DAVID LEVINE (ZHA EWRY) - IBM
MARCH 3, 2008
ROBERT BLOOMFIELD: Welcome everybody to another edition of Metanomics. This is
Robert Bloomfield, Beyers Sellers, in Second Life, sitting here on the CMP Isle. As always,
Metanomics is brought to you by our sponsors, Cisco Systems, Saxo Bank, Generali Group,
SAP, Sun Microsystems, and special thanks to Second Life cable network, SLCN, who
makes it possible to bring us these shows across the grid, live, streamed onto the web live
and archived on SLCN.TV. Finally, Kelly Services and more special thanks to my own
employer, Cornell University: the Johnson School. And I do want to take a moment to point
out Cornell University for the first time in 20 years is home to the Ivy League men's
basketball championship. Congratulations. Go, Big Red. If everyone wants to stick around
afterwards, maybe I'll sing our alma mater.
But anyway, let me also give a special thank you to the folks at CMP Metaverse who own
this amphitheater that is brand new today. I'm hoping SLCN can give a little pan of where
we're sitting now. This is four sims seamed together in quite a nice piece of work. And will
allow us to host very large audiences, particularly on days when Second Life is more stable
than it has been today.
Next week Metanomics will be hosting Richard Bartle, co-developer of arguably the first
virtual world, The MUD, The multi-user dungeon, which brought Dungeons and Dragons
online. And more recently Richard is author of the 2003 book Designing Virtual Worlds.
This week we have another guest with a tremendous historical background in online
communities, online collaboration, and 3D visualization. David Levine has been an
employee at IBM for 23 years. And if you tack on summer and university internships, add
another five years to that. His interests in social computing and online collaboration start in
the early '80s in IBM's online bulletin board style conferencing centers and Tools for Sharing
applications in the very early days of personal computing.
David's currently working in IBM's T.J. Watson Research Center. And his work focuses
primarily on virtual world's technology, with the long term implications of broadly deployed
social collaboration too. He's very active in Second Life, working closely with Linden Lab's
Architecture Working Group, and IBM's OpenSim team. David's work covers the technical
side needed to permit interoperation between virtual worlds, and has been looking closely at
the policy and business and social implications of this work. And I have to say, I for one, am
grateful for his interest in these broader issues because, as ever regular Metanomics viewer
knows, I'm not that much of a technical wonk, and I'd rather focus on business and policy
David joins me today on Metanomics in his primary avatar Zha Ewry. David, welcome to
DAVID LEVINE: Hello everyone. And if you're hearing me in voice, can you hold all
friendships and chat until afterward? I'll gladly grab them then?
ROBERT BLOOMFIELD: Excellent advice. Before we get into too much detail, David, I
assume you have a standard disclaimer--
DAVID LEVINE: Yes, I do.
ROBERT BLOOMFIELD: --that you're not speaking for IBM?
DAVID LEVINE: Right. The IBM tag is there. I do work for IBM. I think that's absolutely
necessary that everybody knows that. I work directly in Second Life and with Second Life
stuff. I'm not speaking for the corporation; I'm speaking here as a personal employee
involved in the work as a resident as well, and I'm not going to be able to or want to speak
policy. So this is purely personal perspective on the technical and social matters. Obviously,
what I think and what I say has some impact on what the company does, but that's all
indirect. So there's your disclaimer.
ROBERT BLOOMFIELD: Well, I'll warn you now, I, and perhaps some of our audience
members, will also want to ask you questions about IBM's plans and vision, and you will
simply have to dance gracefully.
DAVID LEVINE: And I can do that. I'm sure I have my [chim?] someplace around, and I can
ROBERT BLOOMFIELD: Now, we're obviously going to talk primarily about the present and
the future. But can you start by telling us a bit about your early years and what arguably
preceded the Metaverse?
DAVID LEVINE: Sure. I have sort of the horrifying reality of having had a stable email
address since 1981 in some variation or another. email@example.com has found me since
then. And I got involved very early on. And IBM has a large internal network, of course.
There's some funny stories about how big and how fast it grew when at some point
somebody said, "Well, I'll just send that as an email to the West Coast." And the CEO said,
"Oh, we can do that?" And he said, "Yes, we have something like 800 mainframes in the
network now." And he practically dropped his soup.
And we had an internal system which was called IBM-PC, which is initially done as way to
do file sharing and discussion on--obviously enough, the IBM-PC started in about 1981. And
it laid right on top of horrible 1980's mainframe technology. It literally sent virtual punched
cards around the world as conversations.
But all of the issues that we run into in building things like this are there: identity, trust,
whether or not we should believe an answer, how to attribute answers, how to quell flame
wars and bad behavior, and eventually how to organize and collect structures of social
networking so that there are discussions about in-groups.
The first time I got into reading blogs on Second Life and went into Prok's formulation of the
"Feted Inner Core," I just sort of sat there and said, "Oh, nothing's changed in 20 years."
And there were discussions internally about people who were on the inside and had their
own basic insider loop on the discussions and could influence how things changed, and
people who couldn't. So all of that's familiar.
ROBERT BLOOMFIELD: Let me ask you, David--so you're saying, I mean, you were, as
part of your IBM job, studying these social issues?
DAVID LEVINE: Not intentionally. I mean, if you build it--
ROBERT BLOOMFIELD: Sort of by necessity, is that right?
DAVID LEVINE: Right. Exactly.
ROBERT BLOOMFIELD: I assume that your background is in computer technology or
engineering. You don't have advanced degrees in psychology or social work or
DAVID LEVINE: Not per se, no. I am an old-fashioned programming type, with a BA with far
too many majors, including computer science and pieces of masters and PhD's in computer
science lying around. But it's impossible I think to actually build and run and be involved with
building and running something that does social computing. In this case, these are large
threaded discussion environments.
The very first piece of code I built for one of these was a giant indexer for helping us find
things in the tens of thousands of lines of discussion. And as soon as you started being able
to do that, to search things and attribute things back in formed threads, all the social
interaction issues came up, and we had to have discussions on how do you cut off threads.
How do you moderate threads? What do you do when someone wants to retract a
statement, which has now been indexed in search? Flame wars--all the things that the
internet and the social computing world have discovered over the past 20 years come
So the moment somebody says, "It's your job to read everything that gets posted to this
bulletin board system and steer the conversation away from bad topics and quell the flame
wars," you're studying it.
ROBERT BLOOMFIELD: And would you say that you or your colleagues at IBM have made
much progress on that front since the Police and The Arrhythmics were the number one hits
on AM radio?
DAVID LEVINE: Yes and no. I think the maturity of most people's interaction online has
gone up dramatically. And certainly inside a professional corporation it's a little bit easier.
However, that said, a lot of the same issue obtain. We have a joke that was specific--all of
our discussion threads were called forums, and there was a specific forum that got started
to discuss a Byte Magazine article in the mid '80s about the history of the computing field.
And for a while there was a joke that every, single non-technical discussion became a copy
of that thread eventually. Which sooner or later somebody broke out the old hardware that
they had worked on, and it just drifted down into this rat hole. And that hasn't changed a bit.
There are topics that you will drag into. And there are behaviors of people engaged that are
unchanged 20 years later.
On the other hand, people have a much richer appreciation for what happens when you
speak in public. They know that they're going to get quoted. They know at some level that
you can't disclaim things you said five years ago. And I think that's made life a little easier.
ROBERT BLOOMFIELD: Well, if someone will just tell my kids that as they post their entire
lives on Facebook.
DAVID LEVINE: Well, yes. And I think we're seeing generational divides there. That's not
for now. I'm having a rich discussion with a couple of people on what happens when the
social networking tools and the 3D internet worlds blend? Questions like: What happens
when Flickr and Second Life and Twitter all interact? But that's a much messier discussion
than even this one.
ROBERT BLOOMFIELD: Well, maybe we'll get there toward the end of the show. For now
let's move into some of the more technical stuff that you're doing today with Second Life.
And so as I understand it, some of the big things you're working on now in Second Life are
the Architecture Working Group and OpenSim. And I guess drifting throughout both of those
we also have interoperability. I don't want to make this too technical, but I think a lot of our
listeners and viewers are even less technical than I am. So if you could just give just a real
broad brush on what these three efforts are before we dig in a little deeper on the business
and policy side, that would be a big help.
DAVID LEVINE: Sure. Let me touch onto three things quickly. Linden Labs this past fall
formed an internally sponsored working group of residents and involved participants called
the Architectural Working Group because they'd noticed that they were coming up to a point
where they wanted to do some major reworking of the core architecture of the environment.
And as they were doing it, they realized that there was a really good chance to also open up
some of the key interfaces and make them far more web standard.
And so they invited a bunch of people who had been involved and interested in the topics
largely out of interpersonal attractions. It was largely done from people who had attended
office hours or had engaged--and they also had discussion and so on, to come and meet in
Linden Labs out in San Francisco and spend a day talking about this.
That group has led to an ongoing effort to build out a consensus on how Linden is going to
evolve their world to allow real interoperation on the grid. I don't want to dive at all into the
technical details; I want to mention two other things and then talk about some of the social
implications of that.
The two other pieces are--there's a project that is an open-source clone, effectively, of the
server side of Second Life called the Open Simulator Group, or OpenSim for short. And that
has reverse engineered the core of behavior of what Linden Labs servers do, and allows
you to run your own stand-alone island or grid of islands that look just like Second Life to a
large degree. It runs on the unmodified client. It has most of the properties you'd expect. It
has almost none of the content. It's a private little cul-de-sac in general. People can bring
content in, but if you have content in Second Life, it doesn't come in. And that is, in some
sense, the heart of the discussion, which is interoperability, which is the third piece of this,
which is: At what point does it become possible for someone to host their own island or
cluster of islands that in some way connects to other people's clusters of islands and into
the broader grid. That's, at some level, the heart of what I'm interested in because that
introduces a ton of interesting changes to both the technical landscape and perhaps more
importantly the business and social landscape.
So in particular, yes. And if you look in chat, you'll see Saijanai is busy popping earls up so
you can go and click on them in the web. And I highly recommend you do it if you're
interested in that.
In particular, one of the insights--and Linden has been very clear on this from early on--and
as a professional judgment, I think they're right--is that this isn't going to stay a closed space
forever. And in fact, not only isn't it going to stay a closed space, but it's not in their interest
at some level to stay in the close space. And so what they've said is--they had a choice. You
can either fight it or embrace it. To the extent their attitude appears to be to embrace the
fact that we're going to see an open-up world.
And as that has happened, there is now an opportunity to start defining two or three parts of
meta-interaction. One of which is the technical side, which is the hard grunt work. But far
more importantly, the policies and social behaviors that we want to see. And those are, in
my mind, at least as important as the technical work. What happens when you want to go to
visit somebody's private island? What is your expectation? What are the property rights of
creators in the main grid? How do we respect that? How do we get desirable behavior so
that you don't go to someone's private island and have all of your assets stripped or have
your behavior imposed on you from outside? All sorts of questions like that come out.
ROBERT BLOOMFIELD: What you've been saying raises a lot of questions in my mind. I
have the privilege as host that I get to ask them myself. And so I'd like to ask a couple now. I
also want to just encourage the many people who are here on CMP and people that are
event partners, like Muse Isle and Rockliffe University, if you have questions you can IM me,
Beyers Sellers, directly, or you can simply type them into the Metanomics chat. And we will
make sure that your questions to David get posed before--as many as possible anyway--
before we end.
So I'd like to go back, actually, David, to one of the first things that you talked about, which
is the nature of this Architecture Working Group being a collaboration between Linden Lab
and a number of residents. And so last week I had Michael Wilson, CEO of There.com, on
Metanomics. And he talked about something that they do, which is that they have a constant
polling of the top ten changes their residents would like to see within There.com. And so I'm
wondering if you can talk about the difference. Are these similar strategies, or is this a very
different way of the world developer working with the world residents?
DAVID LEVINE: Let me see. I'm not as familiar with the way There.com approaches--so it's
a little hard to contrast it directly. But I certainly think that what Linden is doing here is
somewhat unique in that they're saying essentially, "We are willing to essentially ask any
resident who's got an interest or a concern, and some technical strength, to get directly
involved it the process." And not just in the "This is what I want" but to actually sit down--if
you sit down at Zero Linden's office hours and ask a direct question about some of the
choices he's making: Is this a good technical strategy? What will the impact of this be on the
store I'm running? What will the impact of this be on my ability to do X? They are engaged
and willing to answer those questions. That's an unusual degree of direct involvement.
Now, it's got its plusses and minuses since literally it means any resident on Second Life
can pop into those office hours and ask any question. And as you might imagine, some of
them are perhaps less well posed than you might like. But it's level of openness is really
quite remarkable. And long term, the really interesting thing is that they're also working
directly with people who are in the Open Simulator Group, OpenSim, and saying, "How do
we get a collaboration here that actually, directly impacts what we're doing so that we can
say in this cycle of work, six months from now, eight months from now, nine months from
now, these interoperation points will exist?" And that's, I think, going to be really of the
essence of what happens, which is this collaborative work that leads to a shared
understanding of the way to grow the system out in a way that allows us to actually link all
the pieces together. And that's fairly different.
I was going to grab a question out of the Metanomics chat because it's a real important one.
ROBERT BLOOMFIELD: Oh, okay. Wonderful.
DAVID LEVINE: So the policy issues and the social issues. I'll start by quoting something
that Mark Lentczne, who's a Zero Linden in-world, said right at the beginning of the kick-off
for the AWG, which is that it's his personal understanding--and I think I can safely represent
it as part of Linden's understanding--that the social implications for this are every bit as
critical as the technical ones. In other words, it serves nobody's interest to do a bunch of
technical work that breaks the social consensus and breaks the things that make Second
And there was actually a particularly lovely phrase that they used, which was an avatar Bill
of Rights, which are: What are the expectations of residents living in the world as you build
this greater community? So I think there's certainly an awareness there. And I say this very
consciously that one of the things that I say about the way I approach this is that I am in this
space as a technologist, but I'm also in this space as a resident. I go and listen continually to
live music. I've got friends all over the world who are involved in Second Life, many of whom
are not technology people at all. And if you wander around Second Life, I’m fairly visible,
generally without the IBM architect tag over my head since I'm not working at all in the
professional capacity. And it's the first line in my profile that--I don't know sometimes the line
between business and work, which is when I'm sitting around in the shop talking with
somebody who's upset because they haven't figured out how to effectively address the fact
with Linden that their textures have been stolen. Can I distinguish that from sitting down and
having a technical discussion on how the protocols work? I can't; they're both parts of the
work. And some of those happen sitting around listening to live music in the background.
ROBERT BLOOMFIELD: I'll say I have so many questions coming in. Let me dive in with a
couple more if I could.
DAVID LEVINE: Sure.
ROBERT BLOOMFIELD: So one is--this has been sent to me, and they've asked to be
anonymous. But the question is: "Is there any financial arrangement between Linden Lab
with any members of the Architecture Working Group?" It's a compound question. "How are
people's contributions to the project recognized and credited?"
DAVID LEVINE: Well, let me start with one thing, which I know is public knowledge, but
should be mentioned. IBM and Linden have a joint development agreement for some of
these areas which include some of the work we're doing here. So at some level, my
involvement in the AWG is certainly entangled with that work. In general, as far as I can tell,
and I can't speak for Linden, and wouldn't want to, any work that's done in the AWG is done
through the Wiki in general. And if you look at the Wiki, there is a contribution agreement
there which has some very specific terms. Basically, it's a Creative Commons agreement.
And I encourage anyone who's interested to go read it. If they're interested in contributing
intellectual property, I encourage them to think hard about what the legalisms mean or go
find a lawyer or their corporate lawyer if they live the corporate setting, and go make sure
they pay attention to that.
That said, it basically is a share and share alike environment. If you're going to work in
there, you're saying, "I'm not bringing encumbering intellectual property to this space. And
I'm going to share it freely." There's no expectation of compensation. And I think that would
be unrealistic. It's largely like an informal standard's body; we discuss, we talk about how
we're going to codify this, write it down and build it. And the expectation is that we're going
to end up with a public specification for how the pieces interoperate.
And I think in general, so far, it's been a very informal collaboration. No one is looking to get
paid to do this work, and I think that's unrealistic. The benefits--the reasons why a company
like IBM might be involved, the reasons why companies like Intel and so on are expressing
interest is because we end up with this very rich technical space in which we can build
richer products. It's nothing more complicated than that.
Just to follow-up on a couple of questions that I saw flop by in chat here: "Is it the techies
defining the policy?" It's hard to say. Certainly being at the table as a technologist has an
impact on it. "Do we talk to lots of other people who aren't writing code?" Yes. "Am I
accessible, for example, to anybody who has an opinion about some of these issues?" My
IM is open to you. Give me a holler; I'll go grab a chunk of space on our island; we'll sit and
talk about it.
So I'm not sure that saying the techies are making the social policy is fair. And accountable--
I don't even know how even answer accountable in the sense that--what the right framework
for that is. Our success or failure at this is going to depend on getting that right as
technologists and as residents and as people who are trying to figure out sensible ways to
think about this space. For those interested in those topics, I especially encourage people to
take a look on Tish Shute's blog ugotrade.com.
Back in December there was a very nice discussion that I had and that Tish held with Eben
Moglen where we talked about a lot of those issues about what expectations of being a
citizen in the virtual world might mean, and what right policies are. And the boundaries
between hosting land in a virtual world and what your responsibilities as a hoster of land
might look like. I think those are important topics.
ROBERT BLOOMFIELD: I'll say there are so many questions coming in. I want to try to get
a bunch of these. Here's one--I don't think we want to make this day just a litany of
complaints about particular features we'd like to see or wish lists. But there's one here that's
very close to my heart. It comes from Joia Sands which says--this is actually I think mostly a
complaint about our Metanomics show--Joia says, "Could you animate the presenter and
panel avatars a little more to make for better TV? Watching an avatar is boring." Now, I have
to say, having done a show last week in There.com, actually, my lips moved. It was a very
exciting moment for me. And I've been talking with some folks here in Second Life dealing
with the technology here who are saying that actually the way avatars are structured here,
it's actually very difficult to get that done.
Bevin Whitfield(?) is saying that is done at MetaNetwork News. But I believe that's after the
fact as opposed to live. Do you know anything about that specific feature?
DAVID LEVINE: There are a couple of interesting things there. And they fall broadly into
this category of trying to make the behavior of your avatar more closely match what you're
actually doing. And this is a rich and messy area. If you look at most social settings, most
social settings in Second Life feature dancing or various other forms of animation of the
avatars frankly, because otherwise, it is quite boring. At the same time, while I could
imagine that there would be something that would sync my voice to the spoken inputs and
especially allow expressions and stuff--and there are HUDS--you can buy HUDS that let you
change your smile, change your look, raise an eyebrow and all those things. The effort for
that is simply staggering. And it gets disruptive to the process of actually just talking to
people. And I think that it's a ways away from this.
I joke about this all the time, which is conversation--I'll say things especially when I'm talking
in voice, but also when I'm texting. I'll be raising my hand in real life. I'll be saying, "Me, me,
me." And my hand will be up. And of course my avatar is sitting there floating in a nice lotus
position, bobbing up and down a little bit and nobody can see that. So technically there are
a lot of ways we could add to that. Thus far, most of them, body language essentially, as
someone just commented, require a lot of overt work. And I think that's a real limitation.
If you go deep, deep back into the roots of Second Life, there was some work on what are
called haptics, which are live connections between people's actual physical motion and their
avatar's motions. And they represent some of that. And the problem is do you really want to
go put on something that senses your body motion in order to interact with somebody in
Second Life? Do you want that your avatar's arm is linked to what you're doing in real life?
That's a hard question; but I think it's a great question.
I expect that as we go on we will continue to see more and more of this. And that's exactly
right--another comment that just went by that voice and text is ten percent of conversation.
And it's one of the biggest problems when you hold meetings here is catching people getting
bored, catching people distracted, seeing who is looking and paying attention to you.
ROBERT BLOOMFIELD: I noticed in There.com their software actually does make that
easier. I'm not sure I viewed it as a positive when I was doing that show. But basically, if I
used any other program other than There on my computer, these goggles would go up on
my head and make it very clear to everyone that I was multi-tasking and placing higher
priority on something else, which made it rather difficult to do a show where I'm naturally
paying attention to many things at once.
If you don't mind, let's talk a bit more about OpenSim, which is--you used the term reverse-
engineering. And correct me if I have this wrong; my understanding is that Linden Lab is
open-sourcing--has already open-sourced the client so you can then reverse-engineer a
server that will fit with it. And so I've got a bunch of questions on this, and I'll just lay out a
few of them. And you can take them as you wish.
The first one is it seems like there've got to be some intellectual property challenges here.
Reverse-engineering is often viewed as a path toward patent infringement that you are
getting a functionality possibility in a different way, but the functionality itself may be
patented. And so I'm wondering how that is being addressed.
And then the second issue is do you see--right now we've got a single--Linden lab has--well,
actually they no longer have a single viewer. They're looking at different viewers. And other
people presumably then would be starting to create their own new viewers to emphasize the
features that they would like, which means that we would see divergence of the servers as
well, which may conflict with the goal of interoperability. So let's start with either of those.
DAVID LEVINE: Let's start with the easy one. The reverse-engineering thus far has been at
the protocol level, which is pretty much in most spaces completely safe. In particular there's
no reason to feel that's too problematic. Furthermore, there's been a very rich and
successful amount of collaboration between people in OpenSim and Linden Labs directly.
For example, there are things that are indicative of that. If you were to go down to the
Mountain View offices, you would actually find a small cluster of OpenSim boxes running. I
know Mark Lentczne sat and did due diligence on what's inside, what the capability is, and
what the functions that are supported in OpenSim a couple of months ago.
So there's certainly no lack of awareness or hostility by Linden about the project. And in fact,
this latest round, one of the explicit things we're doing in the Architectural Working Group is
discussing how we're going to do early operability testing and how we're going to be able to
make sure that what we build works across both sets of the world.
Now, part of that is simple pragmatics: you can't actually say we have a product that
interoperates if it interoperates with itself; that's meaningless. And so at a minimum, Linden
Labs needs something else to stand as the other endpoint for the discussion. Does it in fact
interoperate with something else? And OpenSim is obviously one of the early candidates for
Several of the key technology people who have been involved in the work, Adam Frisbee in
particular, have been to Linden Labs and had rich discussions with them. And certainly
there's no hostility there. It's very much an open collaboration.
Legally, reverse-engineering is a funny thing. You actually go in and start prying apart
people's algorithms, you can run into trademark and copyright issues galore. Fortunately,
I'm not a lawyer; unfortunately, I work with lots of them by the nature of the job I do. And so I
know a fair bit about this. And you have to be careful. However, I haven't seen any issues
that have show up that are problematic there.
The second half of the question I think is a little more subtle. And there are a bunch of
issues that have to do with--forking the world is the general phrase for it. A fork being a fairly
technical notion of what happens when your code base suddenly goes down two legs of a
road. You the intersection. And some people go left; some people go right. And the problem
with that of course is you get incompatibility.
The best that can be said is that most communities don't do well with when they fork
because now you get competition for resources and competition for users. And Linden and
certainly the OpenSim community is aware of this. And we are certainly going to watch
closely to try to prevent that. If you go look around at the moment, there are about, to my
count, six or seven fairly diverse clients now.
ROBERT BLOOMFIELD: Would you mind--I mean the ones you names of. I know OnRez
which came out--Electric Sheep.
DAVID LEVINE: There's OnRez, which is a special case. Right, that's a special case.
Nicholaz Beresford done a very extensive set of patches. There's one that shadows that
which is in a different set of patches. There's one which has been done for Linux called the
Cool Viewer(?). There are a couple of others which have added either 3D specific features
out of--I want to say University--I'm trying to remember if it's Michigan. I want to say it is. In
each of those there are somewhat separate patch trees. They're all open-source GPL
because that's required. And each of those effectively--some of those features have been
attributed back to mainstream viewer, and some haven't. And managing that is going to be
an ongoing challenge for Linden Labs. But on the whole, it's been pretty successful. They've
gotten some very real benefits. New features have been demonstrated and bugs and fixes
have been found. That's going to get much worse over time. Managing that richness is
going to get difficult.
ROBERT BLOOMFIELD: Now let me say, it's the second time you've referred to this variety
as a bad thing.
DAVID LEVINE: Not a bad thing; it's a challenge.
ROBERT BLOOMFIELD: Yes, a challenge because it seems like there would be
tremendous benefits. And if we just think of evolution in a very biological perspective, we're
thinking about genetics and variation and competition among similar agents for the same
DAVID LEVINE: Absolutely.
ROBERT BLOOMFIELD: This is how you get progress.
DAVID LEVINE: In fact, that's exactly right. And that's why I think this is so important. The
reason why OpenSim and Interop really work matters to me is that--and I think to
corporations and eventually to residents--is that it allows evolution to happen. It allows us to
in parallel explore five or ten or more variations and ideas.
The first example of that for people who have not yet tripped over it is a project called
realXtend, which is being done by a couple of Finnish companies, which has added a whole
bunch of support for meshes into the OpenSim project, into a modified client. And that's
something that Linden isn't currently interested in looking at, but over time knows they need
to look at.
There's been a lot of discussion in various groups about different mechanisms of hooking
the OpenSim framework into people's corporate frameworks to bring data sources from the
outside world into simulated regions. There's been discussion about building substantial
variance on physics. Right now the world is essentially flat and has very earth-like physics.
There's no reason why we need to do that. And people are going to explore that.
I think launching that creativity, getting to a place where we have multiple parallel groups of
technically interested people pursuing different parts of the space is absolutely essential.
What happens as that point--and this is what happened in the web in the period when
Apache suddenly took off--is all the sudden you get hundreds of people throwing ideas into
the space, some of which stick. Many of which turn out to be bad and get discarded. The
core ones get adopted by the community and become part of the standard. And I think that's
the process at some level we want to get to happen.
I talk about igniting an ecosystem. And since you said evolution, that's exactly the word I
tend to think of, which is if you want to get to a point where there's enough richness and
enough variation that we start getting competition.
Somebody asked me a while ago, "Why is it that Second Life has the most interesting
looking avatars?" And I thought about that for a very little while, and the answer is because
there are thousands of residents who have shops who are trying to sell me hair and sell me
clothing and sell me good-looking shoes. And by golly, the ones that win are the ones which
are better. And every time somebody gets better at it, the other guys have to raise their bar.
They say, "Oh, well, halfway decent prim shoes don't work anymore; now they actually have
to have detail; I have to have shoelaces." And suddenly all the shoes have shoelaces. And
suddenly all the hair moves well. And that kind of evolution is what gets you great content.
We need that same kind of evolution in the technical side. We need people pushing the
frontier of avatar animation of how we build landscapes; of how we interact with technical
tools so that we can do things like bring our calendars into the world and actually work over
At the same time--and this cuts away from the technical and back to the social side--if in fact
that's some of the heart of the success of Second Life that's enabled people to build this
competitive space and compete in it and benefit from it, we have to make very sure we don't
break that. That's really sort of the essence of one of the lessons, which is, it's easy to look
at it as a purely technical problem. If we look at it as purely technical problem, we miss on
things like that. And so balancing those two becomes one of the challenges that we face.
ROBERT BLOOMFIELD: I see JimmyJet Fossett is saying it might be that user generated
content and all this natural competitive process may not exist as much if we have those
closed virtual worlds that are sharing similar technology but we don't have the free
communication and [CROSSTALK]
DAVID LEVINE: And in fact, the analogy that I look at there is if you look at the internet or
the proto-internet, in some sense, back in the late '80s, early '90s when most of the content
and creation was going on in places like Prodigy and CompuServe and AOL, and they were
essentially closed boxes. There was this lovely period where you could significantly get
traction by going to a private company and saying, "I want to put my storefront up, not on
the web, but as part of AOL or as part of Prodigy." And you would go and build this very
customized content. And the only people who could see it were the people who were
subscribers to that service.
Right now that's where we are today. There's lots of cool content sitting around in places
like Second Life and some in There.com and some in EVE online. And some even if you go
look at it, there's some very cool content in the games. If you look at World of Warcraft,
they've got some very visually appealing stuff. None of that interoperates. None of that is
available to be shared yet.
There was this huge inflection point where all the sudden it became obvious that all the
content was going to essentially escape those closed boxes and be shared. And if you were
on the wrong side of that, you were acquired or you vanished.
Now I see floating in the bottom of my screen the canonical question: "People who are
textures ripped off?" That question. And what do you tell them? What we tell them is it's a
nightmare. And we tell them that in fact this is a deep social problem not a technical
problem. And just to sort of touch on the technical for a moment and then circle back to the
The technical problem is the moment I give anybody a texture to sell, and I sell it to them,
and they put it on their avatar, and it gets presented out to the world, that texture is
physically on everybody's bloody machine. There's absolutely nothing I can do about that in
that sense. It's sent over the internet and it lands on everybody's machines, and it's
rendered. And there are about four places along that pipeline where you can snag it. That's
ROBERT BLOOMFIELD: I just want to clarify. You're saying that--not just that someone
can snag it, but they can then reuse it, modify it however they want?
DAVID LEVINE: Right. Now that's certainly unethical. And at some level it's illegal. Exactly
how unethical it is depends on how you define ethics. How illegal it is depends on how you
define copyright. I'm not going to go there.
It's far worse to take a screen shot of it. You can actually intercept it at the texture level in
the content pipeline as it comes onto your client. And you can pretend to be the client and
do that. You can grab it at the open GL level. There are tools that will reach in and grab the
texture pieces in open GL and put them onto your hard disk. So there's plenty of ways to get
at the content. And that's unavoidable.
ROBERT BLOOMFIELD: Someone is mentioning watermarks as a way that--
DAVID LEVINE: Right. So the classic strategy for this is accept that the content is out in the
wild and then figure out how to enforce your desires. And there are two halves to that. One
of them is the system needs to indicate your desires. So as the creator--and Second Life
does this to an extent today. As the creator you can say, "No copy, no model transfer," and
so on. You can set those marks. And those say, "This is my desire for this piece of content.
I’m willing to give you a copy. You can have as many copies as you want, but you can't
share it with your friends." Or "I'm willing to give you a copy of this, and you can have one
copy of it and give it to anybody you want, and they can have one copy of it." The richer
model of that is, "Here's the Creative Commons license I wish to apply to this piece of
content," which is certainly possible.
Now, watermarks are a way essentially of saying, "Ah-hah, I know my content is at risk, so I
want to put a fingerprint on it." and there are a number of ways we can do that, and some
people do. It's a very common way of protecting content on the internet. You put a very
subtle possibly undetectable, expect by knowing how to look for it, digital fingerprint on the
item deep inside the JPEG encoding, such that it doesn't interfere with the actual
visualization of it. But if somebody uploads it to another place, you look for it and say, "Oh
look, they sent me a copy with that watermark on it." And you say, "That's illegal. You've
stolen my property."
Now, the reality is that even doing that you then have got to figure out how to enforce it. And
from my personal experience of [consecrators?] in the world, that's been the weakest
problem. It's not detecting the theft; it's actually getting somebody to do something about it
because of the continual problem.
ROBERT BLOOMFIELD: This puts us back into the social and legal space.
DAVID LEVINE: This puts us back into the social and legal space, exactly. My personal
take on this and the one that I repeat in painful detail to anybody who gets me on the
subject is, as a technologist, we have to make sure we mark things right so people know
what the intent is. We need to make it clear that if you steal someone's texture, you're going
against their permissions; you're violating the social contract; you're probably violating a
legal contract. We then need to, as much as possible, provide social mechanisms for
ROBERT BLOOMFIELD: Now There--you were there last week for my interview of Michael
Wilson. And he was very clear on this in two ways. First he said, "Stealing intellectual
property is simply wrong, period. End of story." The second is that There.com, the world
developers themselves, are taking a very active role in dealing with this, and they vet
content. Now that seems unlikely to work in Second Life. Do you think Linden Lab needs to
be taking a more active role in this as opposed to just saying, "Well, we need to set--as you
indicated--sort of social norms and power residents to protect IP and all that?"
DAVID LEVINE: The way I would put it is this way--and this is again very much a personal
opinion. I would say talking to content creators they don't feel that the current mechanisms
are working. So I think there's a concern there. Now more broadly, it becomes much harder,
indeed, almost impossible to figure out how that would work for in a fully open environment
in that--at that point, once you have large interoperating things, you have to start defining:
What legal relationships do I have between the people hosting various parts of the grid?
And part of that discussion is going to have to be what will you enforce in these spaces.
And there's a lovely phrase that shows up in my interview with Eben Moglen, which he
coined, which I quite like, which is: "At some level, having digital angels looking over your
shoulder," which says essentially that if you share this content with this part of the grid, you
will be at risk. We need to do things like that so that we can mark it so that, "No, I won't in
fact allow my--I will tell Linden Labs never share my content with other regions that don't
have a legal agreement to enforce content protection." We could imagine doing things like
If we don't build structures like that, the content creators of the world are going to be in a
great deal of pain. There's already a lot of pain. And we can't make it worse. Making it
worse, frankly, is not an option if we want to succeed it creating the kind of environment that
I think we need to create.
ROBERT BLOOMFIELD: I'll say this has been a fascinating discussion. We've got about
ten minutes left or so. I just want to say I hope that viewers of Metanomics are learning a lot
from this because this is my hour of education every week where I learn more than I do the
rest of the week combined, usually. So this has been great, David.
I'd like to change the subject fairly dramatically if I could because you have worked
extensively in 3D visualization which is a broader topic than just virtual worlds, and
something that IBM, which spans both the technical and the consulting side of IT, and more
general business consulting, I mean 3D visualization of data is a fascinating idea.
Can you talk a little bit about, I guess first, what it is that you and your colleagues are doing,
like what opportunities you see for 3D visualization. And then maybe, if we get there, where
virtual worlds fit in.
DAVID LEVINE: Wow. That's a huge topic. That's two or three hours right there.
ROBERT BLOOMFIELD: Well, I'll have you back on at some point.
DAVID LEVINE: Okay. Actually, you have me back on, and maybe we should host it on the
island where we have our biggest piece of visualization in-world.
For anyone who hasn't seen it, and I know most people probably haven't, one of the things
we did fairly early-on in exploring visualization in Second Life inside the research division of
IBM was take a piece of molecular simulation we did on a super computer called Blue
Gene--of a molecule. It's a tiny fraction of the molecule in fact--and rendered it as a
molecular model that floats over the middle of the a lake in my island. Anyone who's been to
the Architectural Working Group meetings at Thornbridge Town has seen it floating over the
lake. It's 56 hundred and some-odd prims of visualization. And you can actually fly around it
and look at the way the bond patterns work. And it's color coded to let you know which are
iron bonds and which are copper and so on. And you can look at that and get some sense
of how this very visual structure works.
Since we let the computer people play with it, it's the first molecule in the visual pathway. It's
called Rhodopsin, and it actually grabs photons. And that starts the whole visual cascade.
So of course, you let the computer people loose and you say, "We want a visualization
molecule." And they say, "Ah, a visual molecule." Of course, horrible puns.
At the same time, it turns out to be a incredibly powerful tool for starting to see what you can
do with the space. You can actually stand around it. You can actually sit in it. And in fact, the
first time we built one, we sat down on it. It was really entertaining. It starts to rez. It's
scripted out in a huge scripting thing that actually floats the prims into place. And it takes
about 40 minutes. At the time it took around 40. I think we're down to about 20 now to create
the entire piece.
I think we were ten percent of the way through it, and one of us--and none of us will admit
who it was first. I don't think we even know anymore--clicked on one of the prims and sat
down. And of course somebody sat down next to us. And we probably sat there watching
this thing build in the sky in a social space and suddenly realized that we could hold
meetings here. And we have ever since. It's a really interesting cross-over.
I think what we're going to see over time, frankly, is all sorts of visually rich things are going
to built in these environments because you can collaborate over them. So we've built things
like visualizations of actual software structures. So for example, there are these huge,
messy things called enterprise system busses, which are the gory details of how very large
applications are bolted together. And data flows down them and connects various huge
software components. And they're messy. They will have ten or 20 boxes worth of stuff.
People cover their whiteboards with them when they're trying to discuss them.
You build it as a 3D space and animate it with actual data that shows the traffic that flows
over this thing. And all of the sudden the phrase "hot spot" becomes very visual. And all the
sudden the notion of "that box that I can now point at and stand around is the one that's
causing our performance problem" becomes very compelling.
I will share another one which is notions of--right now we do almost no collaboration in-
world. If you look at some of the other spaces like Croquette and Quack, they do a little bit
more of this. It ought to be possible when we're building a document and editing it, for three
or four of us to stand around, walk around on it and point to the box we don't like, walk our
avatar over to it and say, "This box over here is the one that I think doesn't convey the
message well." And have everybody see the change I'm proposing and share it
There's going to be a huge amount of visualization of data because as this technology
becomes commonplace as opposed to obscure and expensive; everyone's going to have it
on their desktop.
If you look actually at the adoption curve of Second Life, there was a huge hiccup that
happened in late 2005, early 2006. And I think it goes down to a very funny thing. That was
when laptops out-placed desktop machines that everybody's buying, especially corporate,
but also for a lot of people. And the first generation of very common, fairly powerful laptops
didn't have a card that could support Second Life, didn't have the graphics to do it. And
there was a real pause of people about--at this point you can go out and buy a laptop for
$1,500 that will run Second Life as well or better than your sort of mid-range gamer
machine. And that's been a huge change.
The hot spots demo--I don't have a URL handy, but it's been blogged extensively. Look at
Virtual Data Center online. There's been a couple of blogs for it. And if people IM me, I can
get it. It's not, I think at the moment, set up in-world. But we can probably get that done
again. It's fairly compelling when you actually look at it.
And when will we be able to import stuff? Go write code. No, that's not fair. It's something
that's happening slowly. One of the problems is that, frankly, the rendering engine here is
limited for some of that. And over time it will get a lot richer.
One of the reasons why OpenSim becomes important, why getting to rich interoperation is
important is we can build specific versions of sims and clients for high-end stuff in advance
of the world being able to support it. So for example, one of the things that we've talked
about doing with some of our potential customers are very high-end visualizations that we
simply couldn't do today for every client and every resident in Second Life. But if we own
both ends of the pipe, we say, "You're not going to log onto this sim with a low-end device."
We're only going to support clients which actually have very high-end video graphics card or
very high-end ATI card. And all the sudden we can put things into the client which suck up a
lot more resources and still get reasonable frame rates.
Those things become mainstream over time. If you look at the adoption rate of graphics
technology, 18 months is the double-end curve. Roughly every 18 months you'll see the
basic frame rate triangles per second rendered doubles. So what's unsupportable in 2005
becomes very supportable in 2006, and becomes a cheap gaming machine in 2008. And
the question will become: When does that stuff become mainstream, and it's just a matter of
ROBERT BLOOMFIELD: Let me follow up a bit on the visualization issues that you were
describing. So here's a question I've heard a number of people talk about in different guises.
It's: What's the use of an avatar? And I think I heard a lot about this--initially people were
saying, "Oh, virtual worlds are going to be the future of e-commerce." And then the question
is, "Well, who needs an avatar to buy a book from Amazon?" Virtual worlds are going to be
the future of online collaboration. Well, again, why do you need an avatar to share data?
So now you're talking about a 3D--I can see the use of the virtual world and having a point
of view, but what's the need for an avatar here?
DAVID LEVINE: This is a fairly deep social question about people, I think. So let's back up
a little bit and ask a couple questions. Are there things which don't benefit from avatars? I
think the answer to that is, absolutely. So for example, the classic example is Amazon.com.
Do I need to log into a 3D virtual world to go find a copy of a Noel Coward play that I want to
read or buy and order from Amazon? No, I don't. In fact I'm far better off going to their web
page, typing in a text search, and getting the response. On the other hand, when I want to
find out if the latest Charlie Strauss book is any good and has some interesting insights in
virtual worlds, I'd like to talk to a bunch of people. And at that moment, the question
becomes: Well, how do I talk to them?
And if you look at the traditional web today, the best you can do is to go look at the reviews
of the books. You go to Amazon or any of the other popular commerce sites, and of course
there's this huge list of people's reviews and comments and feedback and so on. But you
can't actually interact with them because they posted them two days ago, and you don't
actually get to know what they're doing or what their thoughts were. You get their distilled
moment of truth: "I like this book because I thought the main character was very
compelling." But you can't talk to them.
The moment you get to a social space where there are ten or 12 people, well now the
question is: How do you interact with them? And there's a bunch of different ways of doing
presence. The classical form online traditionally has been what's called green-dot presence,
in my mind, which is, if you go look, if you go onto AOL Instant Messenger or Pigeon(?) or
any of the tools, you get a list of your friends, all their names, and there are little green dots
next to some of them, and there are gray dots next to other of them. And you know, "Oh, so
and so is online." Well, barely. You know that Zha Ewry at Gmail.com is online if you're on
my friend's list. And that sometime in the past three hours I logged in, and maybe in the past
half-hour I ran my mouse over the window so that the dot stayed green. You don't know if
I'm in the office or not. You don't know if I'm interacting or not. And you certainly don't know
if I'm completely distracted in the conversation at Metanomics and therefore paying no
attention to that window at all, which is in fact the case. I have noticed several blinks on that
window, and I haven't answered any of them.
You go to a 3D space and you find me in-world. All the sudden you see what I'm doing. So if
you actually come and find my avatar in-world, I'm floating here in this space. There are
dozen people watching me or more, and you know exactly what's going on. And it becomes
a focus issue; it becomes a way of telling you, at a very deep level, what I'm doing. And if
I'm text chatting, you can also see what I'm saying. If you're busy listening to the stream,
you know what my thoughts exactly are. All of that ties to the avatar. And then it serves
some social needs which is to hook memory, hook visualization, hook people's needs of
interacting with people rather than blank screens.
One thing we've noticed is that in general, not always, but in general, meetings held in these
kind of spaces are significantly more productive than either phone calls or pure text chats.
We have that technology. We've had that technology for years inside IBM and across the
globe to hold IRCs and so on. [SKIP] Oh, is it really that distracting?
ROBERT BLOOMFIELD: Yeah. Actually, speaking of avatars, this is probably--I see CeAire
Decosta pointed out that your avatar is rather distracting. I don't know if you have any
interest in talking about your avatar choice or how that might fit into this discussion.
DAVID LEVINE: There are some very deep interesting issues about how people do that.
But not a topic I'm particularly interested to dive into. People who want to talk to me about it,
you know where to find me. I'm not exactly hidden in-world; go ask me. I will observe that
there are lots and lots of choices about avatars. It's as reasonable to expect, when you
wander into Second Life, that you'll find somebody who looks like a dragon or a small
elephant as it is to find them looking like a human being. And certainly the world would be
far more boring if there was a requirement that we go and went to some booth some place
and have our physical presence be scanned and look like we do in real life. That would be
kind of dreadful. As soon as it stops being one-to-one, the question becomes merely: All
right, so what presentation do you choose? And once you get to choose, people come out
looking very different than they do in real life. And it's, I think, part of the charm of the space.
ROBERT BLOOMFIELD: I see we are really running out of time. Do you have any--I'd like
to just give you a little time right at the end of this if there's anything you'd like to talk about,
predictions of the future, calls for action. I give it to you.
DAVID LEVINE: I think we hit most of the issues I still have on my short notes and what I
was going to talk about today. For anyone who's watching, I'm wide open to IMs. I get
capped of course. If you don't get a response from me, tap me again. And I'm absolutely
excited to talk to people about almost all of the content here that we've been talking about. If
you're interested in the technical side of it, grab me or Saijanai who's here now, and ask an
invite to AWG. If you have insights that you want to share or questions you want to ask, get
a hold of me; I'm very interested in the entire space.
So the major thing would be to--my message to people would be to get involved. If you have
thoughts or things that the Architectural Working Group is going to do something incredibly
stupid, sing out. Tell us. Get a hold of people involved. And we'd be delighted to hear from
you. I think that's absolutely essential. This is not a closed process, and we don't want it to
be a closed process.
ROBERT BLOOMFIELD: So let me ask a question on that particular part, which is, there
are a lot of people who I think aren't technically minded or trained, but are very interested in
having a voice on the more--the policy and social implications of what you guys are doing.
We had a number of question early-on about "What are the techies doing creating our social
world for us?" So if someone is interested in that aspect and doesn't really know much
about the technical side, what would you suggest that they do?
DAVID LEVINE: There are a couple of outlets. The obvious one is to get involved with
some of the people who are involved and express the social concerns. You're not going to
get the same feedback from all of them. I'm probably about as open to that discussion as
anybody is. But very few people are going to say they're totally uninterested in the social
side of the issue. Most of the people who are involved in the work have some degree of
interest in why the space works. I know our people--and don't claim to understand them
personally--why they feel this way--who don't think there is a coupling between some of the
social stuff. But I think it's not merely interesting; I think it's essential. I don't think you can
succeed in this space without understanding it. So come and find us; talk to us.
ROBERT BLOOMFIELD: People who don't see the connections probably need to track
down Lawrence Lessig’s Code 2.0. I believe the most recent addition and re-release of his
book Code, where he discusses the code as law perspective, code architecture as being a
form of law, and therefore policy.
Well, thank you so much, David Levine, IBM researcher, for coming onto our show. And
you've been doing fascinating things for the last whatever--25 years or so. And so I look
forward to seeing what you'll do in the next 25.
DAVID LEVINE: Well, I think there's going to be a lot more that's going to happen. And
again, I encourage everybody who's interested in the discussion to get involved. Anyone
who wanted to toss me a friend or an IM, now is probably a good time. And thanks everyone
who didn't because it was really appreciated that I wasn't clicking away furiously.
ROBERT BLOOMFIELD: You have lots to keep track of in these things. Okay, thank you
everyone. And see you next Monday.
[END OF AUDIO]
Transcribed by: http://www.hiredhand.com
Second Life Avatar: Transcriptionist Writer