There's been quite a bit of effort put into platformer level design in recent years. Hoffstein wrote
that "platformer level design cannot duplicated by any method other than careful design, calibration and testing." However, Game Maker's very own Spelunky
by Derek Yu
and Cloudberry Kingdom
by Pwnee Studios
clearly challenge this notion. There are countless examples of games that utilise procedurally generated environments to enhance gameplay. I feel that Hoffstein has unwittingly fallen into the trap laid by Clarke's First Law: "When [a distinguished scientist] states that something is impossible, he is very probably wrong."
Spelunky and Terraria avoid the limitations of their generation algorithms by having an entirely deformable environment. If there is a problem with the orientation of obstacles in the environment, the player has sufficient agency within the game world to simply circumvent that problem. Note that this does not preclude an enjoyable experience! Cloudberry Kingdom takes a more rigid approach, an approach that arguably leverages more artifical intelligence (though the question is open as to whether or not the design itself is better or more enjoyable). In a similar vein to Mario ROM hacks
, especially the harder ones such as Kaizo Mario World
, Cloudberry Kingdom creates levels that are a complex interaction between rotating hazards, bottom pits and moving platforms. This, clearly, is a sublime example of artifical design. There's a good video
on their Kickstarter page
that show how unbelievably hard, and detailed, the game's levels can be.
How does such a system work? Platformers boil down to two basic activities: running and jumping. So long as both running and jumping follow deterministic rules, you can successfully model a hypothetical series of actions that is guaranteed to get the player from the start to the finish. In a way, you are generating a series of key presses that solve a time-dependant puzzle. Even the more complex platform abilities: wall-jumping, spin-dashing, spin-jumping and butt-pounding can all be modelled. Cloudberry Kingdom's generator simulates action after action as a string of events that must be executed at the right point in time in order to proceed. In a manner of speaking, it's not dissimilar to Guitar Hero
The levels created by Cloudberry Kingdom's algorithms warranted a 68% approval rating on MetaCritic
(at the time of writing). As with many games on the fringes of the games industry, it appeals to a niche audience of enthusiasts. 68% is not an impressive score. Indeed, the semi-procedural Rogue Legacy
by Cellar Door Games
scored a 85% on MetaCritic
. The Destructoid review
does a good job of covering the main flaws: bad music, inconsistent graphics, unnecessary storyline and some uninteresting character abilities are all cited. However, none of these flaws are a direct product of the level generation algorithm.
The Edge Online review
claims that the levels are "cold, sterile gauntlets, straight off the randomised production line." There are two issues here:
1) It is highly unlikely that two players will ever experience the same level or event. Rogue Legacy and, the structurally similar, The Binding of Isaac use pre-designed rooms and pre-designed bosses to create a sense of continuity and shared experience between players. These two games are eminately discussable: conversations that begin with questions such as "What's the best tactic for Khdir?" or a statement like "The Bible is so handy for spike rooms!" are indispensible tools for creating positive emotion around a game.
2) The game does not feel natural, human and organic. It would seem that there is necessarily only a single solution discourages players. The enduring popularity of Rogue-likes indicates that procedural generation is a viable option for games with limited environments. Cloudberry Kingdom, in its pursuit of precision platforming, loses a great deal of character without the multiple paths and dead-ends of a Rogue-like.
Both these issues are a shortfall of the game design rather than the actual technology. Whilst the design decisions made are consistent with the initial scope of the game (to create an "artifical intelligence capable of doing the impossible: designing insane Mario levels."), they are design decisions that inevitably limited the appeal and scope of the game. Furthermore, these are two problems that can be rectified in a relatively straight-forward manner. Firstly, to create recognisable design elements that are consistent between players; secondly, to create a game world that can structurally and thematically accomodate multiple paths and dead-ends to break up gameplay.
Hoffstein has not been disproved yet. I believe it is only a matter of time before games catch up to what humans can design