Difficulty Regulation

Lumines got me again. I hadn't played it for a while, and then yesterday I decided to sink my teeth in once again. It had one mode I hadn't really tackled before: a puzzle mode where the aim is not to score points, but to form various shapes of one color from the blocks. The reason I hadn't tried it before was that I had to actually check the internet to understand how I'm supposed to form those shapes. So there's a quick lesson: sometimes instructions are simply needed. But that is not the topic today. There is a reason I mentioned Lumines and the puzzle mode though, and we'll get to it soon enough.

Difficulty regulation strategies in games are important tools for keeping the player in flow. One quite obvious strategy is to simply include multiple difficulty modes to play the game in, allowing the player to choose a mode he deems suitable for himself. Of course, one problem is: how can the player know which one to pick? Sometimes descriptions can help, such as BioShock's "Choose this if you have never played first person shooters before" description for its easy mode. This is a relatively simple strategy to implement and has the benefit of hugely improving a game's life cycle for advanced players. The only thing a game designer needs to worry about is making the correct adjustments to make the feel more difficult but not unfair. Also, players might need some encouragement that after finishing the game, their skills are on the level needed to tackle the more difficult mode.

Difficulty increase built into progression in the game is most likely the most common way. The next level is typically more difficult than the one before it. Yet it is not always easy to design a suitable difficulty ramp. In many games, some of the most challenging moments are somewhere in the middle - towards the end, player skills and character powers are becoming too big for the challenge. Another thing is that the difficulty curve should not be a ramp. I think it's important to drop in tasks that are a bit more lenient than the previous ones. This way the player, after mastering some tough challenge, can actually feel that he has gotten better, as problems that are still much harder than the first ones are becoming easy. This is exactly how the puzzle mode in Lumines got me so addicted. Every time I succeeded at a hard problem, the next ones felt like a breeze and got me to play more and more. When the next harder problem hit, I already had felt the satisfaction of success, and I was mentally prepared to face it.

Finally there's dynamic difficulty regulation. In this case, the game uses some mechanism to figure out when the player is not doing too good, and then proceeds to make the game easier. Sometimes this difficulty reduction is invisible, affecting only numbers inside the game's engine or reducing the effectiveness of artificial intelligence - other times it is more visible, when the game reduces the challenge itself by introducing fewer enemies for example. So essentially the game cheats for the player's advantage. This is quite typical in tabletop role-playing games as well when the game master makes his die rolls in secret, and can then freely adjust the result to keep the game more enjoyable. The problem with this approach is that some players might indeed feel cheated out of the challenge. Frankly, they don't appreciate your concern. Personally I think there should always be the option to turn dynamic difficulty regulation off.

Finally, multi-player games where players compete with each other are difficulty regulated by player skill. In these games, one important aspect is to try and match players with somewhat equal skill levels to keep everyone in the flow zone (i.e. not bored and not frustrated).

So, these are the most typical ways games do their best to keep players in flow. Can these strategies be used for real-life applications and tasks? It does seem to highly depend on the task. Difficulty is typically in the task itself, so we really cannot touch it. What applications can do though is to suggest a good task order for the user, based on past performance and statistics. This way users can more easily find tasks that match their skills. This approach does seem more suitable for tasks that are part of hobbies - at work the list of available tasks is often narrow, and there is not much choice in what to do and when.

One other way applications can do difficulty regulation is the level of automation, which I have already discussed in an earlier post. At first, tasks can be done in less detail, relying more on the application to work its best guess magic. When the user becomes more familiar with the application and tasks, more and more control is placed in the user's hands. Of course, the user will need full capability to override the machine's decision at any time, regardless of "difficulty mode".

I guess that's it for today.