Sharing some feedback on my observations of the AI in my games @salvor

After playing a few games with Starnet (no friendship), I find it is still too easy to befriend and thus defang your neighbours. The reason is because you can declare rivalry too easily with no consequences. Since the AI will usually max out declaring 3 rivals, you can easily declare mutual rivals that don't border you, thus they will never attack you; then your neighbour will like you, sign a pact and build trust, and by the time the environment has changed you'll have enough trust to stay at peace. Thus you have all the time to tech rush and econ boom, leaving them in the dust. Or take your time conquering them one at a time.

The only cases where this doesn't work is the first AI you meet, because you can't abuse mutual rivals yet, and they always hate you since you're the only target. That is usually not a problem because my strategy is to conquer the first empire anyway. Or if there is a genocidal or empire with strongly mismatched ethics, which is expected.

I think the core problem is the Rivals system, it's basically free influence with no consequence. I think the best way to address this is to limit Rivalry to only your neighbours, or only within a certain distance, so that there is risk of being attacked.

This brings to my second observation: I notice the AI empires are frequently getting into 2-front wars which devastate them, and then a different empire get into the same situation resulting in stalemate situations where no one is dominating. Even Advanced Starts AIs were often stalemated because of this. Ultimately this hurts all of them and made it easier for me as the player, because none of them could get big and snowball.

Since the AI is not declaring war on multiple empires at the same time, I think this happens because the AI is a poor judge on the chance of a neighbour declaring war on them. My thoughts on this:

1) Currently the AI by default hates all neighbours, and only manage to build alliances due to mutual rivals or mutual threats. I think it may work better if it behaves more like a human player, which is to focus on 1 neighbour at a time, and try to maintain good enough relations with the rest until it's their turn. This reduces the chance of them getting into 2-front wars that they can't win. I'm not sure how you can achieve this thru modding, but perhaps it can be manipulated using opinion modifiers and how they declare rivals.

2) How the AI should pick its targets. If it can judge the likelihood of another neighbour declaring war at them, it could adjusts its decision. Currently it can judge whether to attack based on the strength of their allies, perhaps it should also consider your own enemies that are not their allies.

3) Perhaps the Mutual Threat or similar concept should not be purely based on what other empires have conquered. Is it possible for it to judge whether another empire is a threat via the relative power comparison? By mid-game, if another empire is speeding ahead and have Superior or Overwhelming economy or tech (which is usually the Player), it would be a better idea for AI empires to start building alliances with other Equivalent power empires to counter them. Inferior and Pathetic neighbours it should conquer, but the likelihood of conquering similar strength neighbours for the AI is usually low unless they have allies, so they should find one.

What do you think?