Drive on Moscow drifts back onto the iPad with a whole new AI

, | Games

When Shenandoah Studio’s Drive on Moscow came out last November, the AI was, shall we say, primitive. People playing their first game reported wiping out every enemy unit on the map. The previous release, Battle of the Bulge, had left a lasting impression of a revolutionary and flawlessly implemented interface with a game system that fit it perfectly. Drive on Moscow, despite its many merits, had a bit of a sour aftertaste.

After the jump, time for a palate cleanser.

So when Shenandoah said they were going to do an AI rewrite, I was a bit skeptical they could turn a game I was using as a punching bag in solo play into something I would actually have to pay attention to. It wasn’t really on my radar, and I didn’t play the “improved AI” beta versions because I was afraid of screwing up my multiplayer games. When the version 2.0 update came out, I was interested in it for the ability to play on my iPhone. The AI upgrade didn’t initially register with me as a feature, even.

But after installing the game on my iPhone, and updating the one on my iPad, I figured I should probably see where the AI stands. I’ve now played a half-dozen games against the new AI, mostly as the Germans, but a few times as the Soviets. In the past, I had played lackadaisically in solo play, losing interest quickly when I saw it make error after error, and eventually stayed engaged a little longer only by seeing how recklessly I could play without losing. It was sort of like Kobe Bryant playing basketball against you with a basket of fruit on his head. But it wasn’t a long-term strategy for enjoying a game.

There was this moment when I was playing against this new AI and realized, “Hey, am I out of supply? Holy cats, I sure am!” I had played my usual “isolate Moscow” strategy, and the AI had done a nice job of ejecting me from the crucial spaces east of Moscow instead of doing its usual pointless dance in the mud fields south of Tula. But I thought I was doing well, until I realized the AI had cleverly isolated my entire northern flank.

I eventually won that game, and the next one. But the difference between those games and the release version was that I carefully considered every move, and when I examined the map while the AI was “thinking,” I would often think “it has to move to X right now,” and amazingly considering its previous track record, it would do exactly that. Sometimes it wouldn’t, and I would be a little disappointed, until I realized it had actually identified a better move than I had.

The AI still does weird things (as I was writing this, it just abandoned one of the crucial Moscow-adjacent spaces to send an armored unit to its death in Mozhaysk, a space that won’t be important until about ten turns later) but far more often, it does things that I think are quite reasonable. That’s a big change from previously, when it was sort of a contest to see what crazy things it could come up with. The only drawback is that it can take up to two minutes for the AI to find these improved moves.

Troy Goodfellow, Soren Johnson, and I devoted an entire podcast to Drive on Moscow. Soren mentioned that in a sense, AI development begins at release, because no matter how much testing you do in development, only a wide release is going to reveal the whole spectrum of possible strategies. That really puts the developer on the defensive, because the time that everyone is paying attention to the game is at release, and it’s hard to get that attention back. I mean, what effect is improved AI going to have now?

Maybe it will bring back some people who were turned off by the poor AI. It’s unreasonable to expect an artificial opponent to be able to stand up to an experienced player two months after release, but new players trying the game for the first time should be challenged by this iteration of its digital warbrain. Unfortunately for the game, maybe a fraction of the people who would have benefited from this before will end up hearing about it.

What did I say about experienced players? The Soviet AI just came close to beating me. I still won, just because it wasn’t aggressive enough in pursuing its counterattack. This weakness on the attack makes it a weaker Axis opponent than a Soviet one, but either way, it came within one point of a draw against someone who has been playing the game since beta. If you’re new to the game, you’ll probably find it plenty challenging for now.