Brian is off in the San Francisco bay area working for Archimedes Inc. helping them revolutionize healthcare for everyone on the planet. Check it out. We have a mathematical model of human physiology and disease processes that will help doctors, HMOs, pharmaceutical companies, governments, and NGOs make better and more cost-effective decisions about treatment.
He was a graduate student in the EECS department and also was a student at Switzerland (Project MAC).
The value of many software patents to society is debatable. However, my day jobs use them in a defensive manner in the meanwhile.
Basically the Schmuse is an attempt to create a MUSE (Multi User Simulation Environment) using Scheme, and to fix some of the traditional problems with MUSEs. One thing is particular that we noticed is that the scripting languages in most MUSEs were not well designed and lacked a lot of the functionality of a real programming language. So for this system we gave the users access to the full power of Scheme for their scripts.
The project quickly became unwieldy in terms of the amount of code and the amount of labor we have put into it. If the project is to continue we need a specific goal in mind for turning out an end product, rather than building the all around perfect environment. The foundations of the design remain sound in my opinion however.
We have made a couple of attempts to publish a paper on the SchMUSE's revolutionary object system and other features, but have yet to get a paper accepted. So unfortunately I cannot provide a publicly accessible draft of what we have written. Although, I assume there will be a forthcoming AI Tech Memo detailing the innards of the Schmuse.
MIT has been generating data on digital media in significant quantities since the 1960's. The rate at which new bits are produced is steadily increasing, and the knowledge needed to decipher the old bits is vanishing. All of our older media are beginning to deteriorate, and valuable data is being lost every day. The Time Capsule File System (TCFS) project was developed to address the problem of preserving this information, for use now and in the future.
Specifically, TCFS is intended to be a single universal format to which we can migrate all our old files, thus simplifying the problem of dealing with a myriad specialized formats. TCFS is also designed to be simple enough to re-engineer without any previous understanding of it, which reduces the risk that data will become ``stranded''.
Using TCFS, Dr. Alan Bawden and I developed a framework using TCFS to preserve our archives in a durable and easy to use fashion. Pandora Berman and I showed this framework to be viable by using it to rescue data from backup tapes written on the Incompatible Timeshare System (ITS). To this end, I developed specialized tools to translate the data from these tapes into the more general TCFS format. Simultaneously, I developed other tools to categorize, index, and search the large set of TCFS archives that were created.
This mini-project takes all this public data and manipulates it so that you can produce nifty graphs showing the ratings of various courses over the years.