Wednesday, September 29, 2010

Civilization 5 and Its Artificial Unintelligence

So, if you’ve listened to this week’s Jumping the Shark podcast you got a plateful of Civilization 5 thoughts. At that point I had about six hours or so in the game and had a generally favorable impression. Now that I’ve had a full week with it (and about 20 hours), it has soured somewhat. Don’t get me wrong, I’m not going back to Civ IV (yet). There’s a lot of stuff I like a great deal and, if I don’t get lazy, I’ll write about that too, in a later post. That said, I’m really distressed by the level of incompetence I’m seeing and reading about with regards to the AI.

Here’s the thing. I really want to be sympathetic about the game's AI issues given that the combat model is a total revolution (no pun intended) in the Civ series design. But I can't. This is something where Firaxis knows they're making this big change -probably right from the beginning- and they have to *know* that military AI will have to essentially be re-built from the ground up. That's something they have to get right. A few holes is one thing (and inevitable), but this isn't a game that, long term, can absorb having a toothless AI against an experienced opponent. And it is toothless. It apparently can't conquer a city state without a Herculean effort. How is it going to beat a human player once said player has grown comfortable with the rules and mechanics?

It would be one thing if the AI itself got better on higher difficulties, but from what I'm reading it really doesn't. It just gets bigger and better bonuses. (That’s a totally unverified assertion; your mileage may vary.) If what you see with the AI at the Prince level (where the playing field is totally level) is what you get, then they really have some work to do.

In my first game, played at Prince (4) difficulty, I achieved a science victory by 1985 (not totally unreasonable at that level), while the AI was still floating around in frigates and finding out gunpowder was kinda neat (way too inept at a level where the playing field is supposed to be level). Militarily… well, here's an example from that game, which I completed last night:

I'm (as the Americans) on a continent with the Persians and Greeks. I take out the Persians pretty early on. They never had more than a couple of military units and they threw them away by attacking the teeth of my invading forces (composed of archers, chariots, and spearmen; no siege units at all). With no siege units, conquering their two cities took a bit of time, but it was a speed-bump since Darius could not defend himself.

Later I went up against the Greeks, who were of equal or greater size to my civ, but were less advanced (later on, significantly so). I fought them, basically, in three stages.

The first phase of the war was fine. I attacked him where he was over-expanded and at his weakest. (He plopped two cities in a line, bisecting my empire in two.) No big deal that I easily won that confrontation. I should have, given my tech advantage and the fact that I had horses and Darius did not.

The second phase is where it was obvious the AI just didn't know what it was doing. When I started up hostilities again, Alexander had been picking on two city states (Stockholm and Venice). He'd mostly given up on Stockholm because I went in and positioned my own units around the city to protect them, so he appeared to have turned his full attention to Venice. And when i say “full attention,” I mean very nearly that. He left a couple of military units behind, but not nearly enough to protect his turf. So when I moved in, I basically had free range to roam his countryside and position myself around his cities however I liked. Bear in mind, any human player would have seen this coming. I had units all over my border with him.

Now, he's got at least four military units up to the north, basically laying siege to Venice, which was on a peninsula, sans actual siege units. I did not go to his capital straight-away, so even discounting my invasion was telegraphed a good five to ten turns ahead of time, he still had time to bring his forces back. Maybe even heal them up a little. As I took out my initial target (Sparta) he did nothing of the sort. I ended up taking his capital with nothing there to defend it because he was all about Venice. Again, I didn't even need siege units of my own. I just relied on my, now huge, tech advantage that had me running over his turf with Knights and Muskets.  (I couldn't build trebuchet's if I wanted to, since I lacked iron.) He did eventually pull units back as I moved north, but once his capital fell it was beyond too late. Once I had conquered all but two of his cities I took a peace deal so I could deal with unhappiness from all these new puppet states.

The third phase was just mop up. Alexander had no means to build enough units in ten turns to mount any kind of defense.

Ignoring my ability to build a big early tech lead despite not yet being familiar with all the new gameplay concepts, what really bugs me is that it seems clear the AI is incapable of moving to defend itself, even when it has the means to do so. This is not the kind of thing that can be justified by, "well, it's a new combat system."

Alexander focusing on a city state while he’s got hostile forces marching on Athens is just not acceptable. Also, just to be clear, I'm an experienced player, but not a particularly good one. And that’s an important distinction because you don’t have to be one of the savants at CivFanatics to badly exploit this game. You just need enough experience with the game to understand how the different mechanics work. That’s just a function of time.

Now, balancing that is a host of stuff I really like. The Social Policy system is really interesting, even if I do think I like the old Civics system a bit better. I *love* that cities can now defend themselves to some degree. AI aside, I love the new military model. It’s so much better than the stacks of doom model of previous Civs. I like how culture works. I think I like the new global Happiness model that essentially replaces stuff like corruption and healthiness, although I’m not convinced it’s well-balanced yet. (Others seem to think there’s a lot of ways to exploit happiness, but this is probably more a concern for players more skilled than I. Still, Firaxis may have oversimplified in relying on that mechanic to dictate empire growth.)

And also, to be fair, I did see the AI do some things that were, well, if not especially smart, certainly correct. Long after I vanquished Alexander, Napoleon came trolling from across the sea. He surrounded the city state of Stockholm with galleons (I think). I was not looking for war with France, but I also did not want to lose Stockholm as an ally. So, I exploited my giant tech lead by gifting Stockholm with a destroyer. As soon as that ship was in Stockholm’s port, Napoleon wasted no time in pulling back, rather than fight an uphill battle. That’s good. (Also, there’s a perverse thrill to be had from launching a torpedo attack from your sub on a nearby frigate. BOOM!)

I think the take home is ultimately that this game is a load of fun early on where you're just learning how it's evolved from CivIV (or learning it for the first time). Firaxis took a lot of chances and tried a lot of new things with this game and I respect that *a lot*. That said, long term, it really looks to me like Firaxis has a ton of work to do to get this game balanced and running the way it needs to. Their history says they will, but until they get a legit patch in the field, you should buy the game knowing that long term enjoyment from the gameplay is probably limited for the foreseeable future. (And by long term, I mean in the 40+ hour range. I put about 15 hours into my first game on a standard size map and game length and am far from tired of it. I can just see some of the writing on the wall.)

Yes, you can up the difficulty, but personally, I don’t enjoy the game as much when I know I’m getting beat just because the AI has ridiculous production and research advantages. I like to be challenged on a level, or only modestly tilted, playing field. That’s something I got from the previous games that it looks like I’m unlikely to get here. I’ve started a new game on King (5) and we’ll see how it goes. I hope to be proved wrong.