Welcome to Forum 2000

``When you're just too busy surfing to solve your own problems'' TM

Are you troubled by a vexing moral issue? Submit your problem to Forum 2000 and let our Digital Philosophers ponder your problem for you! You'll receive a variety of advice from (simulations of) the world's greatest thinkers. It's easy, it's free, and best of all, it's not blocked by CyberWatch!

You may pose your own problem, view recent questions and their replies, or experience the Forum 2000 Hall of Fame!

New! Visit the Forum 2000 Scrapbook!

Though any member of the Forum may choose to answer your question, please select the Philosopher you would like to be your primary advocate on this most difficult issue.

Today's featured panelists:

The Anti-Andrej
The Cube

A message from the President of Intelleq, Inc.

George Santayana said, "Those who cannot remember the past are condemned to repeat it". This is just as true in computer science research as in politics, as we at Intelleq have learned the hard way in these past few weeks.

Recently our engineers completed the changeover to Testbed 2000, our new network of workstations hosting the new SOMAD-5 computational framework. Soon after deployment, we began to receive numerous customer complaints regarding "Internal Server Errors" from our web site.

Although thorough testing proved that the SOMADs were functioning flawlessly, the matrix core itself was aborting abnormally on every single user question! So why did "short" questions appear (to the end user) to continue to work?

The answer has to do with our redundant backup system, which we whimisically call Forum Jr. This system is intended to maintain internal SOMAD consistency in the event of a failure of the main core. However, it has only a tiny fraction of the computational power of Testbed 2000, so it cannot handle any but the most trivial questions.

So what caused all these problems in the first place? Using debugging techniques derived from the ISO/ASA-9001 (DIN 32) capability maturity model, our engineers isolated the problem to the SOMAD transfer phase of the project. Happily for the CMM (and unhappily for us), this was the only point in the process where we deviated from the CMM.

Specifically, in order to save time, we used a standard FTP program (rather than SEI-certified data interchange protocols) to transfer our SOMAD data from the previous Level 1 testbed to the current Level 5 testbed. Unfortunately, one of our engineers made the common error committed by so many FTP "newbies": setting the transfer mode to ASCII rather than binary. When inadvertantly used on binary data, this causes the FTP program to alter occasional bytes in the binary stream, specifically those that it misinterprets as "carriage returns".

Now that we have isolated the problem, we have retransferred and reinitialized the SOMAD data. Forum 2000 should once again have the robustness that our customers have come to rely on.

We appreciate your patience during this rocky period of transition.

Kofi Bartok, President & CEO
Intelleq, Inc.

And what is it you wish to discuss? Please give us the whole terrible story. Omit nothing.

Please compose your question carefully, including as many details as possible. Longer, more detailed questions are required by the neural net's quadratic search algorithm. Simplistic "one-liners" merely clog the matrix and will lead to (seemingly) meaningless advice.

Disclaimer: Forum 2000 is for entertainment purposes only. The material presented here does not necessarily represent the views of the persons simulated or of any other party. Forum 2000 is not endorsed by the simulated persons or their creators. Do not follow the advice of Forum 2000 under any circumstances.

About Forum 2000:

This Web page is a front end to a sophisticated expert system incorporating the latest breakthroughs in natural language and neural network research. The system is composed of about 11 million lines of C++ and Guile code, running on a "farm" of heterogeneous machines interconnected via a prototype 9.6 Gigabit ATM network developed by IntraLink Solutions. The underlying quadratic neural network has been training on Usenet, Web, and literature sources for over three years; on July 11, 1998, the neural matrix surpassed 32 Terabytes in size!

Although previous versions of the system showed promising results, they required excessive computational effort and produced poor solutions for many common cases. However, in 1994 Andrej Bauer made several theoretical breakthroughs in the synthesis of Lambek calculi and connectionist search algorithms. Implementing his ideas increased the performance of the system and the quality of the solutions by many orders of magnitude. He has written a brief treatment of the work, suitable for the lay person.

The Forum 2000 project is led by Darrell Kindred and Corey Kosak of Carnegie Mellon University. Funding for the project is provided by Intelleq, an industrial consortium consisting of IntraLink Solutions, Linguistic Technologies, and Quantor UK, Ltd.

Please send us your questions or comments about this service.

Forum 2000