Set Your Avatars Free!
Avatars have always been fascinating to me, and I think perhaps it all started back when I first started creating make-believe characters for made-up stories I used to dream up when I was a child. Of course, I didn’t call them Avatars at the time, but essentially that’s what they were. In fact, a few of my made-up characters were so compelling to me, that I would pull them out of my own made-up universes and stick them into universes created by other people, (for instance, I had this great character called Terces Tamoc that I often mentally re-purposed and shoved into universes like Star Wars, Blade Runner and other great “sci-fi” worlds.”
From what I’ve read so far, it appears that much of this effort is still in it’s very infancy, and, I may be completely wrong here, but the methods that I’ve seen so far for how this data sharing problem is being tackled seems to be centered around traditional point-to-point protocol and data translation between the different worlds (i.e. take an Avatar data object in Second Life, transform it into an Avatar data object in OpenSim, and now you have created a duplicate avatar in OpenSim, effectively moving from one world to the other). In addition to this, I believe there is a movement to “centralize” Avatar data as a service that would allow a virtual world to look up the Avatar information much as you would a LDAP entry.
While, I think these efforts are a logical progression from the business and SOA mind-set, I think perhaps if we take a step back, and take a look at the “ideal” representation of an Avatar, we can start to try and tackle this issue in a more abstract and ultimately more widely applicable method.
To me, an Avatar is a facade that represents me to the virtual world (whichever that world I happen to be in). The Avatar encapsulates everything that is unique about my character. My characters “physical” appearance, what clothes he is wearing, his name, and generally anything else we might consider to be “public.” Additionally, there are aspects of an Avatar that make it functional (generally, the reason we move between worlds is either to play games, explore, or more importantly to buy things). Just like in the real world, I wouldn’t generally carry around cash in my pockets, but I would probably carry an ATM card, or a credit card, and an avatar needs to have a private, secure set of core data that can only be accessed by the original Avatar owner, not even if it is copied.
Obviously, the very security measures incorporated into an Avatar definition is what will ultimately make that definition succeed or fail in general (if you can’t have confidence that anyone who copies your avatar can’t steel your identity with it, then what’s the point of even creating such a thing?). 
One way this might be approached would be to have a two-part system. Essentially, what you would have a piece of software on your local computer that is you would use to create your avatar with, in a standardized “Avatar Definition Format,” (it should be sophisticated enough to allow you to graphically design your avatar so it meets your aesthetic needs), and this application also acts as a security vault for financial information, providing a high level of encryption around this core data. The application also allows you to fully define all the common “public” elements described earlier, or even provides a way to randomly generate your avatar (within a set of aesthetically pleasing constraints). The public half of your avatar would be loaded into whatever Virtual World Browser you are using (i.e. Hippo for instance), via a plug-in component that allows the browser to interact with the Avatar Management Application.
When you move from world to world, the virtual world you are in is free to cache the public aspects of the Avatar, but the private data is still local. If a browser connects directly to a world for the first time, the public data for the avatar is uploaded, however if a copy of the avatar exists in the world he is trying to brows to already, the browser needs to compare avatar versions and bring to the users attention and allow him to choose what to do (overwrite, keep server copy, merge, etc.) When a world-to-world transfer is happening, what is sent between the two worlds is an encapsulated piece of data that represents your public Avatar with special unique data that only has meaning to your local browser… this way, if another person copies your avatar, it could be possible for the virtual world (through the specification) to identify an unauthentic use of an avatar and remove it from the system and alert the owner.
When a situation arises where a purchase is necessary (for instance, you want to buy an item in-world, or a fee is required to enter a certain virtual world), then there should be a secure connection created between the browser and the requesting in-world component, much like internet commerce has been done for the past decade, and some visual indication of a secure connection is given to the user. Since the data required to purchase the item is stored within a secured vault in the avatar definition, it would require confirmation and authorization. Additionally, there is nothing preventing the Avatar from being configured to direct payment from a third-party such as PayPal and so on…).
Public Data in the avatar is your “Default Model” that stays with you from world to world, and the particular world you are in can choose to use that default model or give you the choice to attach a local model, overriding certain aspects of the Default Model and inheriting others. No matter what however, the unique link back to the private data is what ultimately ties your server account to your avatar. In this fashion, you could move from a business world where your Default Model is used into a MMOG where an alternate model is attached to your avatar while you are there.
What about inventories? To be honest, I think inventories should be for the most part local to the virtual world (in other words, you can’t take them with you). However, I could foresee the creation of Avatar Attachments that might add capabilities to an avatar (for instance, Grid administrators may need certain tools attached to their avatar for monitoring, or muti-world moderators may need special tools, possibly even custom developed, for their needs, or perhaps for MMOGs, they may allow modding, or the creation of plug-ins that enhance game play, or require mandatory plug-ins to support how they want their world and the avatars within it to function and interact. Obviously, some of these plug-ins are world local, but others could be private, purchased from third-party sources).
I think this model for Avatars allows for a great deal of flexibility, and starts to move us into a model that is closer to how I always perceived Avatars to behave and function. Avatars need to conform to a standardized format and structure; this is the only way to make this sort of thing a reality. Without a standards body taking up the mantle of owner for the definition of an “Avatar Definition Language,” or “Avatar Definition Format,” etc. then none of this will make any difference in the long run.
