The Minsky Project

Table of Contents

1 Project to make theses :

Ship with live code
In general, coding projects should include links to up-to-date source code.
Freely available
Public research should be available to everyone. In particular, coding projects should be distributed under a free licence. Others should be encouraged to release their own modified/improved versions of the code.
With live didactic demos
For especially interesting projects and projects that initiate new ideas, we want to go beyond providing updated source code; we want to provide didactic demos which illustrate the new idea to a wide audience. Our principles include literate code and assuming the audience is intelligent.
Rekindled
One of our first tasks is to resuscitate some of groundbreaking AI theses so that more work can be done on them.

1.1 Refurbishments

(and my thoughts about them)

Slagle's integrator
Good — perhaps algebra or even PEMDAS first. (use as a library)
Winston's arch-learning thesis
Too practical — could necessitate initial investment into visual parsing without much immediate payoff. Perhaps cheat using WebGL.
Sussman's HACKER
Perfect — displays near-miss idea, could be applied to blocks world or any other domain. Crypto? Others?
Bobrow's STUDENT program
The ability to solve word problems might be a good precursor or library for an extended version of Slagle's, perhaps. (A Slagle program that can solve integration story problems?)
A chemistry expert
(Answer questions about the results of certain mixtures.)
Other mathematical experts
  • Mini experts that solve problems such as factoring in a talkative, explicative way. It would be especially neat to have programs that can solve story problems or problems that require guessing, because these appear to be less methodical.

2 Project to teach schoolchildren :

  • Fluency in thinking about programmatic processes and bugs.
  • Fluency in thinking of their cognitive processes as programs.
  • In all subjects, using what we've learned from AI about teaching.
  • Whatever they want to learn, at their own pace, through intelligent software. (e.g. algebra/calculus for anyone who can think about and read programs.)
  • About the accomplishments of AI. (Evans, Slagle, and many others)
  • About how they might want to study AI for a living.

3 Project to make Marvin's work and ideas

  • Easily accessible to a wide audience. (SoM, Finite/Infinite, Wheel book)
  • Easily navigable (the biblio project)
  • A part of a grand research endeavor (The Minsky Institute, support for individual theses)

Author: Dylan Holmes <dxh@mit.edu>

Date: 2013-09-27 16:10:59 EDT

HTML generated by org-mode 6.33x in emacs 23