The Daily Update at the Com

**antisocialmunky** · 07-18-2009, 03:27

Yeah, the AI lacks any reasonable foresight most of the time. I'm not sure what you can do about it, I know the 'conditional statement' response seems like an easy solution but CA talked touted a goal based AI...

I suspect that if indeed CA actually implemented something like, the problem is structural. The issue would be that the AI doesn't analyse a set of options after each iteration, instead it just picks the one.

For those not familiar a basic goal based AI can be described as follows.
-You have a set of X choices that can be performed at each step with a reward function(for strategic thinking, I think a long term and short term function would be best).
-During each step, the AI selects the option in which after evaluating the reward function results in the best reward and adds that action to its todo list.
-It steps through Y amount of times and then starts to execute the to do list.

The problem with this though is that it results in a very miopic AI in that it will invariably pick the list that results in the most immediate benefits.

The simpliest way of fixing that is to take Z sized set of the best options at every step so instead of a list, you generate a tree. After Y steps, you select the single path in the tree that results in the highest reward score. So once you adjust the AI to look at a set of decisions, it gets more intellegent. This is why Chess AIs take forever. They are analyzing all possible good moves.

Now, I can't really tell if this is actually the problem but it seems like this might be part of it. It could also be the function that picks the set of actions for picking 'attack' every time or the reward function for each decision is seriously screwed. I dunno.

**Husar** · 07-18-2009, 14:30

Originally Posted by antisocialmunky

The simpliest way of fixing that is to take Z sized set of the best options at every step so instead of a list, you generate a tree. After Y steps, you select the single path in the tree that results in the highest reward score. So once you adjust the AI to look at a set of decisions, it gets more intellegent. This is why Chess AIs take forever. They are analyzing all possible good moves.

"Everytime I click end turn, it takes hours to finish, Buhuhuhuhu!!!"

In a TW game there are even more options to consider than in chess unless you make them rather abstract in which case you'd arrive for example at the prestige ratings, which reflect a faction strength somewhat well, irf the AI sees that the enemy has twice the army prestige, maybe it should look for a reliable ally before attacking, etc. You get some kind of relatively simple tree using things that are already there.
One thing that would really help though is a "my armies are crushed maybe I should offer money for a peace deal" consideration. Wars should be relatively quick and bloody, not always the drawn out back and forth (or just a never ending forth in some cases) that they currently are because the AI hardly ever accepts or offers a peace deal. People who do like the current wars can just deny the peace deals, no problem whatsoever.

**antisocialmunky** · 07-18-2009, 14:33

It would take as long as pathing for one unit as they do it now.

**Servius** · 07-18-2009, 16:41

I have to say, I haven't really played ETW is a while. Dawn of Discovery has been drawing most of my attention lately. But I think the primary reason I stopped playing ETW was the diplomacy. I don't like the "me vs. everyone" situation that arrises with the irrational diplomatic AI. I don't want to conquer the world. As GB, I want to rule the seas, keep the French bottled up on the Continent, fight pirates in the Carribean, fight frontier wars vs. natives in America, control many trade nodes, and begin to colonize India.

If I take Flanders, the dinky little German states should not declare war on me. I could destroy them in a single move, which is why it's stupid for them to declare war on me, and I don't want to garrison lots of little city states in Europe, which is why I won't declare war on them.

I want a logical diplomatic AI that only declares war if it thinks it has a real chance of winning. Once a nation is at war, I want the CAI to act aggresively to execute that war, achieve its war aims, and if it's successful (or tragically unsuccessful), it should sue for peace. What I don't want is a diplomatic AI that's more aggressive than the CAI. I also don't want a stupid DAI that basically makes suicidal decisions.

When the diplomatic AI acts rationally and with some thought to historic relations between nations, it feels more rational and realistic, which helps with immersion and significantly increases my enjoyment of the game. When it acts stupidly, I see through the facade, the immersion is gone, and it's apparent that it's just me vs. a single (yet multi-faced) AI that just wants to crush me.

**TB666** · 07-20-2009, 18:09

Hi guys,

We have stopped the "Backstabber" and "Dishonouring Treaties" diplomatic factors being applied when joining an ally in war. We have also checked in the first pass of diplomacy improvements. This consists of updating some of the diplomacy calculations to use new functionality that has been added to the AI since release and tweaking some of the factors. In addition, we have made changes to increase AI counter offers.

Thanks,

Mark O'Connell
(aka SenseiTW)

**Fisherking** · 07-20-2009, 18:15

WOW!

I hope this is the end of the Black Knight suicide factions!

Maybe we won’t be at war for the whole game and some factions may even except peace!

**Slaists** · 07-20-2009, 18:58

Originally Posted by Fisherking

WOW!

I hope this is the end of the Black Knight suicide factions!

Maybe we won’t be at war for the whole game and some factions may even except peace!

Well, I think in TW series being at war is the point. The problem was that many of the wars were meaningless and random.