explained behaviour following the orders the commander has been given, which is what happens every other time.


True, if you tried to make an entire army out of just the one unit. But that would be rather silly.



Why would it be silly? Dominions does not reward mixing troop types like you seem to think it does. Add slow troops, and they will hardly even see combat unless the enemy is strong enough to kill all of your centaur warriors. If that's the case, then Pangaea doesn't have any troops that would be able to stand up to the enemy anyways.


Use a heavy infantry core, then the archers as support. And yes, eighty light infantry will be more effective in that role than 20 centaur warriors.



I think you need to actually try it out. 20 centaur warriors will easily kill 80 light infantry. I'd be surprised if they would take more than 5 losses. Pangaea has no heavy infantry worth mentioning, so you'd have trouble making such a mixed force anyways. It's not like you'd ever recruit a minotaur or a satyr. The minotaur will trample into the enemy, killing three or four troops, then get swarmed and kiled. The satyrs are either light infantry, which is almost universally useless in Dominions, or require large amounts of resources, and thus can't be built in sufficient quantities.


Centaur warriors are good for making a fast, sneaking army to scout and take out poorly defended provinces. And I usually play CB, where they're much less expensive.



They are also just about the best non-sacred national troop type available. Jotunheim is probably the only nation that has better troops. Centaur warriors have a good chance of killing an air queen with life draining weapons and quickness.


Ludicrously cheap, actually. The point was just that, with a small improvement to the battlefield AI, they *and many other units* could be usable in a more realistic, thematically appropriate, and fun, manner.



Javelin equipped units already almost always throw two or three javelins if you put them on fire closest, so I'm not sure what your complaint is about javelin equipped troops.

Arker
September 11th, 2006, 02:48 PM
KissBlade said:
Eh ... no it really wouldn't. Assuming we're talkign about the typical e9/n9 white centaurs



Wrong assumption. We were talking about standard centaur warriors. Similar to the white centaurs, but cheaper and not sacred.

Arker
September 11th, 2006, 03:11 PM
Graeme Dice said:
I told you that units do exactly what they are ordered to, which is completely correct except in the case of heroic quickness, and that any deviation is actually you not understanding how the orders work. You responded by telling me that I shouldn't say that with a straight face because the AI "always uses 'stay behind troops' even when it's not the order given". Since the AI does not "always uses 'stay behind troops' even when it's not the order given" but instead only uses stay behind troops when no other order is possible,



Nice backpedalling there.

So it turns out, contrary to your past flames, we *agree* that the behaviour I'm seeing is the behaviour that's programmed into the game, you just want to try and make a big deal out of the fact you don't like how I phrase it. And it takes you seven pages to admit that, and even now you're still obfuscating and trying to claim you were right all along.

That's pathetic. Really pathetic.


I told you that you were wrong as that's certainly not a bug.



I don't believe I ever said it was a bug, although you keep attributing that to me. I said it was annoying.


You seem to be confused about the difference between erroneous or buggy behaviour, which is what happens when a unit has heroic quickness, and fuly explained behaviour following the orders the commander has been given, which is what happens every other time.



Nope, I'm not confused about it at all, despite all your attempts to confuse me.

They're two different issues. They're both annoying. As I said a couple pages back...


Why would it be silly? Dominions does not reward mixing troop types like you seem to think it does. Add slow troops, and they will hardly even see combat unless the enemy is strong enough to kill all of your centaur warriors. If that's the case, then Pangaea doesn't have any troops that would be able to stand up to the enemy anyways.



Not at all. Pangæa has much heavier troops than the centaur warriors. Do you ever play them?



Use a heavy infantry core, then the archers as support. And yes, eighty light infantry will be more effective in that role than 20 centaur warriors.



I think you need to actually try it out.



I have.


20 centaur warriors will easily kill 80 light infantry. I'd be surprised if they would take more than 5 losses. Pangaea has no heavy infantry worth mentioning,



OK, now I know you've never played Pangæa!

That's got to be some kind of record for an absurd assertion!

Pangæa, for your information, has Minotaurs. And War Minotaurs. The latter may well be the heaviest infantry in the game.


It's not like you'd ever recruit a minotaur or a satyr. The minotaur will trample into the enemy, killing three or four troops, then get swarmed and kiled.



So you do realise they exist after all!

Well, yeah, if you send one out alone that'll happen. Twenty five or thirty of the suckers, with some cheap support, cut through enemy armies numbering into the hundreds like a hot knife through butter.


The satyrs are either light infantry, which is almost universally useless in Dominions, or require large amounts of resources, and thus can't be built in sufficient quantities.



Light infantry are less effective than they should be, which is why I suggested a modification to the AI that would improve the situation, yes, but if you think they're useless you're just not using them right.



Ludicrously cheap, actually. The point was just that, with a small improvement to the battlefield AI, they *and many other units* could be usable in a more realistic, thematically appropriate, and fun, manner.



Javelin equipped units already almost always throw two or three javelins if you put them on fire closest, so I'm not sure what your complaint is about javelin equipped troops.



IF you set them up on the battlefield right, right for Dominions that is, not right in any realistic sense, then yes, you can get that out of them. If they're foot troops, at least.

If they had retrograde movement, they could be used more realistically, and more effectively. Since it's possible to fudge it with foot troops, it's not such a big deal there - but horse archers are another matter.

Graeme Dice
September 11th, 2006, 04:26 PM
Arker said:
Nice backpedalling there.



Who's backpedalling? It's not my fault that you were the one who started this entire argument by telling me that "I can't believe you can say that with a straight face." I mean, you're the one who posted "Actually, that's the point, they DON'T just do what they're ordered to do." Now you're claiming that they _do_ act as ordered.


So it turns out, contrary to your past flames, we *agree* that the behaviour I'm seeing is the behaviour that's programmed into the game, you just want to try and make a big deal out of the fact you don't like how I phrase it.



I suggest that you might want to go back and re-read page 3. You'll note that you wrote "I don't see how you could say that with a straight face, knowing that there are several final orders that can be given, yet as discussed in this thread the AI always uses 'stay behind troops' even when it's not the order given?" And you might also want to go re-read page 2, where you state that "Actually, that's the point, they DON'T just do what they're ordered to do."


I don't believe I ever said it was a bug, although you keep attributing that to me. I said it was annoying.



"Actually, that's the point, they DON'T just do what they're ordered to do." If that's not calling something a bug, then nothing is.


Not at all. Pangaea has much heavier troops than the centaur warriors. Do you ever play them?



Yes, I've played Pangaea several times. Their only troops that could be called "heavy" are satyr hoplites, minotaurs, and centaur cataphracts. Hoplites are mediocre compared to centaur warriors. The same gold cost in centaurs would win easily. Cataphracts require huge numbers of resources, as do minotaur warriors.


I have.



Then you haven't run very many test games. I certainly hope you're not using Saber Cherry's combat simulator.


Pangaea, for your information, has Minotaurs. And War Minotaurs. The latter may well be the heaviest infantry in the game.



Both are useless for their gold and resource cost compared to centaur warriors.


So you do realise they exist after all!



I realize that they exist, I also realize that they aren't worth their cost. They might be worth it if they didn't have trample so that they would actually use their equipped weapons instead of fatiguing themselves after a half dozen turns.


Well, yeah, if you send one out alone that'll happen. Twenty five or thirty of the suckers, with some cheap support, cut through enemy armies numbering into the hundreds like a hot knife through butter.



In order to field 25-30 minotaurs in a reasonable amount of time, you'll need an order 3 scale, and probably a fortress or a castle. That puts your pretender at a 240 - 320 point disadvantage compared to the person who takes sloth 3, a watchtower, and recruits centaur warriors instead. Those minotaurs will be taken out very easily by any kind of battlefield magic.


Light infantry are less effective than they should be, which is why I suggested a modification to the AI that would improve the situation, yes, but if you think they're useless you're just not using them right.



If you spend 1000 gold on light infantry, and I spend 1000 gold on heavy infantry, you will end up losing, since I'll have about 75% as many troops as you, yet each one can kill two or three (or more) of yours.


IF you set them up on the battlefield right, right for Dominions that is, not right in any realistic sense, then yes, you can get that out of them. If they're foot troops, at least.



It's not hard to set them up right. You put them in a single block in the middle of the field on "fire closest".


If they had retrograde movement, they could be used more realistically, and more effectively. Since it's possible to fudge it with foot troops, it's not such a big deal there - but horse archers are another matter.



Since foot archers can reach across nearly the entire battlefield, what possible advantage do you hope to gain by using mounted archers? You aren't going to do significantly more damage by having them move closer, the combat model isn't detailed enough for that.

Arker
September 11th, 2006, 05:29 PM
Graeme Dice said:
Now you're claiming that they _do_ act as ordered.



Not at all. I'm claiming the same thing all along, and you have posted numerous times agreeing with it.

When there are no enemies in range of their spells, and no buffs are applicable, they go to 'stay behind troops.' Even when that's NOT the order they were given.


"Actually, that's the point, they DON'T just do what they're ordered to do." If that's not calling something a bug, then nothing is.



Not really. If it was intentionally designed to work that way, it wouldn't be a bug per se. It would still be annoying, and perhaps a design decision that, in retrospect, wasn't optimal. It might be a good thing to reconsider. That doesn't really make it a "bug."


Then you haven't run very many test games. I certainly hope you're not using Saber Cherry's combat simulator.



Never used it.

As far as the minotaurs vs centaurs thing, look, that's a pointless discussion. You use what you like, I'll use what I like.

But I think that IF what you were saying about the Minotaurs, and the entire light infantry line, were true, that would be every bit as close to a 'bug' as the problem with mages I started the thread about. What's the point in having different troop types if the combat engine is so silly that your best tactic is a big pile of one troop type in the centre of the screen?

Fortunately the combat engine isn't as bad, in my experience, as you seem to think. It does have some rough spots, some things that could be improved, of course.



Light infantry are less effective than they should be, which is why I suggested a modification to the AI that would improve the situation, yes, but if you think they're useless you're just not using them right.



If you spend 1000 gold on light infantry, and I spend 1000 gold on heavy infantry, you will end up losing, since I'll have about 75% as many troops as you, yet each one can kill two or three (or more) of yours.



I don't doubt that at all.

I wouldn't do that, though. I'd spend perhaps 700 on the heavies, the last 300 on the lights, and deploy them to screen or flank. I think I'd win, and again, if I wouldn't, that would seem to be a 'bug' at least by your standards.



IF you set them up on the battlefield right, right for Dominions that is, not right in any realistic sense, then yes, you can get that out of them. If they're foot troops, at least.



It's not hard to set them up right. You put them in a single block in the middle of the field on "fire closest".



If that's the extent of your tactical thinking, I guess a lot of this makes more sense now http://forum.shrapnelgames.com/images/smilies/wink.gif


Since foot archers can reach across nearly the entire battlefield, what possible advantage do you hope to gain by using mounted archers? You aren't going to do significantly more damage by having them move closer, the combat model isn't detailed enough for that.



No, actually, it's definitely not that bad. Archers hit a LOT more at close range than all the way across the battlefield.

They could also be used to 'pull' enemy squads into disadvantageous positions.

KissBlade
September 11th, 2006, 05:33 PM
Buying 25-30 minotaurs is just about the biggest waste of gold ever. Also I wonder wat you mean by "cheap support" since 25 - 30 minotaurs are very expensive. Also I assure you, unless you plan on your "cheap support" doing all the gruntwork, 25-30 minotaurs cannot wipe hundreds of troops. Plus at that stage of the game, the centaur's faster movements will prove much more useful at reaching the backline of casters that will wipe your minotaurs easily.

Graeme Dice
September 11th, 2006, 06:12 PM
Arker said:
But I think that IF what you were saying about the Minotaurs, and the entire light infantry line, were true, that would be every bit as close to a 'bug' as the problem with mages I started the thread about.



Bugs are things that don't work as designed, not problems with the design.


I wouldn't do that, though. I'd spend perhaps 700 on the heavies, the last 300 on the lights, and deploy them to screen or flank. I think I'd win, and again, if I wouldn't, that would seem to be a 'bug' at least by your standards.



I highly doubt that you'd win. The lights won't do anything to the heavies, and will be killed in a single hit each.


If that's the extent of your tactical thinking, I guess a lot of this makes more sense now



Tell you what. I'll set up a game on the battle simulation map where we can each send 1000 gold worth of non-sacred troops with no mage support into a province with neutral dominion. You can pick any nation, and any troops you want, then set them up however you want. I'll pick 1000 gold worth of purely heavy infantry from any nation that you don't want and stick them in a single block as far forwards as possible. We can see which wins the battle. It won't be the army that wasted gold on low protection troops.

KissBlade
September 11th, 2006, 06:52 PM
Don't forget the obligatory two or three advance militia as arrow catchers =).

Strider
September 11th, 2006, 08:46 PM
Moderators have sent PMs to some of the posters in this thread. It hasn't had enough of an effect. I'm locking this one up.

Mindi
September 11th, 2006, 10:19 PM
I’d like to make a statement here and I hope everyone is listening. When a moderator asks you to cut out some unacceptable behavior, you cut it out, it’s as simple as that. I don’t care who started it, but a moderator’s intervention should be the end of it.

We’re pretty relaxed around here, I don’t have to step in often and in general I think we have a pretty good community that knows how to act like adults and police their own posts. HOWEVER, I want to be clear here and say that the type of behavior posted in this thread will not be tolerated. Neither will childish behavior like “well he started it!” as an excuse for retaliation or taking an argument to another area of the forums after a thread lock. I will not hesitate to reward childish behavior with childish punishments, such as time outs! http://forum.shrapnelgames.com/images/smilies/mad.gif This is probably a good time for everyone to familiarize yourself with the Board Rules (http://forum.shrapnelgames.com/faq.php=) again. Consider this your reminder of the rules and a final warning.

We’re about to have an increase in traffic and community activity with the much anticipated release of Dominions 3. Lets remember to treat one another with respect so that these forums can be enjoyable for all. http://forum.shrapnelgames.com/images/smilies/smile.gif