Dynamic difficulty is a terrible idea. It reduces the potential difficulty-space of the game to two basic cases: Always too easy and always too hard. It means that if you lost, it's because the game chose to make you lose, not because of an effect on your choices, and if you win, it's because the game chose to let you win, not because of your skill.
Every couple of years I'll play a game with this new, innovative idea, and it's still the same story. It's sort of fun for a bit, until your brain catches up and realizes that as meaningless as most games are, this one is even more so. It's not a new idea and if it were going to work, it would have been made to work by now: http://www.intellivisiongames.com/bluesky/games/credits/spac...
(No, there is no in-between "just right". Simple control theory shows how you can't hover at the "just right" point; as you are at "just right", the system must dynamically adjust the difficulty above your level (or you even just hit a bad patch) and cause you to fail, at which point you're back at too easy. There's nowhere near enough data to get to "just right" even in theory.)
And I have a pretty extreme distrust of "Yeah, I know thousands of smart people have tried and failed, but I'm still sure it's a great idea!" See also: General purpose visual programming, 3D desktop or chat interfaces (and I mean true 3D, not "using 3D hardware for nice effects), "I'll just wrap a natural language interface around this and it'll be easy to use!".
Certainly both of our rules fire off, but for every time your rule hits, mine hits several thousand times.
Case with coin op games: Games around the year 2000 was frustrating to play. A skilled player would run into impossible to resolve situation and lose a "life". There was no way to play for more then about 1 minute without losing and no way to complete a game on a single credit. From the players perspective, the game was cheating and playing too skillfully was a demerit.
Recent coin op games seem to have moved away from this. It's possible to complete a level with one credit. The game board remains the same and the difficulty is selected by the board type or player specs.
I think in general that games shouldn't change difficulty mid board. To the player, it becomes a disincentive to play too well. Let the player decide what challenges they want to face. It should at least be an option to toggle.
I have seen dynamic difficulty done well exactly once: in an obscure freeware game called Quadnet by Brainchild Design. It is a simple arcade game that focuses on honing a single skill. There are three lives. Mastery at the game means speed in accomplishing the game's objective; it ramps up the difficulty during the first life in response to the player such that after about 5 minutes you've reached your max and you die. It then spends the next two lives very slowly rocking the player back and forth over that difficulty threshold they can't pass.
It works because there is no penalty for skill -- better players are guaranteed to get more points and faster -- and because it is true to the game: the advance in difficulty works to train you at the game's main skill, find your comfort level, and they try to push you past it.
In general, dynamic difficulty is done wrong in two ways: it changes the fundamental game or cheats, distorting the landscape players are trying to conquer, and it does not reward them with points or bragging rights for having triggered the harder mode.
Comments
Video game publishers also chicken out and add a difficulty level selector at the beginning of the game, rather than dynamically adjusting the level.
Dynamic difficulty is a terrible idea. It reduces the potential difficulty-space of the game to two basic cases: Always too easy and always too hard. It means that if you lost, it's because the game chose to make you lose, not because of an effect on your choices, and if you win, it's because the game chose to let you win, not because of your skill.
Every couple of years I'll play a game with this new, innovative idea, and it's still the same story. It's sort of fun for a bit, until your brain catches up and realizes that as meaningless as most games are, this one is even more so. It's not a new idea and if it were going to work, it would have been made to work by now: http://www.intellivisiongames.com/bluesky/games/credits/spac...
(No, there is no in-between "just right". Simple control theory shows how you can't hover at the "just right" point; as you are at "just right", the system must dynamically adjust the difficulty above your level (or you even just hit a bad patch) and cause you to fail, at which point you're back at too easy. There's nowhere near enough data to get to "just right" even in theory.)
I haven't personally played it, but people I respect have been saying that the Left 4 Dead series is actually doing a good job of this.
(also, as a general rule I have an extreme distrust of "nobody I know of has managed this yet, therefore no-one anywhere ever will")
And I have a pretty extreme distrust of "Yeah, I know thousands of smart people have tried and failed, but I'm still sure it's a great idea!" See also: General purpose visual programming, 3D desktop or chat interfaces (and I mean true 3D, not "using 3D hardware for nice effects), "I'll just wrap a natural language interface around this and it'll be easy to use!".
Certainly both of our rules fire off, but for every time your rule hits, mine hits several thousand times.
There's a danger to having a dynamic difficulty.
Case with coin op games: Games around the year 2000 was frustrating to play. A skilled player would run into impossible to resolve situation and lose a "life". There was no way to play for more then about 1 minute without losing and no way to complete a game on a single credit. From the players perspective, the game was cheating and playing too skillfully was a demerit.
Recent coin op games seem to have moved away from this. It's possible to complete a level with one credit. The game board remains the same and the difficulty is selected by the board type or player specs.
I think in general that games shouldn't change difficulty mid board. To the player, it becomes a disincentive to play too well. Let the player decide what challenges they want to face. It should at least be an option to toggle.
The danger is doing it naively, and winding up with Rubber Band AI: http://tvtropes.org/pmwiki/pmwiki.php/Main/RubberBandAI
I have seen dynamic difficulty done well exactly once: in an obscure freeware game called Quadnet by Brainchild Design. It is a simple arcade game that focuses on honing a single skill. There are three lives. Mastery at the game means speed in accomplishing the game's objective; it ramps up the difficulty during the first life in response to the player such that after about 5 minutes you've reached your max and you die. It then spends the next two lives very slowly rocking the player back and forth over that difficulty threshold they can't pass.
It works because there is no penalty for skill -- better players are guaranteed to get more points and faster -- and because it is true to the game: the advance in difficulty works to train you at the game's main skill, find your comfort level, and they try to push you past it.
In general, dynamic difficulty is done wrong in two ways: it changes the fundamental game or cheats, distorting the landscape players are trying to conquer, and it does not reward them with points or bragging rights for having triggered the harder mode.