Is the AI cheating?

[Didz]

I thought I'd start a new thread for this as having tried and failed three times to complete the Sapphire Archipelago mission I decided to try and find out once and for all whether the AI is cheating, or just amazingly lucky.

So, basically I've invoked de-bug mode and I'm playing the campaign again and watching to see exactly what the AI is doing and whether it is playing by the rules.

[BBB]

@ Didz, playing the campaign is *not* playing your standard game. The campaign is designed to tell a specific story in a specific manner.


It's a common error to think that each camapign map runs at the players leisure. The player is just one part of a story in these maps, and if you carefully read the briefing notes, at no point does it ever mention that you should be building a massive territory.



In the rmg, you can "turtle" quite well. However, disclaimer, what I call turtling is not what most people seem to think of turtling.


Turtling for me is being established and rolling that superiority onto your opponent, inviting them to waste their strength attacking you, and then destroying them. The UUSR in WW2 to me is this kind of turtling (grind down the enemy, then pushback and never let them recover).


Most people use turtling to mean hide away, leave me alone until I am ready to play the agme.


That kind of turtling never won a single war ever. On lower difficulty levels you can do that though, if the game experience you want is somethign more chilled out.


Actually, come to think of it, you can change the difficulty in the campaign anyway.

[Didz]

Messaggio originale di BBB:

In the rmg, you can "turtle" quite well. However, disclaimer, what I call turtling is not what most people seem to think of turtling.

I think you are missing the point.

The game is designed to favour blitzing, because everything is geared and reliant upon the number of cities one owns. Therefore, by design it is a one track railway, you can't opt for anything else. Thats the basic cause of the complaints about AI cheating, but it really isn't cheating just game design.

It's really that simple, which is why so many strategy gamers are getting frustrated with it. The people complaining tend to be people like me who prefer a slower paced game with more choice and diversity.

The USSR strategy you mention of sacrificing ground to over-extend your enemy would never work in Age of Wonder 3, for the simple reason being that sacrificing three cities to another faction merely makes them three cities stronger, and you three cities weaker.

Most strategy games counter-this by making newly acquired territory a drain on resources, so that the faster you expand the more difficult life becomes. AOW3 doesn't do that, so there is no penalty for rapid expansion. You capture a city and within a few turns it's producing for you.

At the start of the Sapphire Archipelago mission Isabella owns seven cities, some of which are already at metropolis level. Sundren owns none. So, within 16 turns she will lose the advantage of her past experience to Isabella, and by turn 50 she will be completely outclassed.

In my first play through I managed to absorb four cities, and capture one of Isabella's (Nuvina), but that isn't enough. Even with a 5/6 city ratio, Isalbella's cities are fully developed and your's are mostly villages. So, you quickly find yourself swamped with large armies of superior units, and beset by Incite Revolt spells which you have neither the mana nor casting points to disjunct. Game over.

It comes down to simple arithmatic in every game, and the only way to stay ahead is to blitz it.

[BBB]

Are you still talking about the campaign strategy or about the game strategy in general?

[Didz]

Messaggio originale di BBB:

Are you still talking about the campaign strategy or about the game strategy in general?

I'm talking about the game design. The campaign works the same way, but obviously in the early missions the scripts prevent the AI explioting things too early. For example on the previous mission it was obvious that the AI was being held back from attacking the goblin cities beyond the pass until you had captured all of them. That pretty much levelled the playing field before battle proper commenced.

[MobySlick]

I beat this mission first time. As someone mentioned, go for cities ASAP. Also it's better to have one powerful conquest group of 3 - 4 armies together, than to try to maintain multiple capable forces. Don't try to defend cities, just retake them, killing more enemy forces by doing so. Ignore any exploration that takes you off your direct path to the next enemy city UNTIL you surround her safely in her final city. Surround her with as many armies as it takes to prevent her from escaping. Then use a hero army to travel around and explore the entire map. This will help you in the next map by getting items you don't need that you will sell on turn 1 for instant cash bonus. Also it's a great way to level Bormac up to the limit if you got him. If you have Nomlik with the Divine Justicars ability then stick Bormac in his army for risk-free exploring. So to reiterate hit Isabella hard and fast, do not rest after taking a city, just proceed to the next one.

She does quite frrequently cast Incite Revolt on you, but keep disjuncting it. Therefore you need to have researched the techs that give you more casting points. Rely on your secondary heroes' casting pools for battles because you need the main pool (Sundren's) for disjucting. Try to build up many Grand Palaces to increase the main pool. Once you get Greater Disjunction you're in the clear. By the time you get her surrounded it's revenge time as you keep Incite Revolt, Shadow Thieves' Guild, and whatever else you want on her one city.

For city taking, you can always make them come out of the walls by killing 1-2 of their units with spells. While you cast spells arrange your army into a wide line. When they do come out, retreat as far back as possible maintaining formation until the next turn, and the enemy should then come close enough for you to get first strike. I use this strategy to do most city battles without losing a single unit. GL;HF.

[havok]

I follow what you are saying and agree with your AI assesment under the principle of the game being go go go. It does kind of contradict itself sense the game gives you stuff to lolligag around on.

I've always believed AI runs on pre concocted attitudes that work off of varying levels of many variables. Based on proximity, moral alignment, etc. So if a human player or other AI pushes these levels to certain boundaries the AI acts in certain ways. Also core directives. like they are greedy, war mongering, peaceful etc. These should be the things that cause wars.

For me turtling (if thats what we are calling it) is I just want some realistic game play. Where Wars are based on Land usage and complete and utter Moral differences. Not because I Know my ultimate goal is to destroy so i say screw enjoying it and blitz everyone. Which is actually a viable strategy for n AI to have at random. Especially evil ones like orcs or Goblins.

And what I think this posters underlying point is that it would enhance realism if there were ways to increase gold output without domination. Like creating an expensive laborer (or upgrade) who could dig a mine into a quary etc. (No i dont think this makes it like Civ) so that he can compete with the blitz everything strategy on his own terms.

My strategy for dealing with the AI's style is even becoming redundant for me as the games mount up. I typically explore early and try to build my Hero's into unstoppable dieties and their troops. Then steam roll the AI even when they had me on the ropes at the half way point.

And while one can say "There you go then, a solution to how the AI's directive works" the problem is the redundancy of it. and it could be solved by other AIs having varying styles.

Before i go too far though. I'm not sure the extent of everything they've done to AI this last run through but from what i could tell its seems to have more personality now. But with positives and negatives like stupid stuff like not picking up gold in their kingdom or not taking out structures within their borders.

[Didz]

Messaggio originale di MobySlick:

I beat this mission first time. As someone mentioned, go for cities ASAP.

Thats pretty much the conclusion I've come to, any time wasted on other activities is really just playing into the AI's hands as it will simply outpace you very quickly if you don't.

When I try this mission again I think I'm even going to ignore the city quests, as they are just a waste of time. If they don't open their gates immediately it's quicker just to steamroller them kill the garrisons and seize their cities by force, rather than muck about waiting for a bunch of dragons or something to wander into their city zone so you can earn their favor. It's redundant gameplay that doesn't earn you any benefit.

Once you've crippled the AI's development by capturing most of their cites you can afford to waste time wandering about exploring the rest of the map and doing quests etc.

What I think I'm going to do next is try out Gloweye's strategy linked above in debugged mode so I can see the AI's reaction.

Gloweye's strategy is pretty much the blitz tactic I think is necessary so it will be interesting to see how it compares with my first debugged game where the AI outpaced me within 16 turns.

[Didz]

Hi! Thought I'd post a quick update.

As stated above I started a new de-bugged mode playthrough following Gloweye's strategy, which was recommended by jaccovandorpGloweye above.

So, far it's going pretty well. I must admit Gloweye's recommendation that you charm, convert, or befriend all the units in your army rather than waste time and money recruiting them is brilliant. I would never have tried that normally, but it makes a huge difference as long as you don't might having a bit of a hotch-potch of units in your armies. I've even managed to charm a couple of bard units so my battles tend to be a major charm offensive now with the two fire giants acting as bouncers if someone doesn't accept the offer they can;t refuse.

The only problem I found around turn 10-14 was that this strategy was so successful I had trouble finding the money to pay my troops. However, I managed to scrape through by selling artifacts and now that I'm starting to take Isabella's cities my gold income is just about covering the costs.

So, far I've taken three of Isabella's cities (Nuvina; Portama and Craws), which means that she is down to four.

It's currently Turn 27, but unlike the last play-through Isabella hasn't managed to catch Sundren in terms of power. Last play through, when I was taking time to explore and do quests etc. Isabella caught up with Sundren in terms of casting points by Turn 16. This time she is still lagging behind on 50/50 and Sundren in currently on 70/70. So, it seems that taking her cities quickly does cripple her development.

Not only that but she has only managed to disjunct Sundren's Corpse Looting spell once on Turn 19, and Sundren had the CP to simply recast it immediately, and she hasn't tried again since. Sundren has been researching nothing but Cloak and Dagger since the start of the game. So I've really focussed on maximizing her casting points, which seems to have paid off.

Isabella has obviously been deviating somewhat as I notice that she has cast Iron Grip on all of her remaining cities. I've just ignored that as I don't plan to use Incite Revolt, in fact, I haven't even researched it. So, if she wants to waste mana on pointless spells it's up to her.

I also noticed during the attack on Craws that she has Festival of Evil running, but it didn't seem to help much I was still getting 60%-75% on my charms and conversions.

There are two more heavily defended cities on the current island (or forts at least) and a level 1 enemy hero. So, they are my next targets. Isabella hasn't even left her throne city yet, so apart from a few threatening messages she has left her minions to face the onslaught.

No sign of cheating from the AI, so far, and the blitz approach seems to be paying off.

[Gloweye]

Good to hear it's working out for you!

[Didz]

Messaggio originale di jaccovandorpGloweye:

Good to hear it's working out for you!

Well it was but I've got to turn 47 and something weird is going on now that I can;t explain.

Isabella is still sitting in her throne city reading books and generally trying to keep pace with Sundren's research rate.

Sundren is currently Cloak and Dagger VII with 90/90 casting points, and over 1200 Mana.
Isabella is still at 80/80 casting points (I don't know how to check her mana, but it can;t be that high as she only has three cities left (Priatan, Twen Drish and Reamon) I've captured all the others.

Then funny things started to happen. In a couple of recent battles she began throwing long range combat spells to aid her troops. Some sort of immolation spell and Skin of Oil, she is also trying to remotely dispell my charms and conversions.

If figured...fair enough...if she wants to waste her casting points and mana firing off spells at double the normal casting rate she is going to run out of power pretty damned fast. But the weird thing is that when I checked her after the battle she was still fully charged at 80/80 despite the fact that Sundren had drained her CP dispelling her spells during the battle.

I did a quick search online primary to try and work out what her spell inventory was as I'd never seen the spells she was casting before, even though I've played a rogue., and I came across a thread on the official forum entitled 'I RESPECT ISABELLA VAN HELDON'

[LukeBu]

Casting points X1billion so dont bother trying to disjunt or ever use permanent enchantment spells. Other buffs in production are more fair, and make up for ai being dumb, but seriously the computer does auto battle instantly, why not have ai run one before to just see if it will win to keep it from making dumb mistakes, and make CP bonuses much smaller?

[Gloweye]

That CP sounds like a bug. It's not intended to get bonuses...got a screenshot?

Dispel can be used also by a couple of units, and all hero's can get damage spells. Are you sure those didn't use them?

[Didz]

Messaggio originale di jaccovandorpGloweye:

That CP sounds like a bug. It's not intended to get bonuses...got a screenshot?

That's what I wondered. It seems odd that her CP drops on some occassions and seems to be working and on others occassions (well at least one) it doesn't. I'll try and get some screenshots next time it happens. I've been reluctant to post screenshots of my game, as I'm running with BARENTZ on, so that I can see what Isabella's up to, and I realise that might spoil the game for some players. Also, as a screenshot is just snap in time, it doesn't really prove anything in terms of what happened over say an entire battle. A video would actually be better if I could work out how to do one.

The 'I RESPECT ISABELLA VAN HELDON' thread on the official forum that I mentioned earlier is a sort of tongue in cheek complaint by a guy called Rhaeg, who has experienced the same thing in a straight game without the debugging on. (so he couldn't see what was happening behind the scenes) But he worked out by logging all the spells she was casting in a single combat that she must have used at least 300 casting points to be able to keep casting.

I know she doesn't have that many CP, because I can see how many she has, but it seems that for her casting spells doesn't always cost anything. So, she seems able to cast unlimited spells under certain circumstances.

I don't think thats always the case as I've been checking her CP regularly whenever she casts spells and certainly for the disjuncts her CP dropped. But it didn't following the battle and she must have cast Skin of Oil six or seven times on Sundren during that battle at double the cost. I'm pretty sure it cost 8xCP to cast, so that's 96 Casting points expended minimum, and that's without the immolations or whatever they were that she cast when Sundren failed to disjunct the Skin of Oil.

Actually that was another weird thing, in that there didn't seem to be anyway of finding out what the fire spell was which was why I was trying to look it up. I'm assuming it was Fireball which costs 10xCP to cast, or 20 if you aren't on the battlefield. However, the online spell discription doesn't mention that it leaves the target immolated just that it does 12 fire damage.

Anyway, it's all a bit odd.

[Nasarog]

Yes, this is a wargame with a rushing mentality. If you don't rush/blitz, you are done.

[Didz]

This game is annoying. Sometimes it seems that its deliberately playing games with the player.

I thought I'd try and provide some more tangibel evidence that there was a problem with the routine for casting combat spells. I was hoping to find some sort of combat log or something that would prove that the AI was casting more spells than it had the casting points for.

Unfortunately, I couldn't find anything and even if I had I realised that I didn't save the game after the battle I had in Turn 43. So I had no way of going back and looking at the log anyway.

However, I did save the game just before a similar battle on Turn 53 (just in case I wanted to turn back time) and I quit without saving in Turn 54 immediately after that battle. So, the Autosave is still set to the end of that turn.

So, quite by chance I have a before and after snapshot of the Turn 53 battle.

The annoying thing is that it shows a completely different problem.

Before the battle Isabella can be seen to have 90/90 Casting Points.
But after the battle her Casting Points have dropped to 6/90.

The problem is that having watched the battle replay, carefully noting every spell casting by both sides it looks like Isabella has been docked too many casting points.

Combat spell casting log based on the replay.
Starting CP: Sundren 90/90 Isabella 90/90

Round 1: Isabella casts Skin of Oil on Sundren (14xCP) Sundren dispells it (7xCP)
CP end of Round 1 (Isabella 76/90 : Sundren 83/90)

Round 2: Isabella casts Skin of Oil on Sundren (14xCP) Sundren dispells it (7xCP)
CP end of round 2 (Isabella 62/90 : Sundren 76/90)

Round 3: Isabella casts Skin of Oil on Sundren (14xCP) Sundren dispells it (7xCP)
CP End of Round 3 (Isabella 48/90 : Sundren 69/90)

Round 4: Isabella casts Skin of Oil on Sundren (14xCP) Sundred dispells it (7xCP)
CP end of round 4 (Isabella 32/90 : Sundren 62/90)

Round 5: Isabella casts Skin of Oil on Sundren (14xCP) Sundren tries to dispell but fails (7xCP)
CP end of round 5: Isabella 18/90 : Sundren 55/90)

Round 6: Isabella casts nothing. By pure chance I don't think Isabella has the CP to cast her fire spell as I think it costs 10xCP (20xCP remotely). Sundren tries to dispell Skin of Oil and fails again. (7xCP)
CP end of round 6: Isabella 18/90 : Sundren 48/90)

Round 7: Isabella casts nothing. Sundren dispells Skin of Oil (7xCP)
CP end of round 7: Isabella 18/90 : Sundren 41/90)

Round 8: Battle Ends.

Sundren comes out of the battle with 41/90 Casting points as expected.
Isabella ends up with 6/90 - 12 casting points less than expected.

So, either she cast somethig else in the same turn which I never witnessed or the AI was docking her extra this time. Either way it's not consistent with what I saw happen in turn 43. But it was my turn so she shouldn't have been able to cast anything outside combat.

I wonder if it matters who attacks whom. I think she attacked me in Turn 43, so perhaps she got her spell points back at the end of her turn. Before I was able to check them.

BTW: Any idea why steam screenshots have stopped working for this game.