When I built the Skirmish AI for Dawn of War, we had 4 levels of difficulty: easy, normal, hard and insane. The easy difficulty was brain-dead — it built a grunt every once in a while, and sent it your way, making it feel like something was happening. On the normal difficulty, it calculated your current army value and built to about 90% of that value, making the game feel like it was close, but almost always losing to you. On hard, the AI played as hard as it could, and on insane it got a 20% resource boost.

The point is, I tried to make the normal AI play at a level where it felt like you might lose, but almost never did. I don’t believe I quite succeeded (there was often a steam rolling effect — if the player lost a fight, the AI wouldn’t commit suicide to some of its army, so it ended up with a higher army count), but this idea to me is the pinnacle of fun game AI — not one that can wipe the floor with you, although that of course is a very interesting technical challenge.

I constantly push for a larger concentration on the initial user experience, game difficulty being one of them; however, rightly so, these factors are traditionally placed as a “B” feature whereas graphics was placed as an “A” feature. Let’s take a look at how Game Difficulty is starting to become more important by looking at different distribution models:

Distribution: Packaged

In packaged goods, unless the reviews stated the game is ridiculously hard (and even then, who but hardcore gamers read reviews?), a player is going to base your purchase decision off other factors, such as marketing, word of mouth, game play trailers and the like. After purchase, chances are they’ll have a higher tolerance to difficulty, since they’ve already made the purchase.

Now, when you talk about, say, Devil May Cry to a fellow player, are you going to tell them “oh it took me hours to get good at the game, but then it was totally fun”, or are you simply going to say “it was totally fun”? Chances are it’s going to be the latter, because you’ve already forgotten how hard it was to get good at the game.

Distribution: Try & Buy / Free to play

Contrast this to a Try & Buy experience. I remember being very excited that Alien Hominid was coming to Xbox Live Arcade. I immediately downloaded it when it came out, and promptly died over and over again in the first level. Did I purchase the game? Of course not. And you can also expect what my impression of the game is.

The core difference here is the barrier to entry and the barrier to exit. Packaged games have a high barrier on both ends, trapping a user inside. It takes far more effort to attract a new user, but once they have one, keeping them isn’t much of a problem (nor do they care that much if they leave).

Games that are free to play or try, however, have an extremely low barrier to entry, but a proportionally low barrier to exit. Hence, even though it’s easy to attract new users, because they have nothing invested in the game, they’ll quit if their first impression is not overwhelmingly positive. There are lots of different variables to the initial user experience, but difficulty is definitely up there.

This is actually a pet peeve of mine — if I die more than twice in a single level, I’ll probably quit and never play again. So please, make games that feel challenging but actually is almost impossible to lose!