on Jan 15, 2009
The Birdmen post the other day kicked off an interesting discussion on auto-leveling or auto-adjusting difficulty in games. Now, I’m very much against auto-adjusting difficulty, because it solves one problem – the need for a game to provide the “right” level of challenge to all players – by creating a worse one: Taking away the ability of the player to adjust for frustration tolerance.
Ask people if a game is easy or hard, and you’ll see responses like:
Player1: Pfft. That game was a cakewalk. I only died maybe once a level.
Player2: That game was a pain in the ass. I died on almost every level.
So even among players of the same skill, the same experience can lead to very different perceptions. Some people want to play on a level they know they can handle and hoover up the content. Some players want the threat of failure to enhance their excitement. And some players want constant failure to test them and force them to develop their skills.
The answer for me is, “The same way I find it exciting to see Bruce Willis shoot bad guys, even when everybody knows Bruce is going to win.” There is the excitement carried by the story and spectacle, and the other layer of excitement carried by the gameplay. But if a game poses too great a challenge, then both layers of excitement are replaced with simple frustration. Personally, I tend to play the game the first time on normal difficulty (and lately, easy difficulty) and enjoy the story. Then – if the game pleases me – I’ll ramp up the difficulty and focus on the gameplay. Lately, a lot of games have simply not been worth the effort to learn to play them well.
But if I do enjoy a game, I sometimes find myself seeking greater challenges than the ones provided by the designers. During a gaming dry spell a couple of years ago, I went through Half-Life 2 on all the difficulty settings, and then I tried to beat it “Nethack”-style by not saving / restoring. If I died, I started over at the beginning of Route Kanal. I tried several times. My most notable game was the one where I made it all the way to the Citadel, and then died stupidly by blundering under one of those vertical stompers right inside the entrance. Close, but I moved on to other games without ever truly beating it sans-save. Ah well.
Auto-adjusting difficulty usually works by lowering the difficulty slightly when you fail, and raising it slightly when you have success. (Note that this is different from auto-leveling enemies in RPGs like Oblivion, which I also think is a bad idea. But that’s another discussion.) The designer decides for himself what rate of failure is “appropriate”, and enforces that on everyone. Players will either win or lose until they reach an equilibrium at the point where they meet the designer’s intended failure rate. If you need to fail twice in each section of the game to keep the difficulty at the same level, then you are going to fail twice in every section, no matter what. If you get better, so does the game. Your skills may improve over time, but your overall rate of success won’t.
Managing frustration is the most crucial part of enjoying a game. When people complain about difficulty, they’re usually complaining that the game was too frustrating. Auto-adjusting difficulty actually establishes and perpetuates the problem you were trying to solve. People with a low frustration threshold won’t just be annoyed until they can get better. They will be annoyed forever, no matter how good they get. People who want a Serious Challenge can never really get it.
Worse, even for the hardcore players who want to put themselves up against a brutal challenge for the sake of overcoming it, it deprives them of any good metric for progress. They don’t get the satisfaction of acing something that used to be too hard for them, because the game has ramped up the difficulty to match. Their true score – the internal number that governs how much the game is cheating or taking a dive – is usually not shown to the user.
(Think of it like a weight set that doesn’t allow you to set or even see how much you’re lifting. It auto-adjusts until you can do exactly ten reps. I do ten reps. Grandma does ten reps. Dwayne Johnson does ten reps. If you want to set the weight low and lift something light while you watch TV, you can’t. If you want to see the maximum you can lift, you can’t.)
Fighting games usually work this way. In the DOA series the steps up and down in difficulty are so drastic you can usually feel them a few seconds into the fight. (You’ll use the same strategy as the previous fight where you were defeated, but suddenly the CPU foe will stop countering all your moves, stop evading throws, and will use less damaging combos when you make a blunder.) A few shooters have this auto-difficulty built in. I never saw it documented, but I very strongly suspect Max Payne used it as well. Unreal Tournament has an option to have the bots auto-adjust, and I always turn it off.
The delta between the most skilled players and the newbies is massive, even for games which don’t look terribly complex. Providing a fun experience for everyone is a serious challenge for a game designer. You can offer two user-selectable difficulty levels and exclude a lot of people. Or you can offer a ton of difficulty levels and let the user grope around until they discover one that feels right. I can see the allure of auto-adjustment, but it’s solution that will displease nearly everyone.
Topic for discussion: Have you ever ramped up the difficulty by some unconventional means? Limiting yourself to certain weapons, modding the game, not saving, or other handicaps not directly provided or understood by the standard game?
Shamus Young is an old-school OpenGL programmer, author, and composer. He runs this site and if anything is broken you should probably blame him.