Metaverse 2.0 — Topology and Trust


Here’s a bit of synthesis and opinion following several recent blog posts on the subject of Metaverse 2.0 (thanks, Stefan). Start off with these three (Raph’s Koster & 3pointD & OgleEarth).

First, for those who want to see Metaverse 2.0 (from here on, M2) as an open unrestricted peer to peer world, I’ll argue that custom worlds/apps like SecondLife and WoW will always exist, much as AOL continues to exist (and I don’t mean to insult anyone with the stigma of AOL users) despite or perhaps because of the wide open web. There’s a reason people choose MySpace vs. any number of free  homepage hosting services. These private virtual worlds can and will continue to grow with time.

However, I do agree that the pure vision of the Metaverse requires something more like the traditional (though endangered) freedom of the open Web to truly grow and prosper. To have Metaverse 2.0 explode as the Web has done, to allow that level of creativity of approaches, failure and succes, we ultimately need the ability to host our own "Homespaces" and 3D content startups and connect them up to M2 however we see fit. But, let’s just stipulate that both the open and closed systems can exist, and, I’d hope, be connected by an open and toll-free highway.

I was hinting at this with my last post on the analogy of web protocols (like HTML) to what we need for M2 (unlike VRML). And I’ve already covered some thoughts on what one part of that protocol would want to look like. Here are yet two more pieces of the puzzle: Topology and Trust.

I’ll take on Topology first, since Stefan started there. Basically, there’s no technical reason any "web" of user-hosted virtual spaces can’t be connected to any other such spaces in any way we can imagine. The fact that SecondLife uses square servers arranged in a grid has clear benefits in their design, like in being able to see the adjacent servers as seamless, and like being able to see a nice 2D cartesian overhead map. Stefan’s hexagons have one small advantage there in relative equidistance between neighbors, like old table-top war games, but I’m not sure they buy much else. And why limit ourselves to Cartesian/Euclidian spaces, regardless of how we divide them up?

In arbitrary topologies, like the WWW, we go through links:  through a magic mirror to the next user’s space, down a rabbit hole, or even cross a spiral border if we can handle the shapes. There’s no limit, and no hard constraint, even in terms of reciprocity, except when you want to map the resulting mess (mentally or visually, especially for overlapping spaces).

For a technical discussion of how to link arbitrary topologies, have a look at "Cells and Portals" technology for 3D rendering, which could recreate the illusion of one seamless space (I have a little bit on it in my old Scenegraph article or google it for more depth). And note that teleporting in Cartesian space already does effectively create an arbitrary topology, perhaps dynamic, but certainly invisible, for better or worse (I’d say worse — see GE for a likely better way to fly).

Trust is the thornier issue. Some will argue that we can’t have an open peer-to-peer M2 at all because of various problems related to the ways in which people can and do harm one another in virtual worlds (sometimes even spilling into the real world). Experience arbitration and enforcement are two of the essential paid services of current virtual world/MMO developers.

For example, if I host my own 3D "Homespace," I don’t want someone coming in with a hacked client with unlimited powers or infinite resources, or anything more than they’d fairly earned. I don’t want someone making a mess of things with malicious scripts on my CPU. Likewise, if I take my avatar (and my data) to your server, I want to ensure you won’t steal from me, imprison me, or treat me unfairly.

So we have a clear mutual trust issue, even for such minute and fundamental transactions as "I pick up 2gp from the ground" or "I virtually nudge you" let alone: "I bought your laser gun and shot you before you rolled yourself back to a previous state." Good or bad, right or wrong, we at least need to be consistent about what happened in any given transaction if we want to maintain any sense of Capitalism, Democracy or Law. If you don’t care about those things, then have at it.

It’s not as much of an issue with the Web, mainly because web sites are, for the most part, read-only. Not so when you’re talking about your on-line bank or unfiltered blog comments. But even on the static web, we have problems with "web bugs" and phishing and images with viruses inside. It could be significantly worse with any naively implemented M2 protocol.

There is a potential solution though. Developers who implement MMOs already typically separate the "avatar accounting" stuff (your inventory, your money, your stats) from the world simulation itself (physics, rendering, chat). It’s just more efficient from a transactional point of view.

If one of these MMO/VW companies, say Linden Research, can take their considerable expertise in this area and apply it to open M2 protocols even without enforcing their grid topology, then viola–you’d have the equivalent of VeriSign on the web (only not evil). Linden may do well with their land auctions today, and therefore function as a sort of ISP in the M2 of the future, providing paid hosting to those who want a space. But they can also become the arbiter of trust on the open M2 highway. My guess is they’ll wind up there eventually, though I haven’t asked and I don’t know.

For the technophiles, the details of such a scheme aren’t too hard to imagine, though the implementation is another beast entirely. My guess is you’d need to be able to roll back any P2P transactions that were found to violate the agreed-on contract between avatar and server. Real-time monitoring from a third party trust broker may be too much to ask for now. But as long as the trust provider uses journaling for all transactions over time, I believe such an open topology can actually work.

Then again, I’ve been doing other stuff for the last few years, so what do I know?

  1. #1 by Stefan on June 4, 2006 - 3:36 am

    I think we’ve gotten ourselves an interesting conversation going, and thanks for the quick response.

    Just one comment. When you refer to Cartesian space, that’s the same as Euclidian space. So when you ask “And why limit ourselves to Cartesian spaces?”, that is what I was toying with when I looked at hyperbolic space and elliptic space. Just not sure if I got that across properly:-)

    As I see it, there are two real-world constraints we can let go off when building M2: That space has to be flat (Euclidian) and that the topology has to follow the geometry of space (it can be non-contiguous, like the web.) it just took me a very long post to say that.

  2. #2 by avi on June 4, 2006 - 9:49 am

    Yes. The idea I probably didn’t express well was that, while Goggle Earth and most MMOs nowadays are true 3D worlds, at least locally, on the global scale, they’re effectively 2D–meaning their worlds are pretty flat in the 3rd dimension overall, despite being wrapped on a sphere in GE’s case and having mountains and buildings in both. (Here, topology obviously means connectivity, not terrain.)

    Taking any such "effectively-2D" space and bending the whole thing to any non-holed continuous shape, no matter how extreme, is roughly equivalent for an ant on the ground. Flat, elliptic, hyperbolic, matter only insofar as they form new connections (seams, folds, short-cuts, and warps) as in the case with GE’s sphere, which makes it so east and west meet; or even a torus, which does the same for north and south.

    For example, you could take GE and wrap the data to the inside of the sphere and it would work roughly the same. The principal difference would be that the horizon slopes up and if you go to the middle and relax your FOV, you can see roughly 1/2 the world at a time. I did such a thought experiment of a VW in my first novel, but I called the space RealityPrime. Imagine what the metaphor is if the Metaverse is the Earth, just inside out and right under our feet. I’m not sure people would actually like that. :)

    Metaverse 2.0, if decentralized, would need to work more like the web, handling discontinuities and arbitrary connections, where we lose the nice metaphor of one big flat space and wind up using graph theory to draw our maps. But I think we agree on that. The challenge is how to make it all feel seamless, despite the natural discontinuities, where distance may be measured more in "hops" than in kilometers.

  3. #3 by Morgaine on November 16, 2007 - 4:55 pm

    > “I don’t want someone coming in with a hacked client with unlimited powers or infinite resources”

    You really need to bring your worldview out of 1990’s. avi. The only kind of client that’s going to get any significant cross-world mindshare is an open one. The open meme has taken hold, and isn’t going to disappear.

    And that means that if you have powers or resources to guard, then you have to guard them server-side, because nothing that happens client-side is in your power to restrict. This was true even before open-source became huge. It was just a bit harder to modify binaries before, so fewer people could do it, but certainly not impossible.

    Linden Labs understands this, as well as the fallacy that underlies DRM, and refuses to get into a pointless arms race with their customers. Indeed, they have open-sourced their client so that they can ride on the shoulders of giants instead of fighting them. If you continue to see these issues through the eyes of a previous generation, and see users and their clients as adversaries as your phrase above suggests, then things are not going to go well for you.

  4. #4 by avi on November 17, 2007 - 8:08 am

    Ah, Morgaine, I’m going to delete my previous response to you. I think most people reading the article above and your post will conclude that you simply didn’t follow what I wrote and decided to divert with rants about DRM and open source. The scheme I proposed is in fact server-side, but it doesn’t require centralizing the whole world. It is possible to have an open-server architecture where anyone can run their own part of the world and still have a central authority for validating important transactions. Such a model exists today — only the number, speed, and size of transactions would change significantly.

    Anyway, I really don’t like your tone. You’re a borderline troll. So don’t bother responding yet again.

    As for Linden, I’m sure they’ll make money before this latest bubble bursts. However, I personally wouldn’t want to be forever known as the AOL of virtual worlds.

(will not be published)


  1. Ogle Earth
  2. The Other Here » Where 3pointD: Second Life, Google Earth, and Space
  3. Prompt Criticality
  4. baack to the future » tag — you’re it