VW08: Technical Visionaries Debate the Future

Liveblog, excuse trhe typos!

Technical Visionaries Discuss and Debate
The Future of Virtual World Technologies
This session will analyze the future course of virtual worlds technologies. Join us for debate between leading industry technology experts on the future of the technology, where it’s headed and what needs to be done to get there. Don’t miss this lively conversation.
– John Swords, Director of Business Development, The Electric Sheep Company (moderator)
– Ian Hughes/epredator Metaverse Evangelist, IBM
– Ben Goertzel, CEO, Novamente LLC
– Mark Wallace, Chief Executive, Wello Horld, Inc.
– Christian Renaud, CEO, Technology Intelligence Group

Mod: When we brainstormed, a common theme was that we might be at a point where it’s time to talk about VW’s 2.0. The last few years could be described as 1.0, what characterizes the changes and what is 2.0?

Ian: The 2.0 thing kind of scares people, geeky. For me, there’s stuff that is probably going to happen that breaks this avatar model. There may actally be better ways for us to interact as humans over this medium, and it may be that while we try to repicate the world and dress up in interesting ways, that there are better ways, whether it is motion tracking or other ways we choose. That may be a leap we need to take, though avatars has brought some acceptance.

Mark: I wonder if that really consistutes 2.0, and whether we can have an O’Reillyesque conversation. From my point of view there’s some shift that will take place sometime soon — 5 year horizon — at which VW and the tech come to resemble the tech of web 2.0. More plug and play, more easily combined, a more atomic level. That does shift away from the avatrar and island model, but there are these two parallel tracks of richness of interaction and both need to be addressed.

Christian; I think the transtion from 1.0 to 2.0… a lot fwhat w see is evolutionary artifact from MMOGs. As a result, avatars and islands, you go to this other place, not incirporated into our daily lives. When we do away with the MMOG minus G model, that’s when we get this big evolutionary breakthrough.

Ben:  I have been thinking about this a lot lately, been talking to a friend, and our thinking ahs been around a new concept, which we call “existence.” One of the definitions in the dictionary is “that which has its own individual being.’ We’re generalizing the notion of avatar, ifyou take this point of view is that the world sn’t at the center but the avatar, which may have its own state and knowledge, avatars that learn and interact in various ways and are portable across different worlds, flexibly controllable by all sorts of external processes. This is an extension of the interoperability, move avatars and machines and ways of organizing structures, from one world to another, where a world is one way of manifesting a more general existence.

Mark: I agree, there will be a lot more focus on things centered around identity. It’s not limited to VWs, there’s enormous pressure on the web aboutidentity and tying people together across disparate domains. Many different people trying to solve this problem in many contexts. It remains to be seen where the solution emerges if there is one.

Ian: it doesn’t take much for some combination tech and inventiveness to turn a model on itshead. A few weeks ago, the Mozilla stuff, Ubiquity, which was about not making the browser or email client a place to go to, and the web comes to you.

Ben:  If you have trained an intelligent agent to do something, 2d world or 3d world, its the behaviors that are the key thing, not the context in which it is manifested.

Mod: Christian, if we are a tthe point where we are getting close, are there going to be subindustries? Splinters? What does that look like?

Christian: multiplae answers based on environment. If I an doing intelligent sim, US govt simming Iraq, then I want to make a space with AI… that in itself is a splinter that will go on.In the social space, it will be a feature of existing systems, like seeing Facebook in 2d or 3d. What we need to do is diffuse and harmonize with all of those. The SLIM accouncement — yet another walled garden IM — is a step back, not forward. You just continue to multiply that by all the other uses. Education, universities, is another. WYou’re going to get a lot of speciation, and we will see VWs in some uses, and in other uses, it will be a feature, a 2d website with a 3d view, with pesistence. To his point, the persistence, identity, and presence that you can carry with you. We solve it for VWs… Just saw Metaplace, that sort of transparent integration.

Mod: sounds like thispanel is thinking more in terms of augmented reality and moving these social interactions to other platforms and other interfaces other than laptop and brwoser or downlaoded app. Ian, if some of the industry mves away from avatars on islands and a lot merges with social media and SNS tools, what are the redeeming values of VWs 1.0 that carry forward?

Ian (4): the important thing is not technology, it is about people. Whether it is enterpris eor education or socially, that is what web 2.0 is doing for everyone. People can do something and organize themselves. VWs give us a deeper degree of connectivity, we get to know one another a whole lot better. But you need to be able to take that and say what is thenext step? We have done blogging, video, avatars on islands. Is it how I am feeling/ Body language? Something more than gestures on a keyboard? So that when we are all sitting ehre we can see our vitual colleagues. W ehave the technologies, and quite a lot of them are available,s so we have to deive that and have people feel the benefit.

Mark: all that stuff is really important as interactions get even more rich than now, as they are in SL and other places, richer and more colorful and more expressive. But I don’t think that is what will push it over the edge intot he next phase. Thereis a set of tech that are rudimentary or missing around VWs aboutbringing people together. On the web that has been solved in search, for instance, and different methods of doing discovery. In VWs we haven’t even started to think about discovery very well, it’s one of the big problems wih Second Life. You can wander around and see all this very cool content if someone has told you where it is or you know where it is. If you are just dumped into this environment it is very hard to find where this stuff is, and there is a whole set of technologies that ar eonly just starting to be thought about.

Mod: followup question, we think of VWs as very synchronous tech, realtime. As we move more to the web and have more asynch content and people interacting not at the same time, as we move to VW2.0, — everyone smirks when I say that! — how do you see the asynch non realtime stuff manifesting?

Mark: an interesting question. Because I do think of these as synchronous, very much about realtime, and that is the big advantage they have over the web. I imagine that the asynch tools will resemble the asynch tools on the web. There will be ways to discover new info, and form new connections. I believe the #1 use of email is still to arrange face to face meetings. The asynch tools when applied to these real time environments will be tools at the service of the realtime interactions. Things like lifelogging is something totalk about where it fits in, record and store the actions you take. That is another set of tools only starting to be developed for the offline world that will have interesting applciations in VWs.

Ian: We had a guy from Dopplr by… online website that is there for you to share your travel tips. It is a serendipity engine so if you are in LA at the same time but not at the same conference, you realize that you can meet. I asked him are you going to make this a virtual world, will be the serendipity happen? He said I hadn’t thought of that, thank you! That ability to connect people and willingness to share where you are and what you are doing is what makes this stuff useful.

Ben:  responding to thw question o what is the unique vaoue delivered by he 3d worlds… there’s an added component in the same vein. Just as 3d worlds allow richer interaciton they also allow richer interaction between people and non player characters. From an AI point of view, teaching Ais to imitate you or be your digital twin, or having hem solve the real world search probnlem or teaching a virtual dog virtual tricks, it’s all richer in 3d. Imagine trying to raise your baby, seeing it only on a handset or in a 2 1/2d world. I think we need to go further than the current 3d worlds in terms of allowing interaction — detailed control of skeletons, etc, to enable more emotional and and instructive interactions bewteen people and NPCs and Web 2.0 evolves.

Ian: As well as the manifestation of the dog or whateverm you can do a digitl MRI scan so to speak, so you have that extra level of scale and depth.

Mod: we are talking about the value being interacitng with 3d content. How do you feel about using them to interact with 2d content?

Ian: If you break everything down to consuming services that can be consumed anywhere, there are times that you will not be as a full workstation, you will be on a mobile decice, or in your car, or in a field under a tree. it is important that we have all the layers of being ableto interact in comfortable way, so when we cannot be in LA we can see the Twittering so we know what is going on. Whether it is 2d 3d or 4d.

Ben:  In a way there is a limitation of vision underlying that question. A lot of content we are used to viewing as 2d is not intrinsically 2d, and there are lots of ways of interacting with it that aren’ 2d. We are used o looking at software code in 2d, but if anything it is 1d. Maybe looking at a 2d panel for a 3d world, maybe there is a way to project code in 3d and see thigns differently, it is just more work to do it that way. There is huge potential there, just that for reasons of cost it has sometimes been done inan overly simplistic way.

Mod: how do you think open source is going to play a part inVws 2.0?

Ben:  it can pay a huge part. From an AI perspectivem, we have been working with some proprietary platforms and some open source ones, and there are pluses and minuses to each. Some of the closed ones ar emore mature with easier APIs. But we have often ound you want to be able to go in to adapt the platform for the AI, and when it is ia a proprietary server that is a lot harder. So there are two ways it could go. One is that the open source stuff is a kind of vanguard where cool new stuff is prototyped and implemented, and it gradually filters into the closed stuff. The other is that it just gets ahead because of all the funky stuff an dit becomes the foundationfor a metaverse like it is for email and web servers. Which of those two ways it goes is hard to sayat this point.

Ian: As a software company, initially open source would have been looked on as a threat. But most of the large software companies have embraced it, like we ciontrobute to OpenSim. A lot of innnovation happens because people want to do it, and it’s imprtant in something like this. And Sl going open source, it’s not even settled in as a the nasty monolithic branch yet.

Christian: we talk a lot about how to get the J curve, and open source can do this. At the web 3d conference someone made the observation that they could not do VWs yet because they worked for an auto company that had 10 year cycles. So as a risk mitigation thing, they said they woul dbe glad to adopt when there were open standards, andopen source would counter the risks these people take.

Mod: At least a handful of companies have come out with open standards, but haven’t gotten any mass adoption, and haven’t really consulted the larger group of folks building these worlds to see if they fit their needs. Ian, while talking interoperability, does id portability and avatar portability really matter?

Ian: There are two elements there. Technically, I don’t think it is a particularly big problem, We do it in enterprise all the time. When we get downto it, it’s just bits and bytes. But then there are layerson top, what does the data mean, protocols, etc. But the big thing is the social side of it, the acceptability of moving stuff, “I don’t want my WoW character wandering around Sl.” But the point is that you won your idendity, federate it across multiple platforms. WoW may not want you to move your lvl 70, bc it is theirs. But in a free enviromment, anything I make should be mine to move.

Christian: the portability piece, we talked aboutthis at VWForum last year. Avatar portability is a subset of asset portability, which may be the clothes you are wearing. If I paid for the clothes, and someone’s biz model is predeicated on that, then i cannot tak eit into There.com because i did not have an agreement with the creator. Before we dictate to people and radically dictate their business models we needto understand that. That way you can use items without getting into the stickiness there.

Ian: there is a technical middle ground, which is if something belongs inan environment, that business wants the hits and the traffic. So when we talk about portability, there are ging to be ways to say This tghing is somewhere else, bring it in, but it exists in the original place so they get the traffic, but you can use it in some way to adorn what you are doing. Each environment is getting hit to remotely render say a powerpoint.

Mark: in VWs a the moment, th proliferation of different tech protects that business model, the biz modelis founded on the tech not being interoperable. When you look on the web, you see a much smaller number of formats and standards, people have no problem making money out there. That is an indication things will priobably change in VWs.

Ben:  The same issues are with cognitive goods, NPCs. The mindof your avatar or pet dogor machine is the persistent thing. And that mind can be ported from one world to another. But then you have who owns the mind? If you taught the dog 1000tricks, is it the server owner or do you want to download the dog’s mind to upload into someone else’s Ai engine? The same issues arise. The company has a financial stake in letting you teach the dog tricks, but maybe if you pay them something you can buy the file and put it somewhere else. The biz model will have to evolve to support introperability on alllevels, visual goodsand cognitive goods. But I don’t see why the biz models can’t like they doon the web.

Questions: isn’t the wbe full of walled gardens? Amazon, eBay, Yahoo? They use the same building blocks but do not share identity and data. W ehave been talking data portability on the web for years, but it’s not there. Why is this different?

Ben:  in the specific case of NPCs that learn, if I have spent hours a week to teach an AI, that is a lot of investment of my time and life into this thing, and I have more motivation to port it, whereas my Amazon profile doesn’t take that long to replicate. There is more user motivation to demand portability.

Christian: To answer your question; I dont think it is. It is the same old identity problme. The issue here  to mashup your points, s that it is like the Apple DRM issue and who owns your assets. If I go create IP, ai or house, I shoul dhave some claim on that property unless it is prohibited in the EULA (and then shame on me for using that world). It’s about me and these are just tools to support me. The worlds like metaplace (talking about what is fresh in my mind) I can mash stuff up on the Net and I can use those assets elsewhere, I might have multiple identities on different services, and that makes it more attractive because I am not locked in. Too many passwords, it’s too much and inconvenient.

Ian: yes, it is inconvenient, and sometimes you go to a new one fo a fresh start. In some ways you have walled gardens for the brands, but generally you do ave a common id becaus eyou putin your credit card. So that standard has emerged. We just need that extra level. That kind of approach is hitting businesses as well, and how their processes function. Do you have to buy a walled garden, or can you grab a little bit form here and there? The tech is there to do it. We start to get back to a service mentality., making it us-centric, not walledgarden centric.

Q: (muffled) The thing about not being yourself in n a VW is fundamentally different….

Repeated by Mark: the pain point in VWs i different from web, my history of interactions witht he websites is that I cannot connect purchases on amazoin with rep on ebay. But in VWs, the friends in SL cannot find me on WoW without my expicitly telling them where I am. The fact that there is no unified ID across the environments makes it harder to do what the VWs are good at.

Christian: we do it thru twitter now

Mark: do you twitter your WoW stuff?

Christian: I have kids, no wow for me.

Mod: every vw shoul dhave had pointing your address book at the servers to find your friends. Why not?

Christian: because anonymity is built into this,so you ca be Rebecca instead of John. That might be invasion of privacy. You have to ask people, is it OK to get your real data, when interviewing people in SL. And they say no, i won’t give real name, you say I won’t send you a real paycheck.

Mod: I wante dto ask about the future of hardware VWs will run on. Intel just announced a chip with many more cores, moving away from the GPU concept, which is a specialized chip for graphics. Any of you want to comment on that sort of thing, remote video processing?

Ian: Cell is in the PS3, same principle. There is the Moore’s Law thing, we’re able todo more and more processing, better Ais and graphics. But as Jon Landau said, there are some elements we do not need to do more and more of. You can engage very strongly with cartoons, and you don’t need the uncanny valley. I think we will see things driving more tech, games and media will generate better and better physics and graphics, and it is just whether we actually need that for social interaction.

Ben:  I think multicore can have a huge positive impact on VWs, not necessarily in improved graphics… right now one of the issues with AI is keeping cost down, packing AI minds onto one server. With 100 cores you could run your dog and your virtual girlfriend on your PC, and not pay subscription costs. in terms of physics a change that has to happen is the use of physics engines mediating interactions between avatars and objects. You cannot use tools in VWs, stick a fork into a steak, unless a lot of work has gone into building sockets into the objects. If it works mor elike a robot, wherephysics mediates, you can do much more — AI, bvirtual robotics, cybersex. there is no problem doing it technologically, it exists in virtual robotics, the limitation is bandwidth and processing power.

Mod: last thing: name a game changing technology that we may se ein mainstream intenrt users in the next three years, give or take, describe some applications.

Ian: 3d printers. They allow rapid fabrication from digital models to actual object, it is already possible and has ben for years. They have gotten incredibly cheap — $5k that can make something on your desktop. This leads to… we now have the real weorld modeled by people in VWs, and then we can get the virtual world back out to the real world as actualphysical objects. So if I want to make something and distribute it, I don’t need to spin up production in China, I can deliver it cheaply or free and they manufacture it locally hen they need it. They just need the raw materials. No ships needed, sweatshops, nothing like that — from designer to reality. It is gamechanging and really interesting. You can go to Fabjectory, Shapeways, get your Sl avatar. But it has impacts on the economy.Drug companies — can you print drugs from a recipe? Huge economic impacts beyond toys and dolls.

Mark: Tangential answer… I think a couple of really interesting tech that are out there or just coming out have implications. Changes in browser tech — Ubiquity, WebSlices in IE. The content in those environments becoming more atomic. We see as the web start to fragment, you start to be able to slice and dice it and mash it up. One of the thigns will that will make a big difference is doing VW slices. And discovery and identity i think will make the biggest difference.

Christian: bc it is ho wyou interact with others… I think we don’t pay enough attention is not only AI, but the rest of the world and physical objects. The intersection between location based services and smart objects –spimes! — intelligent objects that return data back. Inanimate objects that you can askabout the best curry in LA. Augmented reality and location based services, and the blend of that between meatspace and virtual space.

Ben:  Moore’s Law is going to enable a whole bunch of new tech. More realistic physics, tools, sim of complex machinery. in parallel, Moore’s Law will allow more rapid advance in Ai tech partly for reasons that have nothing to do with VWs. But I think there is potential for them to evolve together, and this will characterize VW 2.0: NPCs that interact linguistically and conversaitonaly. Virtual pets, humanois search engines job interviewers, tour guides. When you talk about intelligent NPCs, in the Metaverse, the collective unconscious is a real thing — the agents can share knowledge. And this willlead to a secondary market incognitive goods. And the virtual goods market will explode anyway in Worlds 2.0.

Question: how does the panel see thin clients working out and affecting the industry?

Ian: it’s back to services. If you need to combine some stuff in a particular space, and you get the processing power elsewhere, good. But if you have the power, do it locally on a thicker client. It shouldn’t matter where the stuff is. If you only have an iPhone, you don’t expect the iPhone to do the work, but there are things that it can do really well. We shoul dhave all, not one or the other.

Christian: one thing you do get, when you mix in the cloud computing, is taking the barrier to entry for startups down. My own startup, the backend is Amazon. It used to be prohibitively expensive. But Ogoglio was able to deploy fast and cheap on Amazon. If it is that easy to scale my business, then new business models come out of that.

Question: Ben you talked abou actively teaching AIs. How much avenue is there for the agents to just learn empirically from observation?

Ben:  That can be a really huge part of it. Most children learn just bywatching. On the other hand, having the active correction and encouragement is an important part of the process, kids need reinforcement or they get confused. So we will need both to have agentds of high levels of intelligence.

Question: Ben remarked that we will have the chance to share knowledge via virtual pets, which makes explicit a lack we have today.Creative Commons enabled sharing, but explaining with simple layman’s terms what you can do, and automate the transactions, to sue them in your own apps… This kind of approach is totally lacking in VWs, and users are not stupid, Privacy issues, issues of content sharing and the like are going to become even more acute in interoperable open worlds. What is your take on this, is it slowing the industry down?

Ben:  We have been thinking about that a lot in terms of cogniive goods. Who owns the content of your digital twin? The creator of the learning algorithm should be compensated, but if you trained your digital twin, then you should be able to own the mind of that twin. If you want the contents in a file, then you pay for it. It is important to be simple and explicit about how you manage it, and a creative commonssort of approach will be needed. You don’t want to spend years building something and then have the startup hosting it fold.

Mark: that means you need a platform that supports it, like the web does. That doesn’t exist in VWs today.

Mod: Thank you!