Layers of Peer-to-Peer Worlds

The first time I experienced a virtual simulation was in 1985. The military systems on which I was being trained were usually nothing more than big interactive screens inside a mocked-up space; not nearly as fun as firefighting training in blazing boiler rooms full of smoke, or ship flooding trainers where you had to team together to escape through a hatch leading to a flooded room above you (tough to do). Those primitive systems were still interesting to me though, mostly because of my passion for filmmaking, with which I saw obvious similarities. It wasn’t until a few years later when I read Gibson’s “Burning Chrome” that I saw the potential which sparked my return to college for a second degree.

Shortly after graduation, in the mid-90’s, around the time “virtual reality” was making its first big splash, I picked up “Virtual Reality, Through the New Looking Glass“. The book is sufficiently dated that it’s now mostly just taking up space, but it taught me a core underlying issue which persists to this day: centralized versus distributed.

Yesterday I happened across a post on 3pointD (Link) regarding Outback: Online and what appears to be its underlying proprietary peer-to-peer framework. From the entry:

First, they tackled the challenge of indexing users in space, designing a spatial index that allows the various clients to discover users in the 3D space around them, without having to have all that presence information contained on a single server. Secondly, they tackled the problem of interacting, using multithreading techniques (among other things that got lost in a poor connection) to optimize communication between clients. Third came a security solution that obfuscates users’ IP addresses while still allowing clients to transmit the necessary information across the network.

I’d earlier speculated that Outback: Online was something like what they’re describing (reLink):

an Australian virtual world project currently underway, which sounds very much like Croquet with a persistent and constantly updated “seed”. In other words, they’re taking the old argument over virtual worlds – centralized versus distributed – and done what I believe to be the logical thing and created a hybrid; note that Linden Lab’s virtual world is centralized. This hybrid model probably has its own quirks, and I’m dreaming those up already

The first “quirk” that came to mind was, of course, how to deal with security. It appears they have a solution to this issue, as indicated in the explanation. It’s too bad this solution appears to be proprietary and probably has a centralized component which reinforces proprietary control.

This morning, I surfed through another post on the Meshverse Journal (Link) regarding the previously mentioned, standards-based “hybrid” distribution system (reLink). What I want to point out is something I’ve previously raised (reLink) and which I raised again only yesterday (reLink): avatar portability. As the Meshverse Journal clarifies to unnamed parties:

All modern PC’s can run webservers, but what Google Gears adds is a database and UI that will not only empower individual applications, but enable organizations of all sizes and even individuals to collaborate in grids … Now that the genie is out of the bottle people will look to deploy other database servers locally … Imagine Google Gears solutions running on these machines which will safely collaborate with other peers as well as the web

So while Outback: Online’s peer-to-peer system sounds interesting, it also sounds to me like a layer that will sit atop something else. And that something else could very well be a secure Croquet.