A framework for adaptive game systems was produced by the same research group that produced the USE model. The AGS framework includes additional detail in the process of adapting the game system to promote an enjoyable player experience. The AGS model Figure in 3.2 is composed of two main sections: an oine stage outside portion and an online stage contained within the dotted rectangle. The oine stage
stores information relating to player types discussed in Section 2.2 and can be used to identify and adapt to perceived preferences of the user's typology. The online stage would also include information pertaining to general gamer preferences to be utilized with their user typology. Knowledge of general gamer preferences would provide hints as how to perform adaptation based on a user's personality and player type.
Figure 3.2: Adaptive Game System model illustrates the feedback loop required for producing adaptive video games, which monitors then modies the game in relation to expected player needs. [15]
The online portion of AGS is composed of a four stage feedback loop. In the initial stage player performance is monitored. This information will be utilized to determine whether the game needs to make adjustments. Monitoring player performance will be discussed further in Section 3.3. The second stage of the online AGS performs adjust- ments to match the system's perceived level of player desired complexity. Performing adaptation to the game world will be further discussed in Section 3.4. Thus the adap- tive system alters the game in a way which it believes the player is capable of handling without demotivating the player. The third stage of the online AGS measures how eective the adjustments were for meeting the proposed goal set in the second stage of the adaptive system. The last stage of the online AGS is remodeling informa- tion relating to the player's user typology. The adaptation occurred in response to
a perceived discrepancy between the player's needs and the game. The adjustments made in the second phase are based on the current user prole and suspected user preferences. If the adjustments do not produce the desired response, it may indicate that we have not correctly identied a portion of the player's user typology or the modied factors do not have the expected eect. If the adjustments are successful it reinforces our believe in the current player information, and provides an example for future use. Additionally player information may change, as player types change throughout the course of the game, as seen in Section 2.2.3 on Bartle's player types. Thus the player model constantly needs to be remodeled, as it is also susceptible to concept drift, which occurs as part of a natural progression in strategy or behavior.
3.2.1 Auto-Dynamic Diculty (ADD)
As previously mentioned Auto-Dynamic Diculty (ADD) is the adaptation of the challenge of the game to match the abilities of the player. If the player is struggling to progress during game play, the level of challenge may be reduced; likewise if the player is progressing too easily, it may adapt to provide more challenge. Matching the challenge of the game to the abilities of the player will increase their interest in playing the game.
Although matching challenge to the player's abilities is the main objective of ADD, it may also provide other benets. ADD may mean providing more specic and helpful feedback when the player is lost or uncertain of the next step. It would recognize repetitive player locations and lack of progress in the game, and then provide additional hints. ADD could also be used to detect imbalances in game dynamics. If the player is consistently using the same strategy it may be the result of an unforeseen exploit, which may result in successful play of the game, which depending on player type this could ultimately cause dissatisfaction due to the repetitiveness of the game. An ADD system would be able to recognize this overuse and could alter the game so that a particular strategy is less successful [13].
A system that performs ADD should consider two main points [30]. The rst point is that the player should not be aware that ADD is adapting the game. If the
player knows that the system is adapting to their playing style, they are likely to react negatively towards it, blaming the ADD system for their failures. Alternatively they may attempt to exploit the system by initially playing poorly until the challenge of the game decreases. Knowledge that the system is adapting hinders a player's immersion and sense of achievement, as it directs the players focus away from the game and towards the ADD system. Thus it is benecial to altering variables of the game that are not viewable by the player. Secondly, the ADD system should not make large incremental changes. This helps hide the modication process from the user, as well as helps avoid drastic changes in diculty that may cause the player to become frustrated or bored quickly. Using small incremental changes also helps avoid the problem of players exploiting the system by performing poorly at the beginning of the game to lower the diculty.