Floyd, as currently implemented, provides the socially useful services he is intended to fulfill. Users can easily download and install their own Floyd. Bringing a new Floyd into the network is simple with the database synchronization protocol. Using the Floyd network to discover where people are or just to see what ccr worlds are running is easy and effective. There are some obvious implementation wrinkles, none fatal. It is clear that the current design of the Floyd broadcast network will not scale well to hundreds of agents: designing a scalable network where each node is unreliable would make an interesting research project.
Given that the Floyd network exists, one important question is the impact Floyd has on ccr social dynamics. To what extent does making it easy to find people encourage person to person contact? Does easy global communication discourage people from leaving their homes to seek each other out? How serious is the current lack of privacy in the Floyd design? As more people join the ccr universe, the influence of Floyd will become more apparent. It is important to make sure that new users see Floyd as an optional thing to add, rather than part of the landscape of the ccr system.
Another possible effect of Floyd is to provide more impetus for a service-driven economy in ccr. Currently, the infrastructure for a digital cash system in ccr is just being created: once currency becomes widespread in ccr, how will people spend and earn money? Floyd is an obvious service one could sell, charging money to have Floyd answer questions. This would serve as an extra encouragement for people to set up their own Floyds. Floyds could also require payment from other Floyds for some types of information: for instance, one Floyd could charge a fee to let another learn from him. With enough participation, Floyd could create a fairly complicated information economy. This economy should be intrinsically interesting, but is it desireable to have services available only to those who can pay?
Another possible use of Floyd is to use his communication capabilities to distribute updates to Floyd's code itself. One Floyd could broadcast to the network the source code for a new version: other Floyds would then choose whether or not to install and run the new version. In addition, it might be possible to give new users a simple Floyd bootstrap, one that has just enough capability to then fetch the full Floyd code from the network and install it. Obviously, the security implications of running source code other people give you are severe. Full safety is impossible, but a digital signature protocol would at least provide accountability.
The Floyd broadcast network could also be adapted to provide many other global information services. A simple application would be to have Floyd help people negotiate netdoors: a new person could sign on to the Floyd network, broadcast a request for a netdoor, and automatically create new doors to anyone generous enough to offer them. Floyd could be used to send a message out to someone wherever they might be in ccr: broadcast the message out on the Floyd network and have the first Floyd who sees the recipient deliver the message and tell everyone else the message has been handled. Individual Floyds could wander around their local domain creating a map of their surroundings: a smart agent could then collect each of these local maps and then assemble them into a global view of the ccr universe.
Local communication restraints are a necessary part of our environment: they reduce both bandwidth and noise. At the same time, people naturally have requests they want to make that require global information: things as mundane as ``what worlds exist'' to as complicated as ``what is the layout of the entire universe?'' Floyd, provides a substrate upon which global information can be exchanged, thereby creating a powerful distributed database.