The (failed) VIP group (Gamescom, again)

[Schuesseled]

I for one will be very much dissapointed if ally's mean nothing to the AI come endgame. Civilizations aren't spotty FPS'ers in a free-for-all, if your allied with a winner, then you are a winner.

Actually it would be really cool to see the list of allies associated with a winning civ in the end of game screens, give credit where its due.

Winner - Washington (And his allies Monetesuma and Elizabeth)

 

[Seven05]

AI that 'plays to win' and AI with 'flavorful personalities' are not mutually exclusive. You guys don't have to take everything to such extremes

All the AI needs to have in order to 'play to win' is a way to devlop and follow long term goals. All the AI needs in order to give it that feeling of 'historical personality' is a foundation of basic weights on how likely it is to perform various specific actions to achieve that long term goal.

Neither of those render the other impossible to achieve.

 

[mercury529]

Because you don't design a game by including hundreds of different optional features to cater to every last little individual taste.

So your focus group determined that this feature is wanted by only a small subset of players? Or right now are you just automatically assuming the majority of gamers share your opinion?

Long-term features... that affect an entirely different game?? On a different world? Its nonsensical.

Absolutely, have the AI respond to how you've played in a particular game. But to how the game was played in previous games played on the same computer??

Leaders living for thousands of years? Nonsensical.
Knowing roughly how leaders you just met will act? Nonsensical.
Knowledge of what special units a country will be able to produce before they exist? Nonsensical.

If it enhances gameplay, it does not have to make sense. But frankly a liberal use of imagination can justify it if you'd like: "In his rise to power, he gained a reputation for being..."

 

[The_J]

Were you able to test the login/logout online/offline features of the steam parasite? Or mod loading protocols?

No, nothing like that.
But to mention: It was installed via steam, and the Gamescom computers sure had no internet.

And did you seriously expect someone to start checking out mod installation for a game that isn't released yet, during a 1 hour hands-on demo at an expo?

Would you expect someone to look at the files in the folders ? Probably not, but i did.

Which reminds me, that i've seen in the folders of the natural wonders different subfolders, named "european", "african", "asian" and "american", all with textures in it.
Not sure what effect this will have ingame.

I believe Azazell had a picture showing the diplomatic history log but it's been taken down now. It basically listed all ongoing and completed deals you've had with other civs, how long they lasted, when they ended, etc. That seems like the only diplomatic information you could have right? All diplomacy is through deals

That's the only thing I'm asking for. The version that was shown at gamescome didn't have this. I couldn't even check one turn after I made it against whom the Secrecy Pact went that I just signed with Oda. Or whether I already had a Friendship Pact with Alexander. That's the only thing that's bugging me.

God, i feared that i didn't see it. Good, that you confirm that i'm not the only one.

Will the AI have visibility into how strong you are? For that matter, will we know how strong they are? (military strength, techs known, number of cities, "score", etc.)

That reminds me on something else: I didn't see a scoreboard at all.
No idea, if an AI is stronger than me.

Isn't it possible that the AI makes up it's mind about a trade or some other bit of diplomacy based on current situation

And how many turns long is the "current situation" ?

 

[Calouste]

That reminds me on something else: I didn't see a scoreboard at all.
No idea, if an AI is stronger than me.

Arioch has quotes on his site from the Military Advisors that gives an indication of your relative strength towards other civilizations. I'm quite ok if the computer keeps track of how many and which type of enemy units have been spotted and compares that to my own army on a scale of 1-10.

 

[Ituralde]

That reminds me on something else: I didn't see a scoreboard at all.
No idea, if an AI is stronger than me.

The scoreboard was there. When you click on your Civ icon and all the Civs you have met are shown with their icon under it, where you can also contact the other civs there was a big number next to your civ and the others as well. I'm pretty sure that was the score, though I didn't hover over it to see what it was composed of.

Also there was a demographics screen, which showed you how much military, production, food, land mass, population and what not you had. That's what got me so confused to the missing Diplomacy Screen as it really didn't offer any features. So my hope is really that some info was just missing from it.

 

[Tatran]

If a "friendly" AI is going to attack you when you are about to win, then diplomacy is completely meaningless.

IIRC, in civ2 at emperor and deity level when your space ship was the fastest one,
all civs, pact brothers or not, would eventually declare war on you within a few turns.
Only the civs with one, maybe two cities, left would not declare.
Rise of Nations had something similar when you were about to win.

I prefer the ''hell breaks loose'' concept, when someone is about to win, over 30 turns pressing enter for a space win confirmation.

 

[The_J]

Arioch has quotes on his site from the Military Advisors that gives an indication of your relative strength towards other civilizations. I'm quite ok if the computer keeps track of how many and which type of enemy units have been spotted and compares that to my own army on a scale of 1-10.

Ah, nice, didn't see that.
Is the foreign advisor in the same way?

The scoreboard was there. When you click on your Civ icon and all the Civs you have met are shown with their icon under it, where you can also contact the other civs there was a big number next to your civ and the others as well. I'm pretty sure that was the score, though I didn't hover over it to see what it was composed of.

Also there was a demographics screen, which showed you how much military, production, food, land mass, population and what not you had. That's what got me so confused to the missing Diplomacy Screen as it really didn't offer any features. So my hope is really that some info was just missing from it.

Same as above: Didn't see it...it was not enough time .

 

[cccv]

AI that 'plays to win' and AI with 'flavorful personalities' are not mutually exclusive. You guys don't have to take everything to such extremes

All the AI needs to have in order to 'play to win' is a way to devlop and follow long term goals. All the AI needs in order to give it that feeling of 'historical personality' is a foundation of basic weights on how likely it is to perform various specific actions to achieve that long term goal.

Neither of those render the other impossible to achieve.

I hadn't thought of it like that. My interpretation of "play to win" is, if your best buddy ally is about to build a spaceship, you nuke the crap out of him. If you don't you're resigning yourself to loss and not playing to win. But I suppose it's possible to make winning a goal in the sense that Greece will try a diplomatic victory, but can switch to culture if diplomacy is a dead end without being ridiculous and ignoring relations whenever it suits him better to play like a computer rather than Alexander (yeah, Alex was a bad example because he was the biggest backstabber in IV anyway, but I'm not retyping >.>) So yeah, I'd be satisfied if winning was the computer's long-term goal that dictated what kind of victory to try for and then working toward it, but not "win at all costs at the expense of personality-- we're all megalomaniacs bent on world domination."

Anyway, we've kind of heard both. We heard that the AI will play with those long term goals and will have preferred styles of playing, but not to the point that they can't adjust to the world if their preferred way is not going to work. But we've also heard that they wanted the AI to play like a human and simulate the multiplayer experience. So we'll have to see how far they go with the latter.

 

[Calouste]

Ah, nice, didn't see that.
Is the foreign advisor in the same way?

Don't know about the foreign advisor, most quotes there deal with city states, and of course most of the screenshots that have been published are from early in the game.

The military adviser has the following to say about other civs:

"I think I saw the only unit the Iroquois Empire had. It looked sad and lonely. It would be tragic for them if they got in a war with someone of our power."

"The English Empire are not to be trifled with. They have a stronger military than us and we should only consider aggression in the most dire of circumstances."

"Our war with the Persian Empire is going well and we have a stronger fighting force than them. Keep the pressure on and let us claim victory!"

That's the kind of level of information that I can use to make strategic decisions.

 

[Thormodr]

Nice avatar Tatran. I assume this is intentional?

 

[Louis XXIV]

If the AI is modeled to act in its best interest over an indefinite number of games, then the AI would not backstab friends unless a genuine opportunity to win the game presented itself. If a friendly AI will win the game by thwarting your efforts, then it is reasonable to do so. But if they have no chance to win the game, there would be no reason to jeopardize their reputation.

I would argue this behavior is ideal. The AI should realize that, if it can't win, it shouldn't try. I don't think this should be specific behavior of consequences for future games, but a general programming philosophy that 100% untrustworthiness by the AI hurts the AIs interests. That one AI who could backstab to win won't be able to do so unless other AIs generally don't do that. The AI that seeks to work with the player can gain an advantage by cooperating. Of course, they shouldn't be sheep, but it should be a middle level of cautiousness that changes throughout the game based on how the player plays. Think of it as playing against an AI player with a medium level of trustworthiness. Whether or not you'd want to cooperate with them would depend on if they are high or low on their random scale. So you could watch for signs throughout the game to see how they're going to act, always taking the chance that they aren't trying to trick you, but not being so paranoid as to avoid helping them period.

 

[Some guy]

People have the wrong understanding of diplomacy for this game. It isn't meant to convince the enemies to be mindless puppets. That's what City-States are for. Diplomacy is used against the AI opponents to form temporary alliances or discourage random aggression by offering something alternatively beneficial.

Being elected president of the world is now possible even on multiplayer b/c of City-States.

It's not realistic for a civilization to be "peaceful" and able to resolve military disputes by diplomacy. Some civilizations just simply had to be ousted before diplomacy could be used in place of war.

 

[sdlufkin]

I guess the good news is if you guys had been missing where to find stuff maybe they just simply didn't have it finalized/in the build you had? Since they probably wouldn't put a pre-Gold demo out, they must still be adding stuff like that. One can hope!

 

[Mercade]

AI that 'plays to win' and AI with 'flavorful personalities' are not mutually exclusive. You guys don't have to take everything to such extremes

All the AI needs to have in order to 'play to win' is a way to devlop and follow long term goals. All the AI needs in order to give it that feeling of 'historical personality' is a foundation of basic weights on how likely it is to perform various specific actions to achieve that long term goal.

Neither of those render the other impossible to achieve.

This.

I hadn't thought of it like that. My interpretation of "play to win" is, if your best buddy ally is about to build a spaceship, you nuke the crap out of him. If you don't you're resigning yourself to loss and not playing to win. But I suppose it's possible to make winning a goal in the sense that Greece will try a diplomatic victory, but can switch to culture if diplomacy is a dead end without being ridiculous and ignoring relations whenever it suits him better to play like a computer rather than Alexander (yeah, Alex was a bad example because he was the biggest backstabber in IV anyway, but I'm not retyping >.>) So yeah, I'd be satisfied if winning was the computer's long-term goal that dictated what kind of victory to try for and then working toward it, but not "win at all costs at the expense of personality-- we're all megalomaniacs bent on world domination."

Anyway, we've kind of heard both. We heard that the AI will play with those long term goals and will have preferred styles of playing, but not to the point that they can't adjust to the world if their preferred way is not going to work. But we've also heard that they wanted the AI to play like a human and simulate the multiplayer experience. So we'll have to see how far they go with the latter.

Also for the purpose of playability of the game it has to be that while AI has a plan to win (long term goal instead of just messing with its worker instructions on every turn), but that also means acknowledging that someone else is better and winning the game. It would be rather annoying if every time you had a 25% lead it meant that all AI would backstab, nuke and dogpile you because they're adamantly against letting you win. M.A.D. is not a fun way to end every game with.

 

[stealth_nsk]

It would be rather annoying if every time you had a 25% lead it meant that all AI would backstab, nuke and dogpile you because they're adamantly against letting you win. M.A.D. is not a fun way to end every game with.

The funny thing is - if many players dogpile the leader, the winner will be a player who don't participate in it. So I don't think AI will try to attack leader unless:
- The win is close.
- AI sees a clear way prevent the win.

 

[vranasm]

I think that on the "AI wants to win and is programmed that way" we can look from 2 perspectives

1) Ai wants to win on their own... when they choose spaceship - they try to build it asap and launch, culture - SP, domination - well they certainly will attack you but you should guess from the ammount of war they had

2) AI doesnt want lose and let someone else win ... well then they will do whatever they can to break everyone else when they are close to win...and that will for sure devolve the game to domination games and firaxis could just abandon spaceship victory and dont invest money, because there would be no chance to win it.

I think that the answer will be 1), because 2) would be really really lame...
I liked the suggestion that Allies of winner should be mentioned in victory screen... that could maybe allow another type of victory... you were ally of victor! so you actually didnt lose and when you are friend with someone else launching ship and you for example did protect him etc. you would be victor too

 

[Lyoncet]

I'm pretty sure you're right on your scenario 1 being the case. Although I would like to (and, actually, expect to at some point a while from now) see an option to make the AI, or at least a few of the leading AI, actively try to prevent wins from other contenders. Yeah, there's the "if I want MP I can play MP" argument, but not everyone can play MP. Don't know how hard that would be to code in/if the devs think it would be worth the effort though.

 

[stealth_nsk]

I really expect what leading AI will try to prevent competitors from winning. But I think lesser AI will seek their chances in leaders' opposition, not trying to stop one of them.

 

[kaltorak]

People keep complaining about there being no information on what you're doing "wrong" with AI diplomacy. It's quite simple when you're making the AI mad at you; you're beating them.

The AI is out to WIN, not suck-up to the humans. And why should we know about what wars are going on, especially if we have no one scouting? Somehow were just supposed to magically know?

Basically, people are mad that the "illusion" of competitive AI was replaced with an actually competitive AI (human-like players that actually WANT to win). Saying if I wanted to play "human-like AI I would play multiplayer" isn't an argument. You just don't want to lose, admit it.

No. People are mad that in a single player game aimed at historic immersion, that the AI seems like it will play like a deathmatch bot rather than an actual nation. Alliances, friendship, vendettas and so forth are meaningless if historic actions don't affect the AI's likelihood of declaring war.

If I didn't care about how the other players acted, and just wanted them to win, then I'd play multiplayer. How is that not an argument? This is a game about history, not just Command & Conquer.

I want a game where real history can happen. If AIs are just playing to win, then the Allies would have annexed France after D-day. After Waterloo, the English would have turned around and marched on Amsterdam. Britain would have refused to bankroll the construction of US industry in the 19th century, fearing a competitor. The European union would never have formed.
In real history, alliances are meaningful, and countries don't attack their friends just because they don't have a big army on their mutual border. Cooperation is possible. Grudges are held, and acted on, whether rational or no.

I want a diplomatic system that feels real, and then I'll crank up the difficulty level to make it challenging.

I don't want to have to look through a history log to try to figure out the state of diplomatic relations. That should be available, at my finger tips.

If an AI isn't willing to trade with me (for example, because its worried about my military strength) then I need to be able to know that, and know why.

There is nothing inherent in having a transparent diplomatic system where actions matter that says they have to matter forever. Its entirely possible for example that declaring war could give a -4 for the next 50 turns, thne -2 for the 50 turns after than, then -1 for 50 turns, then nothing.
But without a transparent system, you won't know this.
And its ridiculous to think that having declare war on a country and razed their cities won't have *any* future impact on their diplomatic dealings with you, even in the short term.

I couldn't have explained it better.