[net.ai] Tail Recursion is Iterative

gary@rochester.UUCP (Gary Cottrell) (03/09/84)

From: Gary Cottrell  <gary>
>From decvax!decwrl!rhea!orphan!benson Thu Mar  8 23:44:38 1984
Date: Thursday,  8 Mar 1984 18:57:59-PST
From: decvax!decwrl!rhea!orphan!benson
Subject: Re: Tail recursion.  Please forward to Mr. Sloan.
To: decvax!harpo!seismo!rochester!gary             
Status: R

Dear Mr. Cottrell:

  I do realize that in most cases (I.E., everyday programming), tail recursion
can be reduced to iteration. However, in my study of this aspect of dog
modeling, I found the underlying MOTIVATION to be recursive in nature. Clearly
this is not a concept which can be applied to programming outside the AI realm.
(And when I say "AI", I of course mean "AI", not "AI"). My canine subject did
not set out to chase his tail for i equals 1 to n. Nor did he intend to chase
it until some condition was met; the obvious condition being "has the tail 
been caught?" In fact, frequent experiments showed that actual tail capture
did not necessarily end the cycle, and it often was not achieved at all before
cessation of the chasing activity. No, a more realistic model is one in which
a bored or confused dog initiates an attempt to catch his tail. During this 
process, the previously unseen tail falls into view as the head is turned. 
The dog's suspicion is aroused; is this some enemy preparing to strike? This 
possibility causes an attempt to catch the tail. This causes the tail to fall 
into view....   and so on. The recursion may be terminated either by some 
interrupt generated by an unrelated process in the dog's brain, or by forced
intervention of the dog's master. The latter is dangerous, and should be 
scrupulously avoided, because it does not allow the dog's natural unwinding
mechanism to be invoked. Thus, the dog may carry unnecessary Tail Chasing
Activity procedure frames around in his brain for years, like a time bomb
waiting to go off. This, indeed, is a subject deserving further study.
  In response to your other question: you are welcome to post my AI reports
wherever it seems appropriate.


					Tom Benson