Ex-Linden vows to battle for interop
Former Linden staffer Meadhbh Hamrick has vowed to continue her fight for virtual worlds interoperability.
However, the commitment on the part of Linden Research is unclear, and the project is currently being overshadowed by the success of the hypergrid, an alternative method of bridging worlds.
Hamrick — also known as “Infinity Linden” — wrote much of the Virtual World Region Agent Protocol, designed to help different virtual worlds relate to one another. A full working implementation of the protocol which would allow teleports from one world to another is still two years away, she said.
Hamrick’s experience with interoperability goes back to the summer of 2008, when she was part of a group that achieved a demonstration teleport from Second Life to an OpenSim grid run by IBM. It wasn’t a full teleport, she said — no avatar appearance or inventory items made the trip. “It was just a demo of where we might start,” she told Hypergrid Business.
That early promise of interoperability hasn’t yet borne fruit, however.
“Me and my team, we took that and started going more stuff with it,” Hamrick said. “But as soon as we got started, we got ‘de-emphasized’ and my team went away. I was the only one left.”
What followed was almost a year of negotiations, in which different people had different ideas of what a virtual world was, and how interoperability should work. Finally, a working group was created last November under the aegis of the Internet Engineering Task Force — the same body responsible for a number of key Internet standards like TCP/IP and HTTP. The group held its first meeting in Anaheim, Calif. on March 23, Hamrick said.
Even though she announced her departure from Linden Research just a week before the meeting — she called the parting “amicable and mutual” — Hamrick insists that the Lindens remain committed to the idea of interoperability.
“The co-chair of the working group and one of the contributors to the specs is Linden,” she said.
Today, Hamrick is working an independent consultant focusing on virtual worlds, network security, and mobile device development.
Meanwhile, the hypergrid protocol — invented by University of California at Irvine professor Crista Lopes — was implemented about a year ago for OpenSim-based worlds. Dozens of virtual worlds have embraced this approach and are now hypergrid-enabled, allowing full teleports between different grids. (Directory of hypergrid destinations is here.) Second Life, could, in theory, connect to the hypergrid, but there are a number of business obstacles that first have to be overcome, as well as some technical challenges.
For example, hypergrid is an implementation of OpenSim, not so much a stand-alone protocol, she said. “You would have to reverse-engineer it” to make it work with Second Life.
However, two approaches are not in competition with one another, Hamrick added, but are designed to accomplish different tasks, and may well be used in concert in the future.
The Virtual World Region Agent Protocol is a way for worlds to communicate a variety of information, such as identities, presence, or assets.
It could allow people to log into a virtual world using, say, their Facebook or Twitter identity, Hamrick said, or to use an avatar from Evolver. In addition, worlds can use the protocol to offload asset storage — allowing users to save their inventories with an online service or on their computers, for example.
One area that the protocol won’t be handling, though, is payments.
“That is something that is specifically not going to be standardized,” she said. “Linden was unsure whether Linden dollars should be standardized, and a lot of the OpenSim people said, ‘I don’t want the Linden dollar anywhere near my grid.’But there are some underlying, base-level protocols, and I’m sure future Linden payment systems will use these underlying protocols.”
Another area of potential controversy is the permission system for objects. Today, Second Life objects have three permissions settings — allowing holders to copy, transfer, or modify the objects. There is no setting however, to allow objects to be transferred to other grids.
“We have identified it as a critical issue for the growth of virtual worlds,” she said. “Having a permission system that works across organizational domains is critical.”
It’s not clear what the permissioning system will look like when it’s finished, however.
“I anticipate that our first specifications and our fist implementations will be interesting, but will also have holes,” she said. “We will need to get them out and deploy them in test mode before we jump up and down and say ‘This is completely safe, bring your content over here.'”
Another function of the new protocol is that will provide a roadmap for Linden Research to replace traffic that goes over the problematic UDP channel and to funnel it through the industry-standard HTTP. The latter is a more standard approach that travels more efficiently over the Internet, plays better with corporate firewalls — and will allow for more logical hypergrid addresses.
Currently, hypergrid addresses are based on the IP address of the server hosting in the individual region — instead of grid.com/regionname, a hypergrid address looks like 18.104.22.168:9000:regionname.
“HTTP is the one that gives you the redirection semantics,” she said.
Implementation of some of the individual pieces, such as the logins, may be completed before the full protocol is done, she said.
Individual worlds can mix-and-match approaches, she added, using hypergrid for teleports, for example, and VWRAP for identity authentication or inventories.
“We understand that there are people out there using hypergrid and are happy with the infrastructure,” she said. “Hypergrid is not going to go away.”
Hamrick said that her group would welcome participation from the hypergrid team.
“There’s an open invitation for Diva [UCI’s Lopes] to participate in VWRAP in any way she wants to,” she said. “If she wants to come in and participate in the standards development process she is very welcome. I would personally love for her to show up.”
Hamrick added that she’d be happy if the hypergrid team adopts any of the protocol.
“If Diva [Lopes] looked and said, ‘Man, we should do that,’ and picked up some aspect of that, that would be a success,” she said.
According to Lopes, one of the differences between the hypergrid protocol and VWRAP is that hypergrid is build with higher security, so that users can teleport to strange grids and not worry about losing their inventories.
Lopes agreed that the two different approaches can be complementary.
“I can totally imagine these protocols used for different levels of trust, and different levels of security,” she said. “From my understanding, the protocol that they have been envisioning is sort of more like the Facebook Connect API, a protocol that only exists if you already assume that there is a certain level of trust between the different grids.”
That might be a good option to connect two grids that know each other, she said. “The hypergrid, on the other hand, is designed as a protocol that’s independent — you don’t need to have established trust relations.”
This approach works for many grids, but some may want a higher degree of privacy.
“We’re pretty sure that in the future there will be lots of people who will want to limit where in the open metaverse their content flows,” said Hamrick.