Yeah yeah yeah, but will the AI work in this game!?!?

[Dearmad]

It's obvious Paradox can design game systems that are complex and compelling, but will the AI be shipped brain dead on this one or not? Perhaps design SOME of the game systems so that an AI can handle them without as much trouble as it gets into in the EU series...

I know I sound a little harsh, but you guys have soooooooooooo much POTENTIAL with these games.

 

[MacRobert]

Most of us like the all the games Paradox puts out. We just don't show it as much as people who don't.

 

[Napoleon XIV]

Sergei has already said that by the time the game is put out it will be passed version 1.00 to avoid the HOI problems and such.

Sergei said once that 1.00 seems deadly to release...

 

[unmerged(502)]

Originally posted by Napoleon XIV
Sergei has already said that by the time the game is put out it will be passed version 1.00 to avoid the HOI problems and such.

Sergei said once that 1.00 seems deadly to release...

huh, why call it 1.00 then?

 

[unmerged(28)]

Originally posted by webbrave
huh, why call it 1.00 then?

Possibkly because 1.00 means it fullfills certain technical specifications.

/Greven

 

[Jos Theelen]

Originally posted by Dearmad
It's obvious Paradox can design game systems that are complex and compelling, but will the AI be shipped brain dead on this one or not? Perhaps design SOME of the game systems so that an AI can handle them without as much trouble as it gets into in the EU series...

I know I sound a little harsh, but you guys have soooooooooooo much POTENTIAL with these games.

I think that AI's have big troubles fighting a war. All those armies, all those possible sieges, all those possibilities it has to choose between, too much for an AI. But the other things, especially relations, look more simple.
So I guess the CK-AI will do a better job then the Hoi-AI.

 

[Murmurandus]

Originally posted by webbrave
huh, why call it 1.00 then?

Well, you have to start somewhere...

 

[Norgesvenn]

0.01? It just doesn't sound like much more than a notepad file.

 

[Endre Fodstad]

I think the solution to AI is to preserve the more simple EU/EU2 model rather than go for the more complicated HOI detailness that efficiently wrecked that game, even in 1.03.

EF

 

[unmerged(485)]

Originally posted by Greven
Possibkly because 1.00 means it fullfills certain technical specifications.

/Greven

The technical specifications for HOI must have been a mess!

On the serious side though, the AI is the toughest part of a game to make. There are so many things which look simple to us because we have years of intuitive understanding about many things - the AI does not. Just because you can see the path needed to get from Anglia to the Highlands (sorry if this seems like I'm not discussing this game) doesn't mean the AI can see it - it needs to calculate it. And this is just moving from one place to another - something we think is simple.

 

[unmerged(547)]

Originally posted by Sonny
The technical specifications for HOI must have been a mess!

On the serious side though, the AI is the toughest part of a game to make. There are so many things which look simple to us because we have years of intuitive understanding about many things - the AI does not. Just because you can see the path needed to get from Anglia to the Highlands (sorry if this seems like I'm not discussing this game) doesn't mean the AI can see it - it needs to calculate it. And this is just moving from one place to another - something we think is simple.

Yep, I mean it is Artificial intelligence after all. It will be a dark day, IMHO, when an AI in a game can play a better game than a human opponent. So...go for MP

 

[Derek Pullem]

Originally posted by Endre Fodstad
I think the solution to AI is to preserve the more simple EU/EU2 model rather than go for the more complicated HOI detailness that efficiently wrecked that game, even in 1.03.

EF

Or alternatively admit that coding an AI is an evolutionary thing and make as many of the AI parameters adjustable (as HoI 1.03 and EU2 1.06 have done) as possible.

Lets face it - the first chess AI's were shit but the later ones are getting pretty damned good (better than me certainly if timed)

 

[Dearmad]

All agreed. but an IMPORTANT part of AI design in *games* is to construct the *game* in such a way that the AI you can prgoram can *play* it somewhat well.

Check out Stardock's game designs- these are solid simulation/game dsigns with *solid* AI. There are other AI's out there too.

I'm not saying copy another company in the details, but perhaps apply the specific and personal game design style of Paradox to a model wherein *game* is looked at a few times OVER *simulation* so that an AI can follow up better.

I returned HOI, I've kept EU2 so that's about where my limit lies. Well, I kept EU1 as well, but stopped playing it rather fast- I kept it to reward a game company that I thought had a future (though I though EU1 was rather bad, it had promise), now it is time to fulfill that promise; I won't be keeping anymore babies that I don't think have at least a 20 IQ this time.

 

[Araanor]

Originally posted by Greven
Possibkly because 1.00 means it fullfills certain technical specifications.

/Greven

Paradox(al) convention?

 

[The Andy-Man]

I'd say take as many lessons from EU2 ad HoI AI as possile, and make the AI as open-source as is possible

 

[Endre Fodstad]

Originally posted by Derek Pullem
Lets face it - the first chess AI's were shit but the later ones are getting pretty damned good (better than me certainly if timed)

Chess is a fairly simple game(in terms of rules, at least). How many years did it take to produce Deep Blue?

EF

 

[unmerged(485)]

Originally posted by Dearmad
All agreed. but an IMPORTANT part of AI design in *games* is to construct the *game* in such a way that the AI you can prgoram can *play* it somewhat well.

Check out Stardock's game designs- these are solid simulation/game dsigns with *solid* AI. There are other AI's out there too.

I'm not saying copy another company in the details, but perhaps apply the specific and personal game design style of Paradox to a model wherein *game* is looked at a few times OVER *simulation* so that an AI can follow up better.

I returned HOI, I've kept EU2 so that's about where my limit lies. Well, I kept EU1 as well, but stopped playing it rather fast- I kept it to reward a game company that I thought had a future (though I though EU1 was rather bad, it had promise), now it is time to fulfill that promise; I won't be keeping anymore babies that I don't think have at least a 20 IQ this time.

I've seen the kinds of AI you are talking about - on Gameboy systems. Designing a game that can be handled by the AI would be great if you didn't have to cope with trying to make it as realistic as possible.

Wanna go from here to there in a space sim? Just move to a stargate and boom- you're there. Wanna move from Pondicherry to Oahu? Little different story.

 

[Dearmad]

Sonny,

those are pathfinding AI's you're talking about. Merchant Prince of Venice (the first one from the 90's) had a near-perfect one. That's simple math taking into account various factors the programmer finds difficult. With today's machines those aren't too hard to program, it's an issue of resources, not programming ability. I've programmed a few pathfinding algy's (some even original) in my day as well- I KNOW they're tough, but they CAN be perfected if you have enough resources (memory & cpu cycles).

And when you crit. my point and bring up gameboy you didn't bother to look at a Stardock game- their Corporate Machine game has an AI that kicks serious butt and is NOT a gameboy game. Do some research before you try to crit. me out like that, please.

I was talking about designing a STRATEGY game in such as way so as not to KILL the AI, which Paradox comes very near to doing each game they put out, and in fact DID do to HOI.

 

[Araanor]

Could you get a link to Stardock?

 

[unmerged(6160)]

Dearmad, I respect your option, but have you coded a Paradox AI file? Have you had to deal with releasing a game when the publishers tell you to do so? Have you run a software company with only 2 programmers?

Give Paradox a break, HoI is a great game, EU2/EU are great games. I am sick of all this wininng.. At least Paradox follows its games....

Go Paradox!