At the recently held Danish National Championship in Retro Gaming video recordings of different players playing the same level in two games were created. Inspired by the Averaging Gradius experiment, I have taken these recordings and combined them into two video clips showing how differently players are able to complete a level.
The first game is R-Type, with four different players:
What I think is most obvious in this clip, is how little difference there is between the players’ paths through the level. This is likely due to the static and narrow level architecture, which effectively guides the movement along a single path. Enemies are also taken down in almost the same sequence. I believe this can be attributed to the fact that enemies appear one at a time, which creates a natural kill sequence.
In terms of skill, one of the players seems slightly less adept against the end boss, but all players use the same strategy; two charged shots kills the boss. This is likely because the players saw each other play, and hence learned from each other.
The second game is DonPachi, with two players:
The reason for using only two players in the video, is that the secondary scroll direction is too dominant making it hard to separate the movement of more than two players. R-Type has a secondary scroll direction too, but it is rather weak and doesn’t clutter up the video too much. In any case, the idea is not to provide statistically relevant data – just to explore the use of this video technique.
What I think is most interesting in the DonPachi video, is that the difference between the movement paths seems slightly higher than in R-Type, and on several occasions the enemies are taken down in different sequences. This can be explained by the total lack of level architecture – the players can fly anywhere they want. Also, some enemies appear simultaneously at different sides of the screen giving the players more ways to approach the challenge.
The most important difference between the two games, besides their scroll direction, is their challenge frequency and amplitude. This could probably be counted and calculated, but just by looking at the videos it is obvious that the players has a lot more to see to in DonPachi – it is called a manic bullet hell shooter for a reason. While these challenges are small and quick, there are many of them and also a lot of ways to overcome them. The players are given more room for creativity in DonPachi, which also explains the slightly higher difference in movement paths.
While there are many flaws in this experiment, it does show that overlaying video clips to average the gameplay of different players can grant some interesting results. First, it becomes very easy to compare how different players play through a level. As I see it, a high difference is good, because it shows that the game allows for several ways to play. Second, when used on two games in the same genre the differences between them seems to stand out very clear.
What these videos doesn’t show, is how the players learn to play the game. These recordings are from players who has already learned to play the game – they all play through the level without loosing. It does show that they have learned to play in different ways, but how they reached their current skill level is not seen in the videos.
In the end, I believe this technique is most useful for observing the amount of different ways to play. That is, observing gameplay depth.