Topology and AI

[PkK]

From posts here on the forum (and experince with other games) I deduce, that the AI provides more of a challenge, when inter-AI-interaction is more limited, i.e. AIs tend to not encounter each other early or often. This would implay that flat world topology provides a more challenging game than cylinder topology. Any anecdotes or even data about this?

Philipp

 

[unmerged(485425)]

Good question and sound hypothesis I think. I have never played on flat either so would be curious too.

It would take longer to get to various parts of the map, but it might mean the human player already has the "super" units that the AI has not worked out how to build yet so it might still be easier at that stage?

 

[unmerged(485924)]

I have seen the AI get totally gimped by being in a corner largely surrounded by mountains; it had a path around but it never did build a second city.

 

[Mardagg]

I have played solely flat worlds so far because i exactly had this thought process,i.e. AI might be better.
Its right that the AI is less likely to encounter the player or other AI`s early on and i had a really good game going on where i was in the middle with one AI far on the left and 1 far to the right.The problem was that the AI,while having many cities and at least some pretty good units, had huge trouble defending his big empire vs the independent monsters.Once i did win the border wars it wasnt a challenge anymore.I had much more troubles beating all those indies spawning everywhere while easily steamrolling the AI.
So,for me,its now clear,that the AI`s main flaw is that it cant counter strong spawns.E.g. there had been 2 Ogre caves near the AI capital and instead of going for these places it always tried to defeat the Ogre`s that came within reach.Once i reached this capital,there already had been like half a dozen pretty experienced ogres that actually did conquer several of AI cities near its capital.
And since i believe indies are strongest at impossible level i think that the AI might actually be better on hard difficulty setting.
But this is kind of a dilemma,since good indies are the one thing that makes the game interesting because they can be challenging at times.
I dont see an easy solution to this.

 

[unmerged(170187)]

The more I play the game the more I think the AI's biggest problem is actually a financial one. The AI does HORRIBLE city design that have huge upkeeps and very poor income. Because of this the AI has NO money to build advance units or to upgrade them. Everything else falls apart if you don't have the resources to do stuff, it doesn't matter how great the other areas of the AI code are since it's can't be tactical with no units.

The reason I came to this conclusion is that I have been playing the 90 Turn Challenge and even when I was playing on Larger Maps with more AI I would notice the same issue. The AI starts with a huge amount of extra cash on Impossible difficultly. The result is it pumps out units like crazy and an average player can have trouble with these large numbers or even get hammered if they are unlucky enough to start too close to AI. Those cases are rare though and the ones we hear about in the forums where AI Rushed them with some massive force.

The problem comes when it runs out of this massive fund of starting cash. Then it comes down to the players level and has to make every gold coin and unit count. This is where the AI ends up running out of steam as it were. It can't keep the pressure on the player because it doesn't have the gold for lots of new units. When the AI is left alone by player and other AIs it has time to simply stock pile units and gold. The result is a sizable force, and thus challenge, for the player to over come. Once this initial wave is gone though the AI has nothing left to throw at the player.

In contrast in a lot of my games I have continuous waves of units bashing up against the AI shores. So many units that at times they get stuck in bottle necks and are backed up a few tiles deep. This is what I would expect to see from the AI if it actually had the funds to continue fighting. It usually has troop training buildings in every city it has which means they all should be building new units to replace the ones you wiped out. Thus if the AI has 6 cities I would expect at the very least to see 4-6 new units popping out every 2 turns plus summoned units since that is the training time for many units. And this is what I often see from the AI when I rush them early on in Impossible but in late game I see the AI replacing it's losses at a slower rate.

In comparison to another game, Civ5, the tactical combat ability of the AI seems to be on par. Civ5's AI I found is not actually that good, or at least not as good as people make it out to be. I'm very good at the tactical game and in Civ5 I easily win most battles even if I'm out numbered. I played Civ5 on Deity a few times and the challenge was not in a smarter AI but simply one that had more resources. The AI often had twice the forces I did but I out maneuvered them every time and would loss 1 unit to every 3-4 of theirs that I killed. The problem was they could build new units so fast it was very difficult to make any head way. Also thanks to their increased income they actually advanced to an age above me and despite being out numbered and out teched I held the line, because the AI sucked.

I mention this because Civ5 has an easier time getting away with this due to its simplistic economy. That is every city pretty much boost the income of your empire in some way. So the more cities the better, even with happiness penalties, thus making it very easy on the AI as all it has to do is manage what +bonus building it wants to make in a city. Even the upkeep on the buildings in Civ5 is so minor that a few trade post can easily offset them. Thus when you add a few +% production modifiers to the AI for harder difficulties the AI simply does what it always did but now can do it faster than the player and has more resources to throw around. So the player is forced to be even more efficient with the resources he earns to make up the difference.

The problem in Warlock is that the economy is more complex and requires a fairly well planned out overall strategy of specialized cities which the AI seems unable to do. If you don't specialize your cities then you will suffer extremely slow growth with only a fraction of the income you could be making. Pretty much all the AI cities I capture I can't help but think, what is the AI doing?!? as I have two buildings from different specialize paths in it. And the vast majority of the cities with higher level troop production buildings but no cities devoted to resource production. Most players can even do some minor specialization and out produce the AI. Then combined with the ability to out maneuver the AI in tactical combat for fewer losses and the game becomes rather easy.

Easily way to test this is some option that gives the AI unlimited cash and see how well it does.

 

[unmerged(485425)]

I think you're right. When checking late game what their resources are like they often have masses of mana but literally only 10 or so gold. As you say, it would be interesting to test them with unlimited gold. Impossible should be hard and as the AI should realistically get some assistance then perhaps an automatic money injection every 10 turns or something may go some way to helping them out in the short term before their city specialism AI gets improved.

 

[Hassat Hunter]

Or just improve them so they don't *just* make mana-cities... that'll help.
Mana drains money a lot...

 

[Mardagg]

For this,the financial problem, there should be some solutions,that I think would be fairly easy to implement.
For a start You could give the AI a much bigger fixed income per turn on harder difficulties,or like ticktoc said 1k gold every 10 turns or so(to let the AI produce more waves).
Another solution,maybe even in addition , would be to make all buildings the AI builds costing no upkeep,maybe no upkeep cost for units as well.

This would make the AI better until it gets patched enough to be able to make enough specialized cities.
You can further expand this idea by adding some new traits at character creation: no upkeep for units,no upkeep for buildings.These should cost a lot and every AI opponent should automatically get this on hard/impossible in addition to the normal picks every character has.Like it was in Master of Magic.

 

[Hassat Hunter]

no upkeep for units,no upkeep for buildings.

Even at 10 points, they would be majorly overpowered.

 

[Mardagg]

Yes,thats right.Maybe 50% discount would be ok though.
To make full 100% discount on upkeep work it has to be more than 10 points.This can only be done if negative traits are added so that one can spend more than 10 points...

 

[unmerged(170187)]

I don't see an need to add an additional trait, especially if it's one the player can't use.

Also no upkeep would be fine for the AI but not for a player. The problem is the AI sucks so badly at city planning that even with no upkeep his income would still be pitiful in comparison with a decent player. I keep saying this, as well as the value of city spam, but it seems people don't realize just how big of a difference it is. So I crunched the numbers and here it is.

This is the city growth and gold output for a single base Human city with no +gold resources in it. First column is size of city, 2nd is number of turns after founding to grow to that size, third is the income of the new buildings added at that size increase, and last is the total income of the city. The 3.5 at the start comes from the +2 for the Castle and the 50% conversion of food to gold from the Farm. City growth also assumes no use of Prosperity type spells.

Size-Growth-Gold-Gold Income
1 --- 0 --- 3.5 -- 3.5
2 --- 6 --- 4 --- 7.5
3 --- 11 -- 2 --- 9.5
4 --- 17 -- 50%- 12.5
5 --- 24 -- 75%- 17
6 --- 32 -- 100% 23
7 --- 40 -- 100% 29
8 --- 50 -- 3 --- 41.75
9 --- 61 -- 3 --- 54.5
10 -- 75 -- 3 --- 67.25
11 -- 92 -- 3 --- 80
12 -- 116- 3 --- 92.75
13 -- 153- 3 --- 105.5
14 -- 245- 3 --- 118.25

As you can see the city's income starts taking a huge jump when it hits size 8 and starts building craftsman guilds. This is because of all the +% modifier buildings that were built previously. At the beginning it is rather small amounts so most people dismiss the benefit an extra city or two can have. But within 32 turns your city gold city is enough to buy a new basic unit every turn. Within 61 turns it can the city is producing almost enough to make a tier 2 unit every other turn which is also about how fast you can train them.

If you throw in a silver mine +6 gold then by size 8 it has 54 gold per turn, with gold mine +12 it's 67, and gems +20 it's 114. And those numbers are with just 1 +gold resource in the city limits. This kind of specialization setup makes it so that by turn 70 I can usually have about +200 gold per turn coming in ever after expenses from couple dozen units with -100 gold upkeep and few perk/troop buildings(which sadly don't have a total anyway and not gonna count them all). With no upkeep on buildings I imagine it would add another 100 gold then throw in the unit upkeep and net gold income is probably close to 400 per turn. The AI by this point I estimate doesn't even make half that, which is mostly eaten up by maintenance cost. Even with no maintenance cost I'd say it's safe to estimate my positive, after expenses, income is higher than the AI's total income before it pays upkeep. So yea I don't really see it having no upkeep as much threat. By turn 100 I can easily see positive gold income of 500 or more.

And for those looking to truly optimize their empire the build order above is not actually optimal. If you build the Craftsman Guild before the first 100% gold building you actually get a total of about 37 more gold over the 16 turns from Size 6 to size 8. I know it doesn't sound like much but that is per city, so across several cities that could be enough for a couple extra units or perk upgrades.

Also keep in mind that because of the +325% bonus in human cities each craftsman is worth 12.75 gold per turn and 9.75 gold in non-human cities. So keep that in mind if want to build any other building besides craftsman in gold farm city as it's basically costing you that much in upkeep(It's the opportunity cost in economics).

 

[Mardagg]

Yes and i dont ever expect the AI to be on par with the player regarding city specialization in this game.
I would be happy if it at least does specialize here and there and builds unit producing cities on expensive special resources at times.
For me its pretty important therefore that the AI gets much bigger boost to its gold income at higher difficulties from whatever sources.

The "no upkeep" or "50 % less upkeep" idea gets pretty interesting btw for summoned units.Especially once there can be cast more than 1 summoning spell per turn (maybe via artifacts already now).Talk about a real ghost wolves spam...

 

[Snap Wilson]

I like the idea of the AI having +100 gold per turn on Impossible (and maybe +50 on hard) as a way of countering it's resource difficulties. It would allow the AI to keep the pressure up on their assaults. It wouldn't be enough to combat city spamming, but I think that's a separate issue.