AI voting behaviour for Global Hedgemony

[phoenix2054]

The purpose of this thread is I want to understand if the AI is geared towards playing to win, or if the AI is COLLECTIVELY motivated to stop the human player from winning above all, possibly at their own expense. I’ll explain

My current game is yet unfinished and I am the game leader overall taking all factors into account. Under normal circumstances I would expect to win from my position. But the AI has just voted unexpectedly and I think also illogically during the hedgemony vote

42 votes were required for a diplomatic victory (Im actually playing as Babylonians). I felt secure knowing that whilst the game’s diplo threat are the Netherlands, they only have 31 seats in WC. But while every other civ voted for themselves, Venice voted for Netherlands! They gave them all 10 of their votes and consequently they came within just 1 vote of a diplo victory. But why did venice do that???

Yes both the Netherlands and venice have a DOF and DP. But so what? I have the same with China and they didn’t vote for me. Surely despite whatever relationships civs have with each other, when it comes to victory conditions its every civ for itself, no? I would think that venice would want to stop the Netherlands from winning just as much to stop me or any other civ.

Why would a civ vote to help another civ win, if for no other reason than to screw over the human player?

The only explanation is that the Netherlands bought the venetian votes. But their economy wasn’t particulary strong and votes are expensive, right? I don’t think 10 votes is realistic unless they were swimming in cash

And shouldn’t the venetians be smart enough not to sell hedgemony votes to the game’s diplomatic powerhouse? Any other WC votes are on the table, sure. But NOT Global Hedgemony. They may as well just give away their capital!

In case its relevant im playing on prince 4 (vox populi) with no other ‘behavioral’ mods involved

Interested to hear the thoughts of other players on this. Because with VP the AI already plays in an intelligent way and provides a good challenge, I think. (Prince: chooses from top 2 best picks for decisions etc)

The AI really doesn’t need to compromise gameplay by trying to stifle the human rather than play fairly and consistently, and of course…intelligently.

Moderator Action: Moved to the main forum. Also, the AI doesn't treat humans differently from other AI players. Civs who can't win themselves might vote for civs they really like. This is not AI exclusive. There is an option to disable it in (1) Community Patch > Core Files > Core Changes > DiploAIOptions.sql - Recursive

 

[CppMaster]

I agree with you, AI shouldn't treat the player differently and Venice shouldn't sell their votes in that case.

However, this subforum is about Vox Populi Congress for changes to the game, not World Congress that is in the game. Move it to the main forum.

 

[phoenix2054]

I see. And how do i do that?

 

[CppMaster]

For sure you can just create a new thread in the main forum and maybe remove this one. Or ask moderator Recursive

 

[Stalker0]

I see it as a form of economic victory. If your close to the hegemony and can afford the insane price of buying the votes, then you win.

And when I say insane, I mean insane. I’ve paid 800 GPT for a vote before, those are the kind or numbers I’m talking about.

 

[CppMaster]

I see it as a form of economic victory. If your close to the hegemony and can afford the insane price of buying the votes, then you win.

And when I say insane, I mean insane. I’ve paid 800 GPT for a vote before, those are the kind or numbers I’m talking about.

But the point is that Venice shouldn't be interested in selling the votes at all in this case if it want's to pursue victory. It makes sense flavor-wise, though.

 

[Stalker0]

But the point is that Venice shouldn't be interested in selling the votes at all in this case if it want's to pursue victory. It makes sense flavor-wise, though.

We have had this discussion a few times. At the end of the day, the decision was that..... if vote trading is going to be a viable option, it needs to be usable. The problem we had for a while is that the AI just wouldn't vote trade, and so the mechanic became useless.

The compromise was very very expensive hegemony votes, aka if you can cough up the insane funds to snag a few votes, well then congrats.

Now the OP does mention he doesn't think Venice has that strong an economy, so its possible something else is going on. Maybe this was not a vote trade but a failure in AI logic. Or maybe vote trading for whatever reason wasn't that expensive this time (is Venice a vassal, that could impact things potentially)

 

[phoenix2054]

Venice are definitely not a vassal. They’re long time friends with Netherlands and venice are militarily the stronger of the two anyway.

I did not know this subject has come up before. If this has been allowed as acceptable vote trading, then it’s a clear error. As I said, vote trading for anything else is fine, but there’s no way an AI should ever trade away votes for hedgemony (possible exception as vassal for flavor). Its throwing away victory and is technically being stupid. I thought that cbp was all about stamping out AI incompetence?

Above is a moot point because as you both say, it costs hundreds for a single hedgemony vote. There’s no way Netherlands had the cash to buy all 10 votes.

 

[Stalker0]

There’s no way Netherlands had the cash to buy all 10 votes.

Actually I don't think you can even buy 10 votes, as its based on the diplomat level. Even at highest level you can get what like 6? I have never seen 10 vote

 

[Stalker0]

Just an example of how expensive vote trading can be. This is not even a major proposal, and America actually likes me. But look at how expensive it is, I'm not even close to meeting it.

Spoiler :

 

[CppMaster]

We have had this discussion a few times. At the end of the day, the decision was that..... if vote trading is going to be a viable option, it needs to be usable. The problem we had for a while is that the AI just wouldn't vote trade, and so the mechanic became useless.

It doesn't have to go to extremes, though. AI shouldn't be forced to trade. It should only trade if it'd actually benefit them. Can you imagine any player taking that deal? Would you? I don't think so, so AI also shouldn't.

Just an example of how expensive vote trading can be. This is not even a major proposal, and America actually likes me. But look at how expensive it is, I'm not even close to meeting it.

This is fine, that could actually benefit American and I could even take that deal myself.

 

[Stalker0]

It doesn't have to go to extremes, though. AI shouldn't be forced to trade. It should only trade if it'd actually benefit them. Can you imagine any player taking that deal? Would you? I don't think so, so AI also shouldn't.

This was and is the debate. Should the AI in every single case play the absolute best way to win, or should there be some concession for roleplay?

This debate also comes up in diplomacy discussions. From a "best way to win" standpoint, the AI should always be at war with the human in the end game if the human is ahead, full stop. However, people have repeatedly asked for long term deal making and diplomacy to factor in, aka people hate it when every AI declares war on them "just because". The compromise was that the AI does have big diplomatic negatives once a player starts to win, but its not absolute, and strong positives can overwrite it.

The same compromise is used in vote trades. It is very very expensive, but its doable. This maintains the spirit of the diplomatic victory actually including diplomacy, rather than just a CS grab fest

 

[ma_kuh]

It sounds like Venice may have sold a bunch of votes, and then Netherlands lost the vote anyway, so Venice just pocketed a bunch of money, a huge diplomatic boost with their militarily superior neighbor, and the game is still going on. They've given Netherlands a vote (or a few? I don't remember how partial hegemony "wins" work) in future win checks, at the benefit of a (presumably) hefty short-term boost in economy. Whether that's worthwhile in the long run...

So in terms of a strategy, I don't think this play is even outside the bounds of what optimized play would be.

 

[pitofanavocado]

I just played through this scenario.

I was Babylon and Netherlands was ~10 votes away from Global Hegemony, but in my case Netherlands actually got enough to get the win.

Myself and Netherlands were Freedom, and everyone else was Order. I knew it was possible for AI to vote for another Civ, but figured since they were different Ideologies there would be no chance of it happening. Incans and Mayans were also getting reasonably close to Cultural victory (~30 turns away). Netherlands was friend with both, and both Inca & Maya gave them ALL of their votes.

Aside from preventing me from winning (I had the tech lead, and would probably hold off the Inca/Mayans, but it wasn't certain), I don't understand why they would do it. If they were all the same Ideology AND friends, maybe, but different Ideologies, no.