MUDs : Perception

The Combination of Art and Science

Descriptions, of objects, rooms, characters - indeed anything that the characters can sense - are of vital importance to a MUD. They are the characters' only means of sensing the virtual world around them, and as such must present as complete and useful a scene as possible. They must also be well-written pieces of creativity; artistic descriptions make a MUD enjoyable to play, since atmosphere is a vital factor in good games. The problem that this article addresses is that created when such creative descriptions are subject to influence from the observer's state of being. By "state of being" is meant the sensory abilities, or otherwise, of the character who is sensing the description. That is, given that different characters will have differences in their sensory capabilities (such as being blind, or having particularly keen hearing), the descriptions sent to their respective players should reflect these differences. The problem is how to achieve this while having each of these descriptions conform to an overall creative standard.

From the point of view of the players who must read the descriptions, the best solution is to have a separate description written for each and every possible combination of senses. However, practical considerations of space and creative effort must be taken into account, and so some method for the computer to devise its own descriptions is necessary.

The first obvious step in addressing the problem of variation in sensory capability is to allow some means by which descriptions can be divided into their separate sensory components. This would then allow for a description to be sent to a character which was composed only of those components which correspond to the senses the character has functioning. Thus a blind, but otherwise normal, character would receive descriptions featuring the olfactory, tactile, aural and taste components. As this stand, it imposes immense constraints on a builder's creativity, since each sense must be described in self-contained blocks of text. Where more than one feature is being described, this leads to a very broken description. After treating one sense on all the features, the description will then treat a different sense on all the features, and so on, rather than grouping the description into features, as would seem normal to most people.

To avoid this problem, it is possible to split a description both according to the feature described, and the sense covered. In this way, each feature described would be divided into a number of sentences describing each sensory component. When the description is compiled to be sent to a character, the required sentences for each feature are grouped together, resulting in a list of sentences for the first feature, another list for the second, and so on.

This method certainly makes the description follow a more logical ordering, but is still artificial; lists of sensory data, with at least one sentence for each sense, is not likely to convey any atmosphere. There is no unity is such descriptions, no linking of the individual sentences into a cohesive whole. Nor is there any mixing of senses within a single sentence. This is, unfortunately, a problem which remains unresolved to any satisfactory degree. It is unlikely that smaller units than sentences can be used, for in such a case the computer would need to be able to construct real sentences from them, in an `artistic manner'. Currently there is no way to happily marry creativity and science in the working of descriptions.

MUDs : Perception
Email: jamie@artefact.org.nz
Last updated: 16 July 1998