* Hey Everyone there are some new high-level models we need feedback on... *

[Mark_Everson]

We need your thoughts on three newly-written-up models. They are:

Government ( http://apolyton.net/forums/Forum21/HTML/000105.html )

Characters II ( http://apolyton.net/forums/Forum21/HTML/000104.html )

and

Diplomacy ( http://apolyton.net/forums/Forum21/HTML/000106.html )

Let's hear what you think!

-Mark

[Mark_Everson]

Ahem...

We Still need feedback on the areas above. Thanks,

-Mark

[SnowFire]

The stuff looks great, but I need to know a little something more about Clash of Civs first: Is it going to be against an AI or strictly multiplayer? Because there's no way that an AI can handle the sort of things in government that well. Then again, perhaps that's realistic.

[Mark_Everson]

SnowFire:

Clash is going to be both with AI and MP capabilities. The AI is going to be tough, but we have some good ideas already. Also, we will probably devote 10-100x the coding time to AI than is usual for the genre. What do you see in the government system that is so much harder than writing AI for Diplomacy, Military Strategy or grand strategy? (warning: I head off on vacation tomorrow morning an may not be able to respond in a timely fashion)

-Mark

[SnowFire]

I apologize for MY delay on posting back. The problem is, well... my vision of how a computer decides what to do is it takes stock of a situation and "ranks" it. If a peasant uprising is sufficiently urgent and their demands too ridiculous, there's an excellent probability it will get crushed through the military. If those values were reduced a little, then appeasing them or granting them privileges might appear a better option to the AI, and if it's a pacifist AI, there's a fairly good chance it will appease them with the more watered-down version of the uprising. The problem comes in on the inter-connectedness scale- there are many more variables that should be affecting the choice here. For instance, the military class might be getting inordinately strong, and pretty soon the government will be beholden to them. And even worse for this calculation is that some items receiving varying amounts of weight depending on the situation. The proximity of a revolting area is vitally important when communications are bad, almost irrelevant in the great age of exploration and early imperialism (save one group of colonies on the Atlantic seaboard), and relevant again in the age of nationalism. Plus, the randomness is an issue again- ridid, inflexible AI peronalties can be far too predictable and thus easy for a human to fool and beat, but ones that guess and use percentage points more to select what to do (this situation favors appeasment, so I'll do an 80/20 coin toss- oh! It came up 18, I guess the military wins anyway to make this more random) can sometimes do the wrong thing in a vitally important situation when the correct reaction is necessary.

Maybe I have a totally flawed interpretation, but that's what I think of when I think of AI-decision making. And that inter-connectivity can be the AI's downfall.

Memo to self: Check spelling BEFORE posting. (edited in)
[This message has been edited by SnowFire (edited July 08, 1999).]

[Mark_Everson]

Snowfire:

I've gotta run so this will be embarrasingly brief. There has already been some discussion of this type of thing in the thread http://apolyton.net/forums/Forum21/HTML/000094.html and also please check out the link in my first post in that thread to go to a site on the web page that describes our current intentions for AI. (although it doesn't directly discuss the govt AI a lot of things are similar).

-Mark