Challenge In Games Is Not A Dirty Word

Thanks to GamerDad for their editorial discussing why there should be more difficult-to-complete games out there. The piece takes difficulty complaints regarding F-Zero GX for GameCube as a starting point, saying "This isn't the first time a top quality, high profile game has kicked people in the butts with challenge and it won't be the last. This kind of challenge is good for gaming and we need more games like it." The article goes on: "Players have grown accustomed to difficulty levels that are far too easy and I think it's contributing to their boredom with many games", but also cites specifics: "The most important thing about challenge, and it's one that F-Zero GX gets right, is that the game must let the player know it was their fault that they lost."

Comments

Re:Adjustable difficulty levels

[whee]

And F-Zero has just that. From the GameSpy review:

But is the game too difficult for most? That's pretty debatable. I personally found the difficulty curve pretty harsh -- particularly on the harder difficulty settings -- but F-Zero has always been about the overall challenge.

I don't understand this reviewer at all. When the difficulty is set higher, the game actually gets harder! What a concept! This amazing technology should be implemented in as many games as possible.

I think gamers are just getting used to games that are nothing more than busy work and no challenge. Obviously, a few hours of NetHack could solve a lot of problems.

On a Slight Tangent

[NJVil]

When will programmers stop equating difficulty with the computer "cheating" to make it better?

As one example, even on the easiest level of CivIII, the computer players somehow, magically, know more about the map and areas they cannot possibly have seen (even by trading maps with each other). Only by clamping down on the computer players' ability to produce units/buildings does it rein in its knowledge. At the higher levels, not only does it know more about the map than a human player ever could, it outproduces you dramatically.

Would it really have been that difficult to come up with an AI that did not cheat by violating the fog of war? I could understand it if at the highest levels this happened, but when it's pretty obvious on the easiest level, it shows a lack of interest in working on making the AI truly challenging, but rather taking a one-size-fits-all approach and altering the difficulty by manipulating non-AI facets of the game.

Re:On a Slight Tangent

[FreekyGeek]

Would it really have been that difficult to come up with an AI that did not cheat by violating the fog of war?

Yes.

People who complain about the difficulty of video games (either too tough or too easy) are almost always unaware of 1) how tough it is to have variable "difficulty" settings in a game, and 2) the real-world restrictions that preclude what might seem like "common-sense" solutions to these problems - restrictions that game developers have to face.

First, what does "difficulty" mean? It's a completely variable term, since everyone has different skill levels, so what seems easy to some seems difficult to thers. There's no objective measure of "difficulty". There's no such thing as a "difficulty meter". In some games, difficulty is easy to adjust - in shooters, you give the player more hitpoints, stronger weapons, or powerups. Or you give the enemies less armor, or whatever. But what about an RPG? A strategy game? Should game designers have to come up with three or more seperate puzzles for each place in the story that calls for a puzzle? How does one gague how easy or difficult a puzzle is? Some people can solve word scrambles in a second, while others can hardly do them at all. Some people think very visually and have good spatial relations while others don't.

Second, there are a huge number of restrictions game developers work under - time and money, mostly. Every developer in the world would love to have the luxury of releasing games "when they're done". But unless you're Id, forget it. Sure, if you had a team of ten programmers with three years to do nothing but write AI code, you could come up with some very sophisticated AI for games, but that's hardly realistic. So games only have to have AI that's good enough to challenge most players most of the time. And often that means "cheating" by the AI. Ever tried to write AI? How would you design a game AI that could provide a good challenege to all players, with a variable difficulty level, without cheating, without spending ten years doing it, without requiring a supercomputer to do so?

Strategy AI is hard to write. If it was easy, the military would just write the Best AI Ever and let it run all the wars. Stategy, even basic strategy, is an incerdibly complex subject with a billion variables. Don't believe me? take a look at Chess, a game with extremely simple rules. It has taken decades of time and many supercomputers to write software to beat a human at chess, and even then it is only through brute force, by analyzing every possible move and permutation. Ever wonder why the game of Go doesn't have too many good computer versions? because writing AI for it is so damn hard. he rules are evn simpler than Chess, yet Go makes chess look like tic-tac-toe in terms of strategic complexity.

Second, there are very serious hardware limitations. Even on fast computers, AI chews up a LOT of CPU time. Programmers have to share the CPU time between all kinds of tasks, and the AI can't chew up 90% of the cycles. There's no point having a near-sentient AI if you the rest of the game runs like molasses. And as much as us high-end, hardcore gamers hate to admit it, a HUGE portion of the gamer audience still uses pretty old PCs. And they buy games, too - a lot of them.

As is so often the case, this issue is far more complex than it appears. Game design does not happen by itself, and every hour spent tweaking difficulty levels is one less hour spent making the game better in other ways. Every cycle the AI uses makes the game that much slower and that much less accessible to people on low-end machines. Every minute of programmer time spent on any of these things costs money and adds to the schedule.

In a perfect world, we'd have unlimited game developement budgets and cycles. Until then, compromises have to be made.