While I can appreciate that the programmers had to do something to maintain the challenge for players, I just don't get the whole "refusal to ally" thing...

Surely a smaller power would look over the border at a massive army of a powerful neighbour and say "Hey, let's have an alliance, we'd much rather be your friends than your next targets."?

Granted a lot of human players may take advantage of the AI's friendliness by launching an attack immediately after agreeing to be allies....but again maybe this could be tackled by the programme slapping "untrustworthy" or "treacherous" traits on the offending player. Maybe there would be some other way to restrict attacks on allies?

I have no idea how this would work, just throwing it out there for discussion purposes. From my own corner of the sandbox, I would like to see the AI approaching the diplomacy thing a little differently.

Any thoughts?