An Opening Response
I've decided to open my blog with a response to an article which I took exception to. I urge you to read it before continuing, but at the risk of misrepresenting it, I shall summarise here.
The article extols the virtues of shared experiences in MMOs, where players use their experience of playing the game to enhance their enjoyment of the social aspects of it. Indeed, I think it shows convincingly that experiences common to sizeable groups of players are quite essential to MMOs. However, it also contrasts shared experiences with algorithmic content as if it's some kind of opposite. Now, the author may be talking about a specific type of algorithmic content, but that's not made clear. It's not shown how algorithmic content supposedly undermines shared experience - it's just asserted that it does. Game flaws which are the result of bad design are attributed to algorithmic content itself, rather than the designers of that content. I think this displays a lack of understanding as to what algorithmic content is and completely misses how it can be used to great effect to enhance the player's enjoyment of RPGs in general, including MMOs.
Algorithmic content is independent of the shared/unique experience spectrum. Non-algorithmic methods can undermine shared experiences, and algorithmic content can be integrated into a game while preserving that essential facet.
To illustrate the first, consider a player entering a dungeon. This dungeon has had a very large number of different designs, each authored meticulously by a human, but each unique. When the players enter, one design is chosen at random for them to play. So they get to play an authored dungeon, with no algorithmic content, and yet the experience of each player will effectively be unique. Of course, this ignores the practical difficulties in authoring that many dungeons, but it illustrates the point that shared experience can be undermined without algorithmic content.
An objection to this could be that the insertion of a random number into the system makes it algorithmic. But this would only serve to show beyond any doubt that algorithmic content does not preclude shared experience – the vast majority of games involve random number generators, and players have many common experiences within them.
For the second point, Fallout 3 is an example of a game where algorithmic content sits alongside authored content quite happily. The monsters that spawn within dungeons vary with the level of the player's character. This doesn't stifle discussion of the dungeon – it can even enhance it by allowing players to share how their experience varied within the common structure.
It could quite rightly be pointed out that Fallout 3 is not an MMO. However, it does offer plenty of shared experience, as evidenced by the Fallout Wiki. It's also borne out in discussions I've had with friends and colleagues about our Fallout experiences. This co-existence of shared experience and algorithmic content makes it an entirely appropriate example.
I would accept that algorithmic content can in some cases undermine shared experiences, but it does not always do so. Also, authored content can be used in a way which undermines that same shared experience. Authored/algorithmic content is completely independent of unique/shared experiences.
I would go further and say that there is no clean distinction between algorithmic content and authored content. Anything authored for a game only makes sense within the very algorithms that make the game playable in the first place. For example, even a simple platformer can be made or broken by its jumping mechanic. The system could be parameterised and the control exposed to non-programmers, or it could be hard-coded within the game binary itself – either way could produce the same result. It could even calculate jumps based on how healthy or strong the player's character is. Is that authored or algorithmic?
This doesn't counter the criticism made in the article of those who get caught up with their 'amazing machines' and fail to focus on the player's experience. But then, this is the fault of the designers who forget that in any game worth its salt, the focus must be the player. Algorithms within the code are just as much a part of the game design as scripts or configuration files, and they can be used to great effect.
I would also like to counter the statement that good authored content is always better than algorithmic. I'm not going to say the opposite is true, but that the distinction is ill-defined, if it exists at all. Returning to the example of monster spawning in Fallout 3, the algorithmic nature of the spawning enables the player to roam much more freely, entering dungeons in whatever order they like. If they come to a smaller dungeon late, more difficult monsters can be spawned to keep it interesting. That just isn't possible with strictly authored content - the algorithmic nature enhances the experience.
The article also posits that authored content can be more complex than algorithmic because the author knows what the characters have been through. This very same Fallout example shows that is not always the case. In fact, the algorithm can have more information at its disposal than the author – it can know exactly what has happened to this particular player, and change things accordingly.
This brings me to address the core misconception the other article. Algorithmic content is authored. The rules are designed and the game works within them. If those rules are badly designed, the algorithmic content will suck. If they're well designed, the game can benefit enormously. The example is given of a newly spawned character getting 'ganked' by a dragon which has been enraged by other players killing its sheep. This shows that game design can be bad, not that 'algorithmic' content is inherently worse than 'authored'. Why not simply spawn the character elsewhere? Why not have the dragon find sheep elsewhere? Fundamentally, the rules of that system are badly authored.
Algorithmic content is not the evil that the article seems to paint it as. People using algorithmic content badly is the same kind of evil as people manually designing content badly. Not every aspect of manual design is appropriate for every game, and neither is every aspect of algorithmic.
The article extols the virtues of shared experiences in MMOs, where players use their experience of playing the game to enhance their enjoyment of the social aspects of it. Indeed, I think it shows convincingly that experiences common to sizeable groups of players are quite essential to MMOs. However, it also contrasts shared experiences with algorithmic content as if it's some kind of opposite. Now, the author may be talking about a specific type of algorithmic content, but that's not made clear. It's not shown how algorithmic content supposedly undermines shared experience - it's just asserted that it does. Game flaws which are the result of bad design are attributed to algorithmic content itself, rather than the designers of that content. I think this displays a lack of understanding as to what algorithmic content is and completely misses how it can be used to great effect to enhance the player's enjoyment of RPGs in general, including MMOs.
Algorithmic content is independent of the shared/unique experience spectrum. Non-algorithmic methods can undermine shared experiences, and algorithmic content can be integrated into a game while preserving that essential facet.
To illustrate the first, consider a player entering a dungeon. This dungeon has had a very large number of different designs, each authored meticulously by a human, but each unique. When the players enter, one design is chosen at random for them to play. So they get to play an authored dungeon, with no algorithmic content, and yet the experience of each player will effectively be unique. Of course, this ignores the practical difficulties in authoring that many dungeons, but it illustrates the point that shared experience can be undermined without algorithmic content.
An objection to this could be that the insertion of a random number into the system makes it algorithmic. But this would only serve to show beyond any doubt that algorithmic content does not preclude shared experience – the vast majority of games involve random number generators, and players have many common experiences within them.
For the second point, Fallout 3 is an example of a game where algorithmic content sits alongside authored content quite happily. The monsters that spawn within dungeons vary with the level of the player's character. This doesn't stifle discussion of the dungeon – it can even enhance it by allowing players to share how their experience varied within the common structure.
It could quite rightly be pointed out that Fallout 3 is not an MMO. However, it does offer plenty of shared experience, as evidenced by the Fallout Wiki. It's also borne out in discussions I've had with friends and colleagues about our Fallout experiences. This co-existence of shared experience and algorithmic content makes it an entirely appropriate example.
I would accept that algorithmic content can in some cases undermine shared experiences, but it does not always do so. Also, authored content can be used in a way which undermines that same shared experience. Authored/algorithmic content is completely independent of unique/shared experiences.
I would go further and say that there is no clean distinction between algorithmic content and authored content. Anything authored for a game only makes sense within the very algorithms that make the game playable in the first place. For example, even a simple platformer can be made or broken by its jumping mechanic. The system could be parameterised and the control exposed to non-programmers, or it could be hard-coded within the game binary itself – either way could produce the same result. It could even calculate jumps based on how healthy or strong the player's character is. Is that authored or algorithmic?
This doesn't counter the criticism made in the article of those who get caught up with their 'amazing machines' and fail to focus on the player's experience. But then, this is the fault of the designers who forget that in any game worth its salt, the focus must be the player. Algorithms within the code are just as much a part of the game design as scripts or configuration files, and they can be used to great effect.
I would also like to counter the statement that good authored content is always better than algorithmic. I'm not going to say the opposite is true, but that the distinction is ill-defined, if it exists at all. Returning to the example of monster spawning in Fallout 3, the algorithmic nature of the spawning enables the player to roam much more freely, entering dungeons in whatever order they like. If they come to a smaller dungeon late, more difficult monsters can be spawned to keep it interesting. That just isn't possible with strictly authored content - the algorithmic nature enhances the experience.
The article also posits that authored content can be more complex than algorithmic because the author knows what the characters have been through. This very same Fallout example shows that is not always the case. In fact, the algorithm can have more information at its disposal than the author – it can know exactly what has happened to this particular player, and change things accordingly.
This brings me to address the core misconception the other article. Algorithmic content is authored. The rules are designed and the game works within them. If those rules are badly designed, the algorithmic content will suck. If they're well designed, the game can benefit enormously. The example is given of a newly spawned character getting 'ganked' by a dragon which has been enraged by other players killing its sheep. This shows that game design can be bad, not that 'algorithmic' content is inherently worse than 'authored'. Why not simply spawn the character elsewhere? Why not have the dragon find sheep elsewhere? Fundamentally, the rules of that system are badly authored.
Algorithmic content is not the evil that the article seems to paint it as. People using algorithmic content badly is the same kind of evil as people manually designing content badly. Not every aspect of manual design is appropriate for every game, and neither is every aspect of algorithmic.
Comments
Post a Comment