Scaling issues in multiuser agents

Agents such as Julia can sometimes avoid this because of their environment: In general, MUDs also scale badly with large numbers of users. This means that Julia never has to contend with a room of more than a handful of characters, or a database of more than a few thousand rooms or characters.

On the other hand, agents such as Ringo find this problem more difficult to avoid. As the number of users in Ringo's ratings database grows, the amount of work required to find a match grows proportionally. Further, as the number of users grows, the rate at which they make queries of the database grows, too. This quadratic growth will eventually exhaust almost any reasonable implementation unless steps are taken either to cluster groups of users together (so that the smaller numbers of clusters can be searched instead) or to distribute the application among multiple servers (which begs the question of how to do the distribution).

Other solutions have certainly been proposed and adopted. For example, hierarchical organization of the entities, as is done with resource records in the Internet domain name system or with newsgroup topics in the Usenet, can help to reduce the inherently quadratic problem into a logarithmic one. However, such approaches depend on some inherent organizational principle that is established in advance, which is neither always optimal nor always convenient (for example, consider the number of crossposted Usenet articles, a clear indication that single-inheritance hierarchies are not necessarily a good match to the underlying topic space).