Makindi slide runner spikes update#

An example of a card which causes information loss is Time Spiral, since it causes cards from the hand/graveyard to be shuffled back into the library. An example of a card which would give card advantage is Divination, which draws two cards at the cost of one. Also, we can attempt to quantify the loss/benefits of an action in terms of variables such as (1) life lost/gained, (2) damage dealt, (3) field advantage, (4) card advantage, (5) mana acceleration, (6) information gained/lost. In general, we can consider the effects of a play as an Aggro, Combo, or Control action. However, since the expected behavior of the Magic the Gathering, when considered as a system, is highly dependent on the behavior of the opponent, it is difficult to use Bayesian analysis to determine the correct play in a given situation.Ī computer should be able to manage Magic data from multiple perspectives.

Makindi slide runner spikes update#

In particular, Bayesian analysis allows the program to make decisions based only on a partial knowledge of the game state, and to update its decision algorithms depending on new information that becomes available. Many of the ways in which Magic is difficult to implement on a computer can be resolved, if only partially, through an aggressive use of Bayesian analysis. Additionally, Magic the Gathering is: (2) highly strategic, that is, players are seldom forced to do anything, and there are many points of time within each “turn” where players may act or react (but usually choose not to), further increasing the time complexity of the game tree (3) difficult to define, that is, as there are thousands of Magic cards in print, many of which have eccentric and difficult to implement rules that alter or create new aspects to the game state, it is difficult to represent and effectively analyze the game state (4) random in outcome, that is, games tend to be won or lost often based upon random chance, and in many cases strategy is less important than drawing the correct card. This will present problems to any computer trying to play a good game of Magic, although it can be mitigated somewhat by considering the fact that the impact of the information of card order within the deck on the outcome of the game is much lower than that of the information that is available to the players. These include the facts that it is: (1) closed in the information domain, that is, portions of the game state, such as the content of the players’ hands, is known by only one player, and, additionally, the order of cards within the players’ decks, which, overall, represents most of the information content of the game state, is hidden from both players.

Magic the Gathering, on the other hand, is, for many reasons, poorly suited to be played by an artificial intelligence. These factors combined make chess a good candidate for computer play. This lack of success can be attributed to a variety of vicissitudes, depending on the characteristics of the game being attempted.Ĭhess is a particularly auspicious candidate for computerized solutions, being that it is: (1) open in the information domain, that is, all aspects of the game state are known to all players at all times (2) highly tactical, that is, many if not most variations involve lines of play in which each player is forced to play from among a very small number of good-looking moves, and where all other moves can be clearly shown to lose, which decreases the branch factor of the game tree (3) simple to define, that is, each piece moves according to basic geometric factors and legal moves tend to remain legal over long periods of time (4) drawish, that is, games tend to produce draws and wins always involve capitalizing on the opponent’s errors. On the other hand, many other strategy games have not been successfully played by computers. Indeed, for some simpler games, such as checkers, computer programs have successfully “solved” the game – creating an algorithm that, along with a corresponding database of predetermined calculations, can provably determine the best move in any position in a relatively short amount of time. The ability of computer systems to play strategy games such as chess has presently surpassed all human players.

Chess, and other chess-like games, are a primary example of this. Recently, deterministic computer systems have been shown to be highly effective in the solution of various games of strategy previously thought exclusive to the provenance of human thought. Of course, the people out there who actually do Magic AI have a better idea how true these things are, but I think there a bunch of good observations. Very, very interesting analysis of Magic AI. (Here’s one of the more technical papers I got.