Avatar, Defined

An avatar is a personality portrait. It is a documentation of personality and community. And like any other form of popular art, it’s an intensely social vehicle constructed for virtual communities.

Though it varies from avatar to avatar, and system to system, we can see some core functions that allow avatars to exist in their very different kinds of worlds.

An avatar is a portrayal, or a documentation, of three primary things:
1) Profile
2) Friends
3) Groups

First is the documentation of the profile. The profile has a picture of some sort, showing what the avatar looks like (which may not have anything to do with its real world owner). After that the profile is a collection of documents; documentation of experiences, memories, thoughts, feelings, and beliefs. Maybe these are poems or essays, maybe they are links to songs or video files. There’s generally something about the physical aspect of the personality – a photograph or drawing (which may or may not have anything to do with the avatar’s real world owner). And often there’s various stats and measurements (age, gender, etc).

Second is the documentation of friends. This is the avatar’s immediate community, family, and close friends. These are the avatars of others that are of immediate interest and are probably similar in style and taste. In Fern’s case it was her husband (the bearded guy in front of the house) and her dog (who got more photos than the husband).

Third is the documentation of groups. This is the avatar’s extended community. These are folks that the avatar considers as important enough to point out. In Fern’s case it was Howard and the other folks that went to Disneyland. This third bit is important because it has a lot to say about an avatar’s social position. We are all born with a family, but the extended network, the extended community, is something that is earned and often times a representation of merit. For example, some users pride themselves on the size of their social network. This is a kind of measurement of popularity and offers many users a measurement of their worth in the community. This is a basic human instinct; if we throw a party we want people to come.

Taken like that, an avatar is a pretty flexible thing. It might be a Playboy Centerfold. It might be a baseball card . It might be a CIA dossier. It might be a Dungeons & Dragons Player-Character sheet.

An avatar is a personality portrait. It is a documentation of personality and community.

Why Your Avatar Is Important [Part II].

… and … Secondly, regardless of whether it is an avatar with a body, or not, you still need an identity to operate in anything that has to do with content, community, or commerce (the holy triumvirate of the internet).

30 million Facebook users can’t be completely wrong.

When social media systems ask us to make personality profiles we are actually making something that very closely resemble baseball trading cards. These are baseball-card avatars. We have a picture that gives some sense of who we are, or what we do (maybe we’re a pitcher or an outfielder, or maybe we’re a joker or an academic), we have a series of statistics that describe who we are physically (age, gender, place of residence), and we have a collection of data that shows what our social interaction may be: which team, or group, are we a member of? What are some interesting statistics that define our role in our society? What are some of the successes we’ve had?

Card avatars have a front and a back. These card avatars are an interface to personality – both private and public. The front being what other people see, the back being what the user writes. When I log into Facebook or Flickr I generally see the backside of my card avatar. If I click on “Profile” I am taken to the back and given the option to change various pieces of text – mostly my stats. When I’m done remaking myself I save that avatar when I save the profile. Then other people see the new front. There’s often times a button I can click on that allows me to see my card avatar as others see it (from the front).
Card avatars are traded. In most systems I am able to trade different kinds of currency with people that indicate we are friends. In some cases these are stars we give each other, or perhaps they are popularity rankings. Comparisons between friends is not uncommon, nor is it uncommon for gifts to be offered. These tradings are generally reciprocal. The process works such that I offer a note to them and if they choose to be friends they offer a note back. As such we are associated on an equal level.1 Users of these systems collect and trade, establishing value and rating themselves and one another. These avatars are usually collected in the “Friends” list of social media systems.
Card avatars are also collected. These collections are called “Groups.” Groups can be organized by an individual, or groups can be self-forming. But the larger the number of avatars in a group, the more valuable that group can become.

Anybody who seriously designs avatars or virtual worlds knows who Richard Bartle is Richard has one.
. Raph Koster has one. Nick Yeehas one.

Raph, in his blog about Richard’s article about the most important people in the virtual world refers to these .. avatars? Something that seems like Richard makes an appearance, at least.

And, after all, The Original Avatar was 2d.

The difference is that these are real-world avatars. The avatars these guys have are their default avatars. These card avatars are more closely linked to us, as real people.

But they serve the same function as our avatars in virtual worlds do: Content, Community, Commerce. The do it with a picture, a profile, a groups list, and a friends list.

Functionally, these are the same.

Why Your Avatar Is Important [Part I].

What do the following have in common?

Business Week is getting twitchy over customer service. The folks over at NASA are using second life to develop open source projects. Barack Obama is using second life to podium-stand. Prim babies are being adopted without background checks and blingtard moms are on the loose. Stroker Serpent is a new-world porn king. (Second) London and (Second) Tokyo were ravaged by (Second) floods. Coldwell Banker, a very large real estate firm, is selling land in SL. And houses for US$20. Long-time SL resident Gwyneth Llewelyn talks a bit about audio.

What do these things have in common? Second Life, for one.

But these things all happened for the same reason Second Life happened.
They all employ what other media can’t: a body.

These things all have Avatars in common. Virtual worlds use a medium that can be visual, auditory, and textual while still including body language. Sure, simple pure-et-dure text is a great interface, and sure, we can use it, but not like bodies.

Virtual worlds such as Second Life continue to improve because of one single idea: The Avatar.

Uncanny Lots-of-Things

Masahiro Mori points out that there’s an emotional response to characters that fall just shy of visually perfect; they are disturbing. He calls this The Uncanny Valley.

Why stop with appearance and motion? Wouldn’t his theory apply to every perceived part of a character? How the character walks, sure (we know this from other research), but also how the character interacts socially, or how the character thinks, or especially emotional responses? Certainly there will be a kind of uncanny valley that will appear in conversation.

After spending the last week at SXSW, in Austin, I had the chance to spend time among the Texans. It’s a strange state with places like the Salt Lick and Emo’s on 6th street offering a range of different character types.

The problem I found is that the uncanny valley isn’t just something that applies to robots.

Emotional User Interfaces

I gave a talk at the Game Developer’s Conference the other day on something I’m calling an “Emotional User Interface.”

The basic argument is this:

Interactive narrative can rely on the emotions of a user to guide the story. Non-Player Characters are especially well suited for this.

It seems an absurdly simple proposition, now that I think about it. The more I take this perspective and look back on it the more I see this as being, perhaps, a defining moment in gaming history; when we will realize how we interface with each other, and apply that to narrative.

The slides are here.

BottomUp & TopDown

Autonomous avatars and NPCs will need both autonomy and dependancy to work properly.

In the AI industry there’s been a bit of a disagreement over whether its better to give a system all the information it needs to do its job or whether its better to let it learn everything from the ground up. BottomUp learning will produce strange minds that live next door, but the TopDown approach gets the short-term job done.

Toshiba and Honda and other car manufacturers use robots and AI to build cars. They had short-term goals (5-10 years) and so the short-term solutions worked. But all the long-term stuff has to do with emergent systems. I’ve been following Rodney Brooks’ MIT-based robotics project, COG, for about eight years, and its nice to see that other robotics specialists are beginning to take similar approaches. For example, here’s “Dexter,” from Anybots Research Group who’s learning to walk. [that poor, poor robot. what an ugly mug, and what a geeky name. his inventors were so cruel!]

But this ‘learning to walk’ work isn’t too interesting to the folks in the car factories. Those car-robots don’t walk, nor do they interact with people. They know all they need to know. They can do all they need to do. They build cars. They don’t run around on stages and act sweet to old people.

But as applications increase, so do design methods. Ultimately, the BottomUp and TopDown methods will grow together. That’s happening already. For example, “Dexter” can’t process his data without some existing top-down information to begin with.

<–more–>

As we develop NPCs and autonomous avatars in online worlds these same considerations need to be made. Giving characters a basic concept (i.e., a greeting, such as “Hi!”) is important, but that concept needs to be integrated with a number of other concepts – and specifically for that individual character – that cannot be foreseen in order for either the character or the concept to have much value (i.e., greeting another person at an appropriate time, such as when they walk in the room, such as “Hi, Paco.”)

We humans are built like this, too. We have instincts, and we learn. And one of our instincts is to learn.

So ultimately, the debate around TopDown and BottomUp will evaporate. The design of autonomous characters – whether they are in online worlds or walking around the factory – will use both autonomous and independant behaviors. Short-term solutions will continue to be pre-programmed (and TopDown) while the long-term solutions will become increasingly learned (and BottomUp).

The trick is for us to now apply these converging design approaches to emotional and social systems in online worlds.

On Using “AI” to Elevate Political Dialogue

Today’s question: How can AI and avatars contribute to political change on a large scale?

Here’s one answer.

First, lets consider an experiment in personality design. Let’s say that you can author a personality. You log into a system and you add a sentence here and a sentence there and each of them have the same tone and tenor. They discuss certain topics with the perspective you type into the system. Is it Ronald Reagan or Ronald McDonald? Is it Marilyn Monroe or Marilyn Manson? It’s easy to determine who says what once you decide on the basic archetype.

Now, second, let’s consider how multiple people could work together on this character. Let’s say that you and a couple of your friends work together to make a Jackie Chan character. You each put in what you think the character should say, and you each alter the responses a bit so that it sounds like that charcter. You all work together and so after a while y’all have a good, solid quantity of work. That Jackie Chan character, like you, is made up of a lot of different contributions from a lot of different people. Your mom, your dad, your friends and lovers. Each of these people made some contribution to you. They each gave you ideas of things that you parrot later on. Now we take Jackie Chan and we have him talk with Ronald McDonald. What would they discuss? Probably the stuff that the authors wrote into the system. And I suspect that the personality authored by a group will have more to say.

Third, let’s consider an experiment in politics. Put ten Canadians in a room and ask them to author a personality. Then take ten Irish, and put them in a different room, and ask them to author a different personality. Iranian and American. Russian and Georgian. North and South. Kurd and Tamil. Rich and Poor. 100 people, 1,000 people, 1,000,000 people – all from the same group. Then have another million people, from a different group, and ask them to do the same.

You end up with a true Hobbesian Leviathan. And then you end up with a second. And then you can make them talk.

What, when we abstract our cultural, racial, financial, and tribal differences, would we have to say to one another? What kind of dialogue could occur? Can a culture have a personality that we can use computers to liquefy, extract, and then mix with another in new and unique ways?

Four Thoughts on Autonomy

Part of the reason I enjoy Second Life as much as I do is because my avatar moves around on his own. He seems to have a few of his own ideas. Like a pet. He’s aware (or appears to be). He’s aware of the itch on his chest (and scratches it). He’s aware of how to walk (all I do is push the button forward). He knows how to land and jump (all I do is push the button up or down). His autonomy gives him credibility.

Spiders, Bots, and Avatars are sometimes distinguished by their autonomy. Search systems, such as spiders, have more autonomy and head out into the hinterlands to look for things on their own, sorting and bringing these items back for your review. Avatars have (traditionally) had no autonomy. They turn right when you tell them to, smile when you poke your macro, and are basically just a body for their user. But WoW, SL, Eve, and most other systems that have some kind of avatar are changing this with small amounts of autonomous behavior.

Avatars will continue to become more autonomous. The more autonomous a system is the more intelligent it seems to be. And we mostly want intelligent avatars we can identify with. We’re only now getting around to inserting autonomy and intelligence in systems that were close-controlled (like avatars in 3D worlds). I call this The Leash of Autonomy.

There are four principles that seem worth thinking about in designing NPCs or avatars.

1. Autonomy indicates intelligence.
   This should be obvious. In case it’s not, consider a stone and a person and ask yourself which is more autonomous. Then consider two people of differing intelligence levels you know, and ask yourself the same question. Autonomy has to do with the ability to evaluate, predict, and act.

2. Autonomy allows cooperation.
   Its impractical that I tell my character where to put his foot, when to move his head, etc. I say “forward” and its up to him to walk. This is the steed concept – he knows what to do, I pull the reigns. The whole idea of the avatar is that he copes with the tactics as I cope with the strategy.

3. Deep autonomy allows a relationship.
   In talking with bots, or conversational systems, it’s their autonomy that allows them to seem independant. It is their ability to throw the curveball that makes them seem “life like” and it is their independance that gives them a degree of independance. This allows me to develop preferences and dislikes, and for an interactive system to develop between us.

4. Autonomy is the basis of morality.
   I’ll save this for the book.

In the coming years avatar autonomy will be as important as virtual world persistence. It is what allows us to believe they are real.

A Talk in the PARC (with Nic and Bob)

The other day I was at Xerox-Fujitsu’s Palo Alto Research Labs (FXPAL) for a guest lecture (my slides are here). Nicolas Ducheneaut (who works at PARC and helps maintain the Terra Nova blog) also presented some of his work.

Nic’s a precise and pleasant fellow and, like many of his French countrymen, quick to understand virtual worlds and non-verbal texts. He’s been looking at the user statistics of World of Warcraft, and running a /who command that helps him determine what class, race, level, guild, etc is online at the time (his work is posted here).

Meanwhile, Nic introduced me to Bob Moore, and Bob (also a PARC researcher) is working with Nic, doing research on conversational analysis.

And Bob’s work, or at least the method for some of his work, is here.

As far as I can tell they’re building models for large-scale forms of conversation. Interesting stuff that will certainly be of use in the coming years (just ask any politician, business manager, world builder, or advertising director).

Good luck, gentlemen.

A World in Every Pot, A ‘Tar in Every Garage

An open source Second Life server was released today.

I think this is great. First it means that this thing is about to really catch fire. It means that we are seeing the beginning of something like a 3D blog appear (3Dlogs? ugh.) … I think we’ll have lots of these spaces. Everyone can have their own world and avatars will have to be able to move from one world to another, and we’ll have to have some way to keep track of who’s who.

And it will not be a browser war as much as it is going to be an interface war: Pages or Worlds. Fortunately they’re compatible. For a while this basic interface will stick… the interface of the 2nd-person camera in a 3d world of random symbols. What this also means is that the lack of metaphor will probably remain prevalent… what I like to think of as the “Dragon With Sunglasses” metaphor, the NONmetaphor, such as Second Life’s, where Furries battle Goreans for blackmarket PsiTech weapons over BladeRunner oceans, dressed in 8th century codpieces. Then, in about 6 years, when the framerates and bandwidth get to what current consoles look like we’ll learn to integrate the interfaces.

Webpages will appear in these worlds (so you read a book in-world) and more interestingly vice versa (where you visit an interactive world in a comic book frame). Storytelling will get insane and potentially beautiful.

But the basic page &/or world interface won’t go away.
And there won’t be any real metaphor to help guide us.

We’ll just have a lot and a lot and a lot of 3d worlds.