Encouraging AI to vote for others in World Congress host votes

In my experience in BNW, I have found that, if I don't have the delegates to win the World Congress host for myself, and I don't like the current host, it's best I vote for a civ I would prefer to host the World Congress. While everyone should retain the ability to vote for themselves (especially for UN World Leader votes; under no circumstances should anyone help an opponent win the UN World Leader vote), sometimes it's advantageous to vote for a player other than yourself. Human players understand this, but the AI at present does not, and the AI will always vote for itself even if it has a snowball's chance in hell of winning. This also results in the World Congress being hosted by the same civ for most of the game.

I think there is a very simple fix to this for the AI's decision-making algorithm (understand, though, that I have not read the code). The AI could be programmed so that if the AI's delegate count is lower than the arithmetic mean of the World Congress delegate counts (the arithmetic mean being preferred because of its sensitivity to outliers, since the highest outlier will win the vote), the AI player will choose to vote for a candidate other than itself (I'm thinking the one it has best relations with, like in G&K). This will make World Congress host votes more competitive and help prevent one player from dictating affairs in the World Congress (in my last game every civ hated me and I just kept railroading resolutions through the WC, reversing every prior decision and stacking the votes in my favor with World Ideology and World Religion; it was fun ), making the organization more dynamic and competitive.

For example, take a congress with the following delegate counts: China: 4; India: 2; Russia: 1; America: 1; Germany: 1; Siam: 1; Aztecs: 1; Japan: 1. The arithmetic mean of the delegate count is 1.5. At present, China will win the Host vote, without challenge. But under my proposed change, Russia, America, Germany, Siam, the Aztecs, and Japan would all vote for a player other than themselves (they each have one delegate, less than the arithmetic mean of 1.5); China and India, however, would still vote for themselves (both have delegate counts higher than the arithmetic mean of 1.5). If China is unpopular, this could result in a civ other than China winning the vote for Host; if India, for example, earns half of the undecided votes, they could win the vote for Host; even America could win if five of the six undecided votes prefer America. So while China still has the advantage, the WC Host vote is much more competitive than it used to be.

Thats an interesting idea.

AI's without a snowball's chance of winning a World Leader vote giving their delegates to their closest ally makes sense.

Alternatively, each civ could be required to vote for two civilizations each World Leader vote: so they vote for themselves and whoever they like the most.

Yes, I agree with this proposition, I pretty much suggested exactly the same the other day. The only thing I want to say is: I don't think a civ should *always* vote for someone else, I think if there are no other civ that they have friendly relationships with, they should still vote for themselves. Not sure if you implied that, but just wanted to add my 5 cents.

It has to be defined gradient on how they split votes and I think it should always be a split vote.

Allowing the AI to throw entire blocks of votes around would result in frustrating stalemate or a rather good game going down in defeat because of AI logic/a technicality. It may also encourage even more CS hoarding.

Neither a fun, nor warranted just to satisfy the need to feel like the AI is voting for someone else.

but picking up 1-3 votes per Civ from good deeds done is good enough to overcome the proposed idea of nerfing CS delegates down to 1 and slightly increasing win threshold by +1 per map size.

I should also add that it MUST be a 2 way street with mechanisms in place to discourage narcissistic voting and encourage a trading of votes. If the human player just wants to vote for themselves all the time then expect the AI to pitch in ti make the difference, then NO THANK YOU. I'd rather we just rebalance how delegates is doled out in the current model than add yet another human only feature.

dexters said:

It has to be defined gradient on how they split votes and I think it should always be a split vote.

Allowing the AI to throw entire blocks of votes around would result in frustrating stalemate or a rather good game going down in defeat because of AI logic/a technicality. It may also encourage even more CS hoarding.

Neither a fun, nor warranted just to satisfy the need to feel like the AI is voting for someone else.

but picking up 1-3 votes per Civ from good deeds done is good enough to overcome the proposed idea of nerfing CS delegates down to 1 and slightly increasing win threshold by +1 per map size.

Click to expand...

I'm not entirely sure, why would that happen? Quite on the contrary, I would say that if you force smaller nations (in terms of delegates) to split up there votes, you'll end up with a situation where they completely lose their importance, because their votes end up being evenly split over the larger nation.

I think each civ should always only vote for one civ, being that themselves or someone else. That way, diplomatic victory will be a battle to actually win the favor of other civs, and isn't that what it should be about?

dexters said:

I should also add that it MUST be a 2 way street with mechanisms in place to discourage narcissistic voting and encourage a trading of votes. If the human player just wants to vote for themselves all the time then expect the AI to pitch in ti make the difference, then NO THANK YOU. I'd rather we just rebalance how delegates is doled out in the current model than add yet another human only feature.

Click to expand...

Again, I don't agree with the perfect symmetry idea. Game is *not* totally equal between AI and human player, and it shouldn't be. The symmetry does not consist in human voting for AI, the symmetry comes from the AI can *either* vote for the human *or* for another AI. This ensures that it's not a guaranteed win ticked for the human, because the system can work in favor of another major AI who has kept good relationships with other civs.

I think it's important to acknowledge that we're talking about a game, and that the gaming experience is focused on the human player. Part of that gaming experience arguably is potentially losing to the AI, but that does not mean there can't be differences. I often see the "perfect symmetry" argument envoked when it comes to vaious espionage/information features also, because people say that this or that can't/shouldn't be in game, because AI will not be able to get the same information from the human player, but that is imo. a wrong approach.

kaspergm said:

I'm not entirely sure, why would that happen? Quite on the contrary, I would say that if you force smaller nations (in terms of delegates) to split up there votes, you'll end up with a situation where they completely lose their importance, because their votes end up being evenly split over the larger nation.

I think each civ should always only vote for one civ, being that themselves or someone else. That way, diplomatic victory will be a battle to actually win the favor of other civs, and isn't that what it should be about?

Click to expand...

Split votes as in voting for themselves + sending some of their votes for the preferred candidate to win. Similar to how they only send their core delegation to vote yea or nay from vote trading rather all their other extra votes.

So they always vote for themselves with extra votes going to someone else with the number of votes graded towards the amount of good things/reputation/warmonger hate a Civ has. This will work for human and AI. Then again I'm not really that hyped about this idea myself. Read below.

Again, I don't agree with the perfect symmetry idea. Game is *not* totally equal between AI and human player, and it shouldn't be. The symmetry does not consist in human voting for AI, the symmetry comes from the AI can *either* vote for the human *or* for another AI. This ensures that it's not a guaranteed win ticked for the human, because the system can work in favor of another major AI who has kept good relationships with other civs.

I think it's important to acknowledge that we're talking about a game, and that the gaming experience is focused on the human player. Part of that gaming experience arguably is potentially losing to the AI, but that does not mean there can't be differences. I often see the "perfect symmetry" argument envoked when it comes to vaious espionage/information features also, because people say that this or that can't/shouldn't be in game, because AI will not be able to get the same information from the human player, but that is imo. a wrong approach.

Click to expand...

Sure, but I don't see this having an advantage over what we have now which is fairly transparent, achieves what you are describing above for being human centric, but is slightly unbalanced due to the weight of city states and the gold nature of relationships with city states. So we're discussing how to fix it.

What you want is a whole lot of coding to reinvent the DV voting system essentially for a vanity exercise so you can see votes show up coming from another Civ.

Why not just rebalance how delegates are doled out, and or reward extra delegates for behaving nicely.

A 'peace prize' proposal with a winner + runner's up would work, granted it needs to come with a slew of other delegate and wonder adjusts. That way even peaceful AI civs get to partake and benefit and it works within the context of the game.

Let's keep in mind that to rework how AI approaches DV 1) there is no guarantee results with be satisfactory 2) it will require way more coding only. vs. incremental reworking of delegate counts/win threshold which is ultimately where I feel the work needs to go. Just rebalance it. Make it more interesting with proposals that add delegates, or maybe even remove delegates. Something the AI can already handle.

dexters said:

Sure, but I don't see this having an advantage over what we have now which is fairly transparent, achieves what you are describing above for being human centric, but is slightly unbalanced due to the weight of city states and the gold nature of relationships with city states. So we're discussing how to fix it.

What you want is a whole lot of coding to reinvent the DV voting system essentially for a vanity exercise so you can see votes show up coming from another Civ.

Why not just rebalance how delegates are doled out, and or reward extra delegates for behaving nicely.

A 'peace prize' proposal with a winner + runner's up would work, granted it needs to come with a slew of other delegate and wonder adjusts. That way even peaceful AI civs get to partake and benefit and it works within the context of the game.

Let's keep in mind that to rework how AI approaches DV 1) there is no guarantee results with be satisfactory 2) it will require way more coding only. vs. incremental reworking of delegate counts/win threshold which is ultimately where I feel the work needs to go. Just rebalance it. Make it more interesting with proposals that add delegates, or maybe even remove delegates. Something the AI can already handle.

Click to expand...

Well I agree with many of your concerns in the compromise between what would be ideal and what is realistically achievable. I do think a major focus point of a change like I suggested would be to actually get votes from other civs into play when being chosen as World Leader. One might get around it by having something like your peace prize, but then again, that only focuses on one aspect, namely who's been most peaceful. What I want to have is actually civs voting for the one they like the most.

I don't think the AI should vote for anyone other than itself for world leader. The recommendation that the AI should vote for others for Host is based on the fact that sometimes that there is some strategic value in doing this (I've done this when I have no hope of becoming Host, and I don't like the current Host); there is no strategic value in doing so for World Leader.

Circuit said:

I don't think the AI should vote for anyone other than itself for world leader. The recommendation that the AI should vote for others for Host is based on the fact that sometimes that there is some strategic value in doing this (I've done this when I have no hope of becoming Host, and I don't like the current Host); there is no strategic value in doing so for World Leader.

Click to expand...

Yes, that is true if you think if it from a sort of realism pov, but what is vote for World Leader anyway? It's purely a game aspect, and from a gameplay pov, the current system is not good, because it means that winning "diplomacy" victory is essentially a question of earning enough money to buy out all the City States. Not a lot of diplomacy there. In fact, some of the most abusive wins I've had in civ was buying out all city states on the turn before vote and then DoW'ing everybody else in the world to prevent them from buying them back. That's some great diplomacy there!

But I do agree that my original suggestion of only having the top contenders up for world leader election would probably be a less contrived way of achieving the same.

I like the idea of only top contenders being up for world leader elections: it would be nice if diplomatic victory had more to do with actual diplomacy than bribing city-states.