User input of "you" is not properly activating the "I" concept #50
in the psi conceptual array. The JavaScript artificial intelligence
(JSAI) does not have this problem, so MindForth must be doing
something wrong.

Actually, the only problem is that the input of "you" followed by
a carriage return (CR) somehow keeps MindForth from recognizing
the word "you" and activating the "I" concept. We need to
troubleshoot the issue and perhaps turn a carriage return into
the entry of a space-bar.

Meanwhile, we have changed a variety of things. In Tutorial mode,
the AI was not showing the contents of auditory memory during the
Rejuvenate process. We discovered that in bypassing SVO and in
using Predicate instead to generate a sentence, an SVO flag was
not available to display the Rejuvenate process. We solved the
problem by moving the flag into Predicate.

We also discovered that the AI was not detecting duplicate
thoughts, because it had been the job of SVO to do so. We are
thinking of using the system for the detection of duplicate
thoughts to implement KB-traversal via the Rejuvenate module,
because the oldest (post-vault) concept is oftentimes changed
by the backwards sweep of the Rejuvenate process. It also
occurred to us that we could use the Conjoin module to
consolidate the knowledge-base (KB) by detecting multiple
objects of the same subject-verb combination and joining
the direct objects together with the "AND" conjunction.
We were thinking that we could do KB-traversal and
KB consolidation simultaneously. But when we tried to
compose a list of entelechy goals for the Conjoin module,
we discovered that Conjoin had been missing from MindForth
since 17 August 2002 -- six years ago. Apparently, in August
of 2002, or perhaps later, in 2005, we accidentally loaded up
and worked on the 3aug02A.F version of MindForth instead of
the 17aug02A.F version -- thus losing some days of work in
which we had ported the JavaScript Conjoin module into Forth.
Today we have put the 16aug2002 Conjoin module as-is back into
MindForth so that we may soon start to update the module.

MindForth needs knowledge-base (KB) traversal to
demonstrate multiple lines of thought and thus make
the AI interesting for human users who interact with
the AI.

There are several ways to achieve KB-traversal (KBT).
We should design KBT as if the AI had unlimited memory,
and also as if the Rejuvenate module were going to
truncate the oldest memories in a cyclical fashion.

We do not want erroneous user input to be reproduced
over and over again by a KB-traversal algorithm.
Instead, we would prefer for concepts and ideas in
the enBoot sequence to be reactivated. In fact,
the kbTraversal module could contain a rotating
inception-point so that a different idea would be broached
each time that the AI went into KB-traversal. Ideally,
each reactivated enBoot idea would spawn unpredictable
chains of thought.

There could be a "kbtv" variable that would cycle
in value from one up to, say, seven. A test for a
kbtv value over seven would reset kbtv down to one.
For each of the seven cases of kbtv, the KBT module
would send a particular recall-vector into the
enBoot auditory memory and activate either a single
concept or a whole idea. In that way, there would be
some variety in the visible thinking of the AI
during KB-traversal, and as an actually added benefit,
users would begin to recognize ideas or concepts
resurrected from the start-up sequence. Such ideas
could even be of a didactic nature or in the nature
of an advertisement.

We could make the kbtv variable serve dual purposes.
At the outset of running the program, kbtv could be
set to a value that causes the AI to say,
"I AM ANDRU"
"WHO ARE YOU"
Such behavior might prompt input from a human user.

We would use each Rejuvenate cycle that had no user
input as a trigger to increment the kbtv variable.
In that way, we would bypass the question of whether
or not there are duplicate thoughts occurring --
although we already have elaborate code for detecting
duplicate thoughts.

We can safely assume that, if no user input occurs
during one Rejuvenate cycle, then the AI is either
following a meandering chain of thought or has lapsed
into a repetitious thinking of a monotonous thought.
Either way, it seems okay to resurrect one idea from
the enBoot sequence -- especially if the resurrected
ideas vary from one inputless Rejuvenate cycle to
the next.

This design for KB-traversal has a very important
bearing on the likelihood of people competing to have
the oldest or the longest-running AI Mind. If the AI
does not perform any interesting behavior, there is
no perceived value in having the AI run forever.
However, if the AI sits there and changes its line
of thinking, even among ideas contained in the enBoot,
then not only are the thoughts themselves interesting,
but also the build-up of various indicators on-screen,
such as the claimed IQ of the Mind, the number of cycles
of rejuvenation, and the obvious age of the AI entity.

Now we have introduced the kbtv variable and into the
user-input phase of the Audition module we have inserted
the following comments and code.

\ 3sep2008 The KB-traversal trigger "kbtv" starts out with
\ a value of one so that the incipient AI will say something.
\ Any user input at all between Rejuvenate sweeps will
\ prevent kbtv from triggering KB-traversal. If there is a
\ prolonged dearth of user input, kbtv will increment and
\ reactivate a different enBoot concept after Rejuvenate.
0 kbtv !  \  3sep2008 User input obviates need for KBT.

Next we create a kbTraversal module sandwiched in between
enBoot and Rejuvenate. We borrow some code from the
EGO-resuscitation module to give KBT its marching orders.

After considerable experimentation, we get the kbTraversal module
to cycle through the activation of four different concepts from
the English bootstrap enBoot sequence -- a different concept
each time after a mind-sweep by the Rejuvenate module. As a thought
for each concept comes up on the screen, human users have the
opportunity to engage in dialog with the AI Mind about whatever
the AI is thinking about, or about whatever the user wants to say.
For the first time ever, the AI Forthmind exhibits a wide variety
of thoughts and does not simply repeat the same chain of thought
endlessly. If we lengthen the English bootstrap to include more
concepts, we have greater latitude in knowledge-base traversal.
We upload the AI to let it propagate across cyberspace and the
noosphere.

For discussion of MindForth, visit the Usenet newsgroups
comp.lang.forth
comp.robotics.misc

For discussion of Win32Forth, visit the Yahoo win32forth group
http://groups.yahoo.com/group/win32forth/messages