Dynamic Difficulty Adjustment
In a game with dynamic difficulty adjustment, the level of difficulty (or challenge) the the player faces is modified during the gameplay experience depending on the players personal ability. If the players abilities are very high, the game will become increasingly harder, or easier if the players abilities are not very high. DDA can be technically resolved in many different ways, in particular since the definition of success varies from game to game. This concept is also known as Adaptive Difficulty.
Existence of this DDA implies that there is some way of keeping track of the degree of success a player is having with a game. These can be multivariate and include parameters such as time taken to complete a level, number of lives lost, etc. The difficulty can be adjusted by modifying things such as making enemy entities more/less impervious to damage, adjusting the damage inflicted by weapons, altering the sequence/number of game levels, providing additional feedback to alert the player, etc.
In game design terms, dynamic difficulty is useful in ensuring that each player has a game-playing experience that is tailored to his or her skills and abilities. This is often referred to as the sweet spot where a game is neither too easy so as to be boring or too hard so as to be frustrating.
In some games, particularly racing games, dynamic difficulty adjustment is used to keep things competitive. If the player falls too far behind, the AI-controlled cars will often slow down noticeable to allow the player a chance to catch up. If the player is too far ahead, AI cars will sometimes "teleport" forward. In the context of racing games, dynamic difficulty adjustment is often referred to as rubber-banding, because the idea is that the group of vehicles racing should stay pretty close together. Any car that deviates from that, "stretches" the rubber-band, and is either "forced" back, or the other vehicles speed forward.
Dynamic difficulty adjustment does not refer to the learning process a player might go through wherein he becomes better at playing a game. In this case, the difficulty of a game has changed with respect to the player, but the game itself remains the same.
In the first person shooter Max Payne, a system was implemented that explicitly measured the degree of success a player was having with the game. Amongst other things, it automatically adjusts the strength of the enemies and the amount of auto-aim assistance based on the players performance. The adjustment of the difficulty is transparent to the player.\
In Crash Bandicoot a player is awarded by the sudden appearance of Tiki masks if the player fails to clear a certain section too many times in a row. A Tiki mask is basically a shield that allows the player to take damage (once) without losing a chance. This is considered a weak example because the help the player received is extremely overt as well as applying only to certain specific challenges the player may be facing. This would be in opposition of facing a general change in the amount of challenge.
Crimson Skies (PC)
In Crimson Skies (PC), players are given the option to skip a mission after three failed attempts. This is a weak example because the difficulty of the game is not adjusted, but in the overall context of the game the player is allowed to proceed as if a certain challenge had been met.
Final Fantasy Tactics
In Final Fantasy Tactics, a player earns experience points from attacking/killing enemies and successively levels. In accordance with the player's level, the enemy monsters also level to compensate despite whether you are in the first area of the game or the last area of the game. This is a weak example because this kind of level adjustment has absolutely no relationship to the player's skill level; instead it is a gauge to ensure the game is always challenging regardless of whether it is too difficult or not.