Up until now virtual reality has been in "walled gardens." A provider creates a world, provided "content" and allowed users to interact with it, often in very directed ways. The most common kinds of VR over the internet were combat games, and "3D Chat." The Second Life™ world was unique because it offered a different model. It was not multiplayer combat, nor 3D chat, but, essentially, what VRML had promised so long before: a VR version of HTML, where people made places. The last few months have seen rapid changes in this situation.
The idea of the Open Grid Protocol is to make it so that one user account can seamlessly move between regions hosted by many different companies and in many different locations. There are two pressures forcing the evolution of this idea, one is internal, in that as people grew more adept at working with this model, there was more and more desire for more and more control. From renting "parcels" on Linden Lab controlled mainland, to having a server, to making server ownership a commodity with Openspace™ sims. The parallel track was to allow reverse engineering to produce various kinds of "open" simulators, OpenSim being the most famous. The second pressure was external, as cloud computing, 3D graphics, and end user devices changed, the world of 3D chat was always going to make a large leap upwards in both quality and availability. The recent announcement of a cloud photorealistic VR, rendered by the server and served to even small devices marks merely the next level of entry into the world of 3D chat VR.
The response from the Linden Lab-IBM partnership was to look at the internal and external pressures as being the same. That the problems with concurrency, and the demand for personal control over simulator spaces could be solved by opening the platform. Opening the platform, however, created a series of obstacles, from Zero Linden's prediction of "scary numbers" of virtual reality users, to Zha Ewry's, the IBM architect, understanding that interoperability would be deeply related to continuity of identity within a VR. The famously patchwork nature of the underlying code and legacy decisions made to ship code quickly became visible rapidly. Despite having straw man log in procedures for months, it was only in July that the first OGP intergrid teleport occurred.
The result was a series of efforts, one being the effort by Zero to re-architect the grid itself, the other being what is now called the "Open Grid Protocol" or OGP for short. OGP's first step was to reduce the dozens of pieces of information exchanged between a client and a server down to a few simple pieces of authentication. The second step was to change the very meaning of what we know as the word "Grid." Instead of one singe source hosing both the sims, or regions, and the clients logged into them, there would be three parts. A Viewer, an Region Domain, that is the sim or group of sims in a grid, and an Agent Domain that would negotiate trust between the two. This would mean that a user would log in through a Viewer to an Agent Domain, and the Agent Domain would negotiate presence with the Region Domains. OGP will be the means by which presence, license, and trust are negotiated between these pieces.
The test of this concept was initiated by Zha Ewry at IBM, making the first OGP log in between an Agent Domain and different sims. This test was then opened to a "beta," though more accurately alpha test, where users signed up for the test, were invited into a group, and could then teleport from a region in one place, to a region in another. The "beta" grid, Aditi, was used for this, and by now dozens of avatars have jumped from a sim hosted in different places.
This testing of OpenSims, and making them compatible with OGP, proved to be a success, even though the OGP client did not, intentionally, have inventory, so our "gridnauts" were all ruthed, or clouded. The lack of inventory comes from the still evolving legal question of how to handle intergrid trust, and as a consequences of this, inter-grid transport. Infinity and Enus Linden are both involved in dealing with the technical specification of trust, but there will be an inevitable marrying of law, technology, and custom which will have to be debated, sometimes harshly, along the way.
The other track however, was the building of an easier way to test, and to log into, this new protocol. This work is encapsulated in PyOGP, or Python OGP. This has already gathered enough notice to have one integrated development environment to offer it's professional version of its python environment free to those doing OGP work. One important contributor to this has been Saijani Kuhn, who first attempted to sort through the legacy log in process and then produce the first python OGP log in. He has also been crucial in keeping the sometimes fractious interface between LL and interested residents moving.
The current state is that this process is not in an advanced state, but has the clear commitment of Linden Lab and IBM. The next steps are to begin creating tests for OGP log ins and expanding what is supported. Under this is a radical simplification of how the client talks to a simulator, and the possibility of more, and easier, text based interfaces to Second Life compatible sims.
The business motivation for this from LL's stand point can be outlined simply. By allowing others to host compatible sims, but with an Agent Domain which is hosted by LL, it gets around current concurrency issues, eases the need to be an ISP as its main business, and creates a much larger base for content creation. Having been the most successful of walled gardens, LL is now trying to escape being out prettied by newer even better walled gardens.
The motivation for others is also clear. Being an ISP is something that many people know how to do, there are large numbers of techies who could simply offer Second Life compatible servers in addition to other applications, and there is the possibility of hosting farms of sims without paying for the overhead of support and service from Second Life. OGP however is also a threat to entrenched interests that have made their living by offering what amounts to middle man services between LL the ISP and end users. What need there would be for this large class of people in an OGP world is not clear, except that actually using a sim is a complex business, and service providers with the ability to do more than flip land would have the ability to offer this as a value add.
By the time this article sees print, it is very likely that OGP will have taken several more steps to creating tests, integrating a sophisticated trust brokering protocol, and perhaps even the outlines of content licensing enforcement. The largest hurdle is that trust is a more elusive concept where people, and their avatars, have thousands of bits of independent content to manage.
This model leads farther down the road of having the Linden Lab's simulator be the web server for interactive poly-dimensional content, and fits in with the transition to open standards such as Mono for various aspects of the platform. There has already been no small degree of controversy in how this is to be executed, and on what time table, there is likely to be more. The upside for the Second Life community is that it will forge further into a unique spot in the VR universe, the downside is that there is a wave of "Second Life killers" being prepared which stake their success on the proposition that what people really want is a more easily used walled garden.