New York Times on the Web Forums Science

Missile Defense

Technology has always found its greatest consumer in a nation's war and defense efforts. Since the last attempts at a "Star Wars" defense system, has technology changed considerably enough to make the latest Missile Defense initiatives more successful? Can such an application of science be successful? Is a militarized space inevitable, necessary or impossible?

Read Debates, a new Web-only feature culled from Readers' Opinions, published every Thursday.

rshow55 - 06:59pm Aug 23, 2002 EST (# 3946 of 3948)
Can we do a better job of finding truth? YES. Click "rshow55" for some things Lchic and I have done and worked for on this thread.

Brent Staples describes major problems and some available solutions vividly and perceptively in Mayor Bloomberg's Test: Teaching the Teachers How to Teach Reading by BRENT STAPLES http://www.nytimes.com/2002/08/23/opinion/23FRI4.html

Is it possible to do much better than we've done? Maybe.

3925-3926 rshow55 8/23/02 10:29am

3931 rshow55 8/23/02 4:55pm describes random presentation drills, which can work for letters or words, according to the following pattern:

Words in frequency order 1-6 - the, of, and, a, to, in , randomly presented:

of a to the in of and the a of the a in and to of to a and in of the of the a in in to of in to of the to in and and of the of and and in and of and the to the to and in the to in the a in a in and to a and in the of in a in the to the in to of to a in of a the to of to of to the and in a of a to the to and the to and

Since some words are MUCH more common than others - and provide the "skeleton" of the language readers have to interpret -- perhaps these words are worth learning by "see-say" drills.

Whether they are or are not is a question of fact - - and a question connected to some key issues at the interface between the statistical and the symbolic in psychology.

In the 1950's, Herbert Simon and others developed "artificial intelligence". With computers that are laughably small by today's standards, they were able to prove very many of the proofs in Russell and Whitehead's Principia Mathematica -- do much other logic -- and play games like "the Towers of Hanoi" and Chess.

Simon, H.A. MODELS OF THOUGHT Yale U. Press, 1979

AI did a great deal - and very economically - and then this "symbol based" approach stalled - though it remains useful, and central to "intelligent" computer programs that are actually used.

Nobody doubts that symbol processing, when it is set up, can do things VERY efficiently. But it couldn't "learn." By the mid-1970's people were casting about for other approaches (approaches that Rosenblatt, of Cornell, pioneered in the 1960's). And so an essentially statistical approach arose - and came to command truly huge levels of attention and funding. Key results in the field of "connectionism" are set out in

Rumelhart, D.E. and McClelland, J.L. PARALLEL DISTRIBUTED PROCESSING: Explorations in the Microstructure of Cognition (two volumes) MIT Press 1988

By the early 1990's "connectionism" was becoming disappointing to some - because it was computationally very expensive to so things people and animals did much more easily. And impossible - in a strict mathematical sense - to use the approach to do a lot of things that people did without much effort.

Judd, J.S. NEURAL NETWORK DESIGN AND THE COMPLEXITY OF LEARNING MIT Press, 1990.

Even so, the importance of statistical approaches at the "microscale of cognition" is plain to just about everybody -- and the POWER of statistical approaches has been clearly shown - and embodied in search engines on the web.

Landauer T.K. and Dumais, S.T. “A Solution to Plato’s Problem: The Latent Semantic Analysis Theory of Acquisition, Induction, and Representation of Knowledge” Psychological Review, v 104, n.2, 211-240, 1997 --- draft: http://lsi.argreenhouse.com/lsi/papers/PSYCHREV96.html

The connection between the statistical and the symbolic in human reasoning remains a key problem. Perhaps the key problem in understanding how reason can work as well as it does. A problem that has concerned philosophers since Socrates, and to the present day.

rshow55 - 07:00pm Aug 23, 2002 EST (# 3947 of 3948)
Can we do a better job of finding truth? YES. Click "rshow55" for some things Lchic and I have done and worked for on this thread.

I hope, and believe on the basis of just a few tests, that random drills for "see-say" facility can be very useful in improving reading instruction.

Whether they are or are not is important because reading instruction is important.

But these drills also seem to be about the right size and form to test a great deal about how statistical and symbolic patterns interface as people learn.

I believe that lchic and I, working together, have come to focus on something new and hopeful. Under a lot of circumstances - the odds of getting orderly answers, and discarding mistakes - is much better than people have understood. Enough better to give good reasons for disciplined hope.

I'm taking a break.

lchic - 07:50pm Aug 23, 2002 EST (# 3948 of 3948)

Well deserved.

I'll be taking a late August Break - out to the Pacific - next week.

New York Times on the Web Forums Science Missile Defense