With the new patch, I've been kicked down 1 1/2 difficulty levels (from Deity to Emperor/Immortal) without, as far as I know, any adjustments to AI handicapping.

In my book, that is an unequivocal improvement in the game.

I think there are more possible changes that would do the same. Of course, the obvious one is to "improve AI." The problem with that comment is that it is pretty vague (which AI? what do you want it to do?) and in general is darn hard (from a programming point of view) and resource intensive. Certainly we all want better unit tactics from the AI. But I thought I'd throw out at least one suggested AI change that seems easy (yes, it's easy for me to say easy) and could improve challenge without AI handicapping:


Have the AI stop buying open borders and resources that it doesn't need. It's something that a human never does, is always a bad idea, and (as an Emperor/Immortal player) is something that I've come to rely on as a sort of a crutch. I could be wrong, but it seems to me (with my limited programming experience) to be something that could be fixed relatively "easily" (I really hesitate to use that word in reference to someone else's work, but there you go). I'm pretty sure this would drop many mid- to high-level players about one difficulty level.
Better unit tactics from AI (seems to be improving but much improvement still needed; I've listed a few things below that don't really involve "planning" -- these are just "bad moves" period):

Stop moving archery/artillery/GGs onto unprotected tiles (and move them off tiles that become unprotected and exposed).
For CS under attack, DON'T move that archery/artillery unit out of the city! (This is often an example of above, but sometimes the tile is clear this turn but exposed the next.)
Don't attack into a hornet's nest unless/until you have a lot of backup coming that turn.



(If folks find this line of discussion productive or interesting, please make your own suggestions below...)

The AI at deity has huge bonuses already. They just nerfed human strategies-weaker longswords, weaker research agreements, lower happiness, etc.

Yes, I know that the AI has bonuses already (at everything above Noble, I think...). But my assertion is this: any improvement to challenge without increasing those bonuses is a good thing. Nerfing certain strategies is one way to do this. Alternatively, you could teach the AI to play these strategies (but this is harder to do, clearly).

As far as specific "AI" things, I would think that protecting siege weapons & great generals better would be a good starting point. It shouldn't be too difficult to at least get the great generals to consistently stay safe. I can understand the occasional blunder with siege given the complexities of planning an invasion, but for me it really breaks the immersion when the AI just leaves their great general unprotected on the front lines.

The AI at deity has huge bonuses already. They just nerfed human strategies-weaker longswords, weaker research agreements, lower happiness, etc.

I'm sorry, but this is so Not True. Have you played after the patch? The AI is *massively* better post-patch. Before the patch, AI units did not move with purpose. They loaded up on the units and then kind of milled them around.

Now the AI will come in force toward your cities and pick them off and if you aren't one/two tech levels ahead, god help you.

The AI at deity has huge bonuses already. They just nerfed human strategies-weaker longswords, weaker research agreements, lower happiness, etc.

A) They did improve the AI.

B) Most good AI design is still taking tools away from human players. You design the game in such a way that the "exploits" don't work. THen you don't need as large AI handicaps.

Yes i think they improved the AI and don't forget that changes for the human player affect AI as well. If only they can fix the ''sell free library from GL each turn'' trick.

Nerfing human-favored strategies is often a tough call. I'm always in favor if it is some "gamey" mechanism that adds no flavor, such as RA blocking (now nicely killed) or "one-way" resource exchanges (human always sells/AI always buys, with very rare exceptions; still present although weakened a little due to human need for happiness). Rifle rushes or rushes to certain wonders are still powerful (see strategy forums) but they now are either risky or are no longer "automatic" wins. This is as it should be.

(Edit: The library thing mentioned above is just a bug. I'll predict here that this will be one of the fasted bug fixes ever. Although if you ask me, it should be at the bottom of the priority list because it is so easily avoided.)

The main thing needed is improved tactical combat AI. Open border abuse etc matters nothing as long as you can easily defeat any army due to the AI's non-existent tactical sense.

Even Prince would provide more of a challenge than current Immortal level if the AI had at least the tactical capacity of a newbie player. As long as the AI will pull off things like rushing artillery units in front straight at enemy cities, you'll always be able to easily defeat it at any level possibly excluding the obscene Deity level.

Tactical combat AI improvements: Absolute top priority thing to work on for Firaxis. Frankly, it is shameful how bad it really is.

Removing infrastructure "exploits" isn't necessarily a good thing. It's analogous to removing supply line mechanics in a wargame - it forces you into a symmetrical contest, which is all about mass - and boring.

If the AI has some nice open land with few obstacles then I'm seeing some sensible maneuvers. It's when the lay of the land is tricky, with mountains, hills or choke point type areas, that the AI still struggles. But I suppose we can't have everything all at once, give it time.

One thing I hoped would be better by now, but isn't, is the AIs use of GGs. I am still seeing them commit suicide on a regular basis.

I have a feeling that the AI is teching a little slower than before the patch. Without RA blocking you can't get to the strong military techs as fast anymore, but I don't think it's too much of a nerf to warfare as I seem to see AI rifles and such coming a tad later now.

On a bright note I saw France execute a nice late game invasion in my last game. Atom bombed my army and border city, then took out a few of my artillery with fighters/bombers and then crossed over with a horde of foreign legions and lancers. I'm happy enough with how things are progressing.

The main thing needed is improved tactical combat AI. Open border abuse etc matters nothing as long as you can easily defeat any army due to the AI's non-existent tactical sense.

This is a lot better than it was in the early game in two ways:

- The AI builds more units
- The AI often throws the kitchen sink at objectives

It still has plenty of head-scratching moments, but you have to take early DoWs a lot more seriously than in the past when they were merely inconveniences that precluded luxury improvements on the frontier.

The ai should properly develop it's land. I see too much of the unimproved cows, and untouched swamp-land thing happening.

Moved to Ideas & Suggestions. :)

The most common complaint I've seen is the AI sending vulnerable ranged units in front of melee when attacking. Not knowing how the AI works, is this something that can be fixed? The AI needs to know where the "front" is (axis of attack), decide that the melee should be in front, with ranged units right behind, to be in range of the target.

I think just this one fix would improve the combat ten-fold, or at least give everyone something else to complain about, which would be refreshing.

Against my better judgement, I've added "better unit tactics" to the list (with some specific issues noted by others or myself). I agree with everyone here that this is the #1 problem. However, I don't think it is particularly "easy". I'm ranking these by a sort of ratio of added challenge / difficulty to implement (my subjective estimation of course).

Removing infrastructure "exploits" isn't necessarily a good thing. It's analogous to removing supply line mechanics in a wargame - it forces you into a symmetrical contest, which is all about mass - and boring.

I have to disagree here. Look, the game is balanced a certain way with exploits present, forcing players to play a certain way (unless they want to self-handicap) and forcing the developers to boost the AI in certain ways (e.g., absurd happiness/gold bonuses). If you remove exploits, then that balance changes, perhaps even (in some cases) to the point where the game becomes unfun. In that case, I say remove the exploit and re-balance the game and/or reduce AI bonuses (e.g., if you remove the one-way-human-to-AI resource selling then maybe you need to increase gold from marketplaces, trade routes, etc., or perhaps reduce AI gold bonuses). I don't see any reason to ever maintain an "unfair" advantage for humans -- all that does is necessitate bigger bonuses for AI.

In any case, I'd prefer a symmetrical war where my army and the AI's army are matched in size and yet I'm challenged. The joy of beating the AI with 10:1 asymmetry (against me) is very fleeting. Heck, I'd even swallow my pride and drop down to "Prince" difficulty to play such a game.

The ai should properly develop it's land. I see too much of the unimproved cows, and untouched swamp-land thing happening.

Can any others comment on this? I haven't particularly noticed this, but then perhaps I haven't been looking. (Untouched swamp makes perfect sense if the city has other tiles to work, but a cow within cultural borders should be improved very early.)

OK, this subforum kind of reminds me of The Island of Misfit Toys.