CODING FUNCTIONALITY BEYOND THE 28.MAY.1998 LEVEL
As of Tues.24.Nov.1998, the "AI in 30K" Mind.forth has been keyed
in so as to reconstitute the functionality of the 90-screen Mind.-
forth from Thurs.28.May.1998 and any changes made in June of 1998
have simply been abandoned, for the time being.
The 30K Mind.forth is the most "diagnostic" to date. Within each
diagnostic message, it shows the actual screen-number of the sub-
routine which is reporting a status. The explicit identification
of each screen enables the Forth programmer to call up any prob-
lematical screen in order to attack any problem as soon as it be-
comes evident. Within the tight, carefully coded, 30K Mind.forth,
the screen-numbers are neither likely to change because of "drift,"
nor difficult to change when there are only thirty screen numbers
to worry about.
In order to advance the state of the art in Mind.forth beyond its
May or June of 1998 level, we now run the program in order to as-
certain the chief obvious glitch ("COG"). Right now the COG is
that EXAMINE SCR #17 is reporting garbage instead of clarity for
the content of the "lexicon{" array in which NEWCONCEPT SCR #9 is
supposed to be storing each new concept and assigning to it a num-
ber "n" as a consecutive concept number. Some of the data are a
row of zeroes, others are obviously RAM address values in Forth,
and others are inexplicably the positive number "191" or the neg-
ative number "-192."
In the grand tradition of Linus Torvalds as analyzed by Eric S.
Raymond in http://www.tuxedo.org/~esr/writings/cathedral-bazaar/
"The Cathedral and the Bazaar," let us try to "harness the collec-
tive IQ" by reporting our state of the art on Usenet.
--
http://www.geocities.com/Athens/Agora/7256/ "The Cyborg Syllabus"
http://www.scn.org/~mentifex/ The Mindmaker Project AI for Robots