Terra Nova Welcomes Jim Purbrick

One of the fortunate few who have been able, at this early stage, to acquire both academic and industry experience in analyzing and implementing virtual worlds, Purbrick hails from Nottingham UK. Primarily a network technologist but also a designer, his credits include experiments with time-linking in virtual worlds (recording and replaying events), persistence mechanisms (the PhD thesis), and real-world projects like Dragon Empires and the now-retired Warhammer Online. In the latter, we would have been treated to Jim's version of a seamless load-balancing technology - a holy grail for us lag-tortured players.

You can see a collection of Jim's writings at this link. Welcome!

Comments

[Betsy Book]

1.

Welcome Jim! I look forward to reading your comments on Terra Nova.

[ren]

2.

Welcome aboard :)

[Dave Rickey]

3.

Welcome to TN, Jim. If I could ask a technical question that occurred to me reading the paper on MASSIVE-3: When you're resolving a spacially oriented event, is the distance it propagates fixed, or based on the magnitude of the event, the sensitivity of the observer, some combination, or something I can't think of? The paper didn't seem to make it clear.

Also, does MASSIVE-3 have a means for resolving interests that are spacially based? For example, the team in your stadium example, can the agents only communicate via the spacial system, or can they have a direct channel?

--Dave

[Dave Rickey]

4.

Welcome to TN, Jim. If I could ask a technical question that occurred to me reading the paper on MASSIVE-3: When you're resolving a spacially oriented event, is the distance it propagates fixed, or based on the magnitude of the event, the sensitivity of the observer, some combination, or something I can't think of? The paper didn't seem to make it clear.

Also, does MASSIVE-3 have a means for resolving interests that are *not* spacially based? For example, the team in your stadium example, can the agents only communicate via the spacial system, or can they have a direct channel?

--Dave

[Cory Ondrejka]

5.

Welcome, Jim. Absolutely fascinating thesis that I'm looking forward to reading more carefully.

[Jim Purbrick]

6.

Thanks for the warm welcome :-)

Nice to have some questions about MASSIVE-3 too! Events in MASSIVE-3 are propagated to every agent interested in the locale in which the event occurs. There are no restrictions on which locales an agent can be subscribed to. An agent with lots of CPU and bandwidth resources might subscribe to more locales than one running over a dialup connection. Locales don't need to be adjacent either. Entering a CC TV room might subscribe you to a bunch of locales in a distant car park for example. There are a number of general purpose policies in MASSIVE-3, which subscribe agents to the N-nearest locales or which use a cost benefit mechanism to subscribe an agent to a set of locales.

Locales needn't be spatial, so one way to get chat channels or other non-spatial communication mechanisms would be to set up chat locales, but in practice there are a lot of other technologies that can be used for non-spatial communication along side a virtual world infrastructure.

In general agents supply an interest expression which the framework uses to subscribe the agent to publishers. Virtual world literature calls this interest management or awareness management, but it's really just a special case of publisher/subscriber mechanisms.

One potential problem with locales is their granularity: when you subscribe to a locale you learn about everything going on in it. With MASSIVE-3 we added aspects on top of locales, so you could subscribe to certain aspects of a locale. With Warhammer we allowed agents to subscribe to individual items in the world, which gives the granularity you need for dealing with things like invisible characters.