I don't think for a second that Firaxis INTENDED to "not have diplomacy".
Well they did and didn't. Personally, I think Kaltorak is being extreme in the assessmant, as I have had favorable "luck" with my A.I. interactions, especially as you play the game a couple of times because you get a grasp for understanding who behaves in what ways. Like knowing that Montezuma in civ4 will probably attack you at any point for no reason. In civ5, however, they all have that potential logic, except I wouldn't call it entirely random dice rolls.
We've been told various snippets as the game was talked about that suggests why decisions will be made at seemingly random points. For example, it was said that
every turn our A.I. neighbors make an assessment of us, and of us in relation to their game win. Likely, this assessment also takes into account what turn it is. So while you and a "friend" may have had a fantastic relationship for 100 turns, it's highly possible that at 120 turns, their assessment of you begins to suggest that you're no longer important, or they're reached a point where they don't need to work with you. This is where we start seeing demands come into play, or declarations of war.
It
is like playing a human player in the sense that ultimately, human "relations modifiers" in a multiplayer game are meaningless unless the teams are already picked. Any time there is some agreement of any form, it'll usually result in inevitable backstab somewhere. So in this sense it's like playing a game of FFA civ.
Again, whether people want to buy into this logic or not is their decision. I think it makes sense and falls into what I've observed. Others may not have had the same "luck". However, to suggest that it doesn't attempt to work this way because the A.I. is makes bad decisions is a little odd because A.) Humans make poor decisions too, and B.) If computers never made bad decisions, Chess wouldn't come standard in almost all PCs, since it'd be impossible to beat.
So, you wanted tips, here come my tips;
Habit - Learn what A.I. are more aggressive (Wu Zeitan) and more peaceful (Ghandi). It isn't to the point where Ghandi will sit by and be your friend all game. He'll probably come in to conflict with you at some point.
Pacts -Anytime anyone asks for pact of cooperation, if you have no desire for conflict with them, accept it. I've yet to discover for sure if there is some kind of "you're working with our enemy" penalty, but I don't think there is. Now, this pact is not a ward against conflict, it's to act as an alarm. I've observed one thing from civ's ~ usually, they tell you what they're doing if you join them in playing their little system. I don't think I've ever seen an A.I. declare war without first telling you that their done with a Pact. So, naturally, keep pacts with civs to serve as a warning that you're headed into eventual conflict with someone that ends it.
Don't Lie - This may sound silly, but I feel that you have a world reputation here. Just like if you're a bastard in an MP game... people will probably gang up on you. Admit to war if you're called out on it. If you're at someone's border with an army and they say "you're attacking me aren't you" - admit to it. The diplomatic ramifications of this are widespread.
Misc] - just like in past games, gifts help. As do trades, and research pacts. Interacting with the A.I. still causes them to "like" you. You just have to understand that ultimately they are your competition and they are no longer bound ball and chain to some code that prevents them from ever turning against you. So be sparing with your gifts, and just don't get caught off guard. As in past games, having a big military will help in the A.I.'s assessment of you. I've still been attacked but I've had the A.I. actually say to me (paraphrasing) - "We probably can't beat you... but we've got to try", and this was at a point where the other civ's were ganging up on me due to my city-state aggression, which they asked me to prevent, and I said "mind your own business".
All in all, I feel the "diplomacy" is pretty intuitive if you get off the notion that there's some way to truly manipulate A.I. behavior, and rather you can only influence it. So many people say, "I was friends with X civ, and they attacked me!" and lament about the system... Except, for myself, I've had the opposite happen. I've actually made friends with life-long, mortal enemies. I had one game where I was enemies and at war with Persia for half the game, and the relationship turned around and we were trading and sharing research pacts together.
