MPK20
I went to visit the Collaborative Environments group at Sun Labs in Burlington to learn more about the MPK20 project. Nicole Yankelovich and her team gave us a nice demo and told us about their motivations and future plans. If I heard correctly, more than 50% of SUN’s workforce is ‘remote’, and one of the major complaints is the quality of remote collaboration tools.
MPK20 is an attempt to improve remote interactions, both to improve efficiency, but also to bring back some of the social aspects of working in the same location that remote workers currently miss. Built-in from the start is voice support and application sharing, and though the project is still at an early stage, the development team thinks they are very close to being able to use it in day-to-day meetings.
I tried downloading and running MPK20, which was remarkably easy - as long as you have ant and a recent jdk, you can check out the source yourself, build and run the system in 2 steps. The system downloads the environment from Sun’s servers, which didn’t work through our firewall, but on my home machine everything came up smoothly.
The virtual world you get by default is limited to one building (MPK20 refers to building 20 at the Sun’s Menlo Park campus. There are 19 physical buildings, and building 20 is the virtual one). When you enter the environment, you’re in the big hall shown above. There are a number of “non-player characters” standing around having conversations or doing demos. Unfortunately, the voice-bridge is not yet available for download, but what happens is that as you move closer to different groups of people, you start hearing their conversations (similar to Second Life’s voice support).
![]()
There is currently nothing indicating who is speaking, and though the NPCs in the environment are gesturing, there does not seem to be any way to animate your own avatar. The avatars themselves (and the whole building) are rather primitive looking, though I’m pretty sure that it’s because the development team is focusing on features rather than graphics (since everything is built on Java3D, better graphics should be possible).

This image shows my avatar (live avatars are labeled with their names, NPCs are not), in the “team room”. This area has information all along the walls, and the intent is to have automatic updates of various artifacts, and perhaps even have whiteboards that can be updated in both the real and the virtual world. All around the team area are smaller rooms, that could be specific to individuals, for example. As in a physical office, a closed or open door can be used to indicate whether it’s ok to disturb the occupant.
There certainly is a lot of demand for a business oriented virtual world platform. Second Life and other available platforms have many problems, either in terms of security or in cost of content creation. MPk20 is open source (the voice bridge will be released soon) , Java based, includes application sharing and claims to be scalable because of the underlying Project Darkstar architecture. It will be interesting to see the future development of this project, especially if an open-source development community forms around it. So, if you’re a Java hacker looking for something to do, helping out the MPK20 team could be an option. If you want to make me happy, start working on some of the below:
- Easy content creation. Right now you have to use Blender or something similar to create 3D objects. The ability to import from Sketchup is supposedly being developed, and in-world building tools are also being considered.
- Better avatars. Better looking, more customizable, and more controllable avatars are needed before it becomes possible to feel comfortable with interacting with others using this environment.
- Better camera controls. There are currently only a few camera angles possible. If I’m looking at an application window in-world, I’m going to need better ability to zoom-in and to quickly look at other information being displayed around the room.
- Windows Application sharing. Currently being able to display and interact with desktop apps in the environment is only possible if the applications are X11 based.
Perhaps we should start a list of the minimal feature-set needed for virtual worlds?

I like the idea of developing a feature set. I think all the ideas above are quite good, and I’d tend to identify a couple of areas.
First is identity and portability of social networks. There has been a good bit of back and forth lately on open social networks, such as in wired (http://www.wired.com/software/webservices/news/2007/08/open_social_net?currentPage=1)
This makes me wonder about bringing real identities, which could cut both ways. At any rate, there are things that make the 2D web successful, and those elements should be carried forward.
Another area that interests me is around SL/RL mash-ups, and involves treating a sim as a presentation layer on top of an application server. I’d love to be able to grab the ‘context’ of a virtual world and be able to discover and touch objects and people in a sim, as well as subscribe to events happening in the world (like a 3D swing application). This would include the requisite hooks for security.
Anyhow, those are some big areas where I’m looking for features. The whole L$ thing seems significant too, essentially a micro-payment system, which is a big deal in the mobile industry right now. It allows an economy to flourish, and provides an easy way for non-builders to put together compelling content.
very interesting, but I don’t agree with you
Idetrorce