Posted by Muncie on September 25, 1999 at 18:35:04:

In Reply to: UTK -- A Challange posted by Reilly Atkinson on September 24, 1999 at 17:34:09:

UTK – Some Considerations

First, I would like to acknowledge what an interesting, generous and forgiving forum this is. All the contributors are to be commended on the space created here.
I have been following the forum for about a week now and have gone back and read quite a few of the previous postings. I hope that I have done enough homework to be able to add to this dialog.

The initial part of this is in the spirit of Gregory Bateson’s “metalog”. The metalog is a dialog about a dialog. Bateson used this term to describe a way to aid a dialog when it appears to become stuck or less productive. The idea is to shift the focus of the dialog from the base topic (in this case UTK) to the dialog itself. This gives the participants a chance to clarify underlying assumptions and express requirements to shape the dialog into a more accessible format when it returns to its original topic. The aim of a metalog is to keep the participants playing with the ball (albeit a slightly larger and off center ball) rather than the man.

So, Mike, please take these next comments as pointers on how to improve your presentation of the concept.

Early on in the dialog, some patterns developed in Mike’s responses to criticisms of the UTK.
1) Assigning the critic the task to either use UTK to dispel the criticism or find an expert in the field to join the discussion in order to defend UTK.
2) Suggesting the critic had not assimilated the material and should reread and rethink the existing pieces.
3) Making claims that UTK applications will revolutionize other fields of study, or that existing fields of study are subsets of UTK (and therefore supportive of UTK), but not demonstrating the claim.
4) Promising that all shall be made clear in subsequent postings, but those postings while adding some detail, do not add sufficient clarity to resolve current issues and concerns on the table.

These four response types have a common strategic thread, diverting attention to somewhere other than detailed and clear explication and demonstration of UTK. I believe that response types 3 and 4, Claims and promises without sufficient detail or demonstration to back them up, are what denham typifies as “marketing”. From my reading of the discussion, while people are very interested in Triadic Analysis and UTK, Mike has not yet provided sufficient information in the form of details and demonstrations to convince. I would suggest a different presentation strategy that focuses on less complex examples with more detailed deconstruction and analysis showing the advantages of the UTK approach. After the basics have been shown, then proceed into more complex deconstructions, superconstructions, etc. After several specific examples you will have a better basis to convey general principles of using UTK.

I think a good application area to begin with would be:

T(tic tac toe):
R(2 players, 2 sets of markers(x’s and o’s) and a 3*3 grid)
P(alternating turns where a turn consists of a player placing his marker on an empty cell of the grid)
= O(winning, consisting of a player aligning three of his markers in a single row, column or diagonal, or if a win is not possible, forcing a draw. Avoid losing)

What I would hope that UTK can demonstrate is how to arrive at turn valuations without doing an exhaustive tree search of possible moves. This should lead to identifying Best and Worst practices. For example, what are the best and worst opening moves (if any) and what are the best and worst second moves(if any). I would measure the utility of UTK by how few look ahead levels are needed to answer the best and worst moves questions.

As a follow up to Tic tac toe, I would suggest go-moku, a five in a row game on a 19*19 grid, or Qubic, a 3-D, 4 in a row game on a 4*4*4 cube. For extra credit you can analyze Go, a territorial strategy game that computers have a difficult time playing well because of the computational limits on tree search algorithms.

---

Follow Ups:

---

Click Here to Post Follow Up in New Forums

    Knowledge Management Think Tank (New)

  

 Subject: 

Message: