From Rogue to lootboxes: two faces of randomness in computer games

In this paper I try to look at two opposite ways that randomness is employed and, more importantly, evaluated in computer games. It seems fascinating that the same mathematical and statistical phenomenon could lay at the roots of some of the most acclaimed and despised design principles of contemporary games. On the face of it, this may not seem to be paradoxical at all as it could be argued to be true of many other objects and processes. After all, the same knife could be used to prepare a meal and to hurt people. What makes the case of randomness interesting however is that this connection between its heralded and criticized sides is never examined and that they are often treated as completely separate phenomena.
One initial clarification that may be needed is that the paper does not cover instances in which randomness has been used only during production of the game. In other words – I am interested only in cases where random values and patterns appear during the game execution. To see this difference, think of any game which contains a map generated in random or partially random manner via some additional development tools, but where the game itself uses the map as a fixed asset. To use a concrete example, compare Just Cause 2 and Minecraft. The former contains a fixed map which has been (at least partially) procedurally generated by the developers (Blomberg 2013), the latter contains a map which is generated by the game itself, whenever it is started anew.