Just to show that I’m big enough to admit when I may have been wrong (the jury’s still out, but it’s not sounding good), someone sent me this link to an interesting paper published in a recent issue of the IEEE Computer Graphics & Applications journal on a project called ‘Astrosim’.

Specifically, our AstroSim system supports important activities and features for research and education, such as data zooming, playing back stellar simulations, and color-coding properties or stars. AstroSim’s salient collaborative features include the ability to manipulate stars’ visual properties for annotation and to point and refer to stars.

Nothing too revolutionary, SecondLife already offers text chat, voice chat, and environmental features, so they just needed to add the stars, physics, and playback controls.  They admit encountering some of the same constraints I mentioned:

We tested our application with a simulation involving 1,024 stars. Although this number is impressive for this type of application, the number of objects is much less than normally expected in a realistic astrophysics simulation. In addition, Second Life restricts the total number of objects, or prims, that users can create per island to about 15,000. So, although AstroSim has no hard restrictions in the number of processed objects, this variable constrained us.

As well as other problems such as the dependence on libOpenMetaverse and the XML-based communications, they seem quite positive about the whole project. From a presentation of the project to astronomers from MICA not considered experts in Stellar Dynamics (the focus of the particular simulation under scrutiny):

All the astronomers were extremely pleased to see the simulation visualization displaying smoothly. Some of them said they’ve tried to create the same settings using Second Life’s internal scripting methods and had somewhat disappointing results owing to the stars’ jerkiness and the overload of computing resources on the Second Life servers. From this conversation that the astronomers had among themselves and the opinions we collected after the session, we conclude that AstroSim was a useful tool for serious research analysis.

You can get more details on their site, or in the journal article.