[om-list] 4C-MTShell Plans, 2001.11

[Tom and other Packers]

Hello 4C-associated people

    Here are the news and plans, as of 2001.11.  I hope to hear from anyone
who has the slightest hint of motivation in helping me, including Mark
Butler, Dougy Adams, Ben Oman, etc.

    I expect this effort to overlap OM's plans and goals considerably, so
I'd like input from them as well, including Lee Howard, Luke Call, etc.  I
also know that this effort will overlap Thoughtform's plans and goals, even
if Thoughtform doesn't see it now.  So I will update Dan with the
development from time to time.

--  News:

    I have registered three domain names, all of which will point to one
place on
Mark Butler's machine:

    www.4C.tech (which must be accessed through www.4c.tech.new.net
    www.CompCog.Com
    www.ComputationalCognition.Com

    The full name of the organisation is being debated: should it be
"Computational and Cognitive Consilience Consortium" ("4C" for short), or
should I just use the domain name: "ComputationalCognition.Com".  I am open
to suggestions.  I take the term "computational cognition" from a major
sub-discipline within cognitive science, which I am quite fond of.

    From now on, I will put subsequent additions to the information in this
letter (e.g. news and plans) on the webpage above, and will probably only
use email to tell those who have expressed an interested when there are new
additions or editions.

--  Plans:

    I want to develop a complicated AI-type system, with applications to any
field that has heretofore benefited by human expert knowledge, human skill,
human problem-solving, human logic and reasoning, etc.  In other words, I
will eventually target any arbitrary collection of fields, using AI, or as I
call it, computational cognition.  Examples are numerous (innumerable).

    For now, I will start small, developing one small package of fairly
generic data analysis tools, called Informatica A.  It will be composed of a
collection of small Unix-like utilities, and a
utility-executing-and-coordinating shell/interface called MTShell.  The
shell will begin as a simple command-line interface, like a shell on Unix,
which can execute commands as you type them in, or which can execute a large
collection of commands recorded in an interpretable text (script) file.  The
language will be called MTL (mathetically traversing language).  In the
future, I would like to develop a compiler for the MTL code.

    Most of the utilities will be written under a licence similar to MozPL,
and as such will be "open source".  I will freely accept contribution by
programmers, in whatever language can support the basic design (of using
pipes and command-line parameters, etc.).  In fact, I will be asking many of
you if you would not help writing individual utilities, of the appropriate
structure and interface specifications, which I will put on the web page as
soon as possible.  Some of the other utilities will be written by me
specifically to interface a proprietary software package (TFStudio), and
will not be open-source.  The MozPL-style licence will have the effect of
making it possible to mix open-sourced source files with non-open files.  I
am planning that the utilities which start out MozPLed will stay that way
forever.

    There will be additional files (a library) shared by the utilities.
This library is meant to abstract out the commonalities of the utilities'
interfaces and OS-specific code, and to facilitate porting to other
operating systems -- other than WindowsXP, which, by the way, is the OS I
intend on starting out on.  These library files may be MozPLed forever; I'm
not sure.  It may be considered part of the shell, described next.

    Like the majority of the utilities, the shell will be MozPLed.  Unlike
the utilities, I will not accept any contributions in source code to this
part of the project unless I am given the copy right, because I plan on
converting these files into a non-open-source (proprietary) software package
some day, maybe from 5 to 10 years from now.  It either will be called
"Informatica" at that point, or perhaps "Informatica B".  (Perhaps I should
call them "Informatica Lite", and "Informatica Pro".  Any suggestions?)

    I will probably allow all of the original MTShell files to continue
being freely available and modifiable.  The nature of the shift to
proprietary-ness will be limited to my using derivatives of these files in a
proprietary package.  For what it's worth, I promise to not make this shift,
(i.e. to not stop promoting and supporting and building the open-source
version of MTShell) until after MTShell reaches a "sufficiently useful"
stage of development.  I will probably take some sort of vote from all
invested individuals as to the timing of my shifting from open MTShell to
closed Informatica.  And even after the shift, I may still support the
divergent development of the open-source MTShell, but it will not have as
many features as Informatica.

    4C-Informatica will be proprietary, and Thoughtform Corporation and I
may
have exclusive use, copy, and licensing rights, if TF chooses to accept
them.  I will have to discuss with Dan what will happen to those few
proprietary utilities which are made to interface TFStudio specifically.
But I will probably push to make all "previous" versions of Informatica free
versions, such that any version that is not the most current version can be
freely copied.  If TF doesn't like this, I might try a compromise, such that
only the two latest versions are not freely available.  Whatever we do, I
don't believe that the source code of any version of Informatica will be
open and free after I make the "shift", but the fact that the original
utility files will still be MozPLed, and the fact that the original MTShell
will always be "out there" and available for free (even after I make a
proprietary derivative), will have some value in this regard.

    MTL Plans:

    I plan on implementing many, if not all, of the "mathesis" ideas I have
developed (in my spare time) over the past few years, in the form of a
programming language.  Any one who wants to understand those ideas
thoroughly can read the three essays on this subject (once they are
finished).

    Briefly, Mathesis is intended to become very generalised, AI-centred
construct, and a superset of both mathematics and logic, both inductive and
deductive inference tools.  As such, it will be both a language and a
method, both static and dynamic, both quantitative and qualitative, etc. and
some day, it will be both symbolic and geometric.  As such, I have some
weird ideas I should bounce off of Mark some time, to make sure I'm not
dreaming impossible dreams as I try to write an interpreter for this
language.

    That is the long-term goal.  The short-term goal is MTL (or S-MTL,
symbolic mathetically traversing language), which is a drastically
pruned-down, watered-down version of Mathesis, with not much more than
simple syntax for piping processes together and running them in loops.
(There my not even be any conditional-syntax to start with.)

    I may have to take a compiler-writing class if I'm going to write this
language myself.  I'll have to see.  Mark, et alia, do you know of any good
compiler- or interpreter-writing books?

ciao,
tomp