From: Trans <transfire@...>
Date: 2006-04-02T12:43:44+09:00
Subject: [ANN] Lirpa 1.0

Introducing Lirpa 1.0!

Lirpa 1.0 is an advanced lie detection system based on a heuristic
reverse engineering engine. It can take any set of data points and find
the underlying functional bell-curve system(s) within, or contrarily,
detemine if the suppiled figures constitute a misprespentation of the
facts --hence a lie.

Curiously, Lirpa may provide a direct test of Godel's Theorem of
Unprovability. When Lirpa 1.0 was fed itself as a parameter:

  Lirpa.eval(Lirpa)

We expected it either return Lirpa, i.e. identity, or never to return a
result, running up against the Halting Problem. Any result would
disprove Godel's theorm, so we certainly expected the later. Oddly
enough, there is no way to determine how long this computation would
take. Nontheless, being the forever optimistists, we hunked down with a
fresh brew of coffee and waited as our  druby cluster churned. To our
surprise Lirpa 1.0 returned a result almost immediately. After further
testing we varified the results. But they did not come out as we had
expected. On subsequent runs the results actually VARIED.  We had
somehow managed to create the first true computational random
generator. That's a pretty good feat, but what's really amazing is that
we could not have done it with out Ruby's dynamic eigenclasses!

For fun, we tied the system into a little web application that
partitions the complex results into a simple 0-7 scale and assigns each
value a user friendly "magic 8 ball" natural language representation.
You can try it out here:


http://www.degraeve.com/cgi-bin/magiceightball.cgi?q=Lirpa.eval%28Lirpa%29

On the whole Lirpa 1.0 seems, well, pessimistic. I guess it's just
clarified an unfortunate fact about the world: most of it's full of
sh*t.

T.

P.S. If you feel brave, try feeding Lirpa 1.0 this email message and
see what it says.