In [Magerkurth, C., Engelke, T., Memisoglu, M. "Augmenting the Virtual Domain with Physical and Social Elements"], it is stated that in simple games, such as Backgammon, it might be better not to implement the complete game rules into the digital world. If no computer logic forbids rules' variation, it would be easier for players to modify them in the context of a particular situation, and to develop novel house rules. On the other hand, when the game rules are complex, e.g. the winning condition takes into account many parameters, computational supports would be helpful, which can not be achieved without letting the system be aware of the game rules. These two aspects motivate us to explore the possibility of redefining dynamically the game rules at a high abstraction level. Dynamic rules redefinition is particularly suited for games that exist in many different variants e.g. the Mancala games (Awele, Awale, Ayo, Ourin, Wari, etc.). Players should be allowed to experiment one variant by simply redefining the corresponding rule before a game session, or even invent a new variant defining a new rule without prior knowledge of the underlying programming language but using a high level user interface. In our opinion, letting the human natural intelligence and imagination interact in a natural and direct way with the artificial intelligence embedded in the virtual world could create a new synergy and lead to a novel interaction paradigm for games. The player will be able either to play a game in the standard way defined by the game creator either to free her/his imagination and experiment variants or fully novel ideas without being limited by the constraints of the physical world (materials, definition of new elements, etc.) and avoiding the lack of flexibility of the digital board by letting the system be aware of the new rules and of the role of the new objects (virtual and tangible) in a natural way. An interesting use case could be an interactive test environment for game developers where novel ideas and game mechanics could be dynamically and easily evaluated. Our major objective is to define a general framework in which users can easily and dynamically redefine game rules in a natural way.