Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!ucla-cs!flowers
From: flowers@ucla-cs.UUCP
Newsgroups: comp.ai
Subject: Re: daemons... where's the name from?
Message-ID: <5265@shemp.ucla-cs.UCLA.EDU>
Date: Sun, 29-Mar-87 17:20:49 EST
Article-I.D.: shemp.5265
Posted: Sun Mar 29 17:20:49 1987
Date-Received: Tue, 31-Mar-87 00:44:11 EST
References: <442@cord.UUCP>
Sender: root@ucla-cs.UCLA.EDU
Reply-To: flowers@CS.UCLA.EDU (Margot Flowers)
Organization: UCLA Computer Science Department
Lines: 28

>this came up in a class last week; we came up with a few interesting ideas but
>no real answers. Why are "daemons" called "daemons"? that is, what is the
>derivation of that name?

From "Pattern Recognition by Machine", by Selfridge and Neisser,
Scientific American 1960, in describing the Pandemonium model they proposed:

     In parallel processing all the questions would be asked at once,
     and all the answers presented simultaneously to the decision
     maker.  Different combinations identify the different letters.
     One might think of all the various features as being inspected by
     little demons, all of whom then shout the answers in concert to a
     decision-making demon.  From this conceit  comes the name
     "Pandemonium" for parallel processing.

This paper was reprinted in the seminal and still useful book
_Computers and Thought_, Feigenbaum and Feldman, eds., 1963.  Anyway,
Selfridge and Neisser have some earlier publications about pattern
matching and the Pandemonium model which probably introduced the idea
of demons.  I don't know if their use of the term was inspired by any
prior specific use.

Around 1970 demons were utilized and popularized by Charniak's Ph.D.
thesis.

Margot Flowers, Asst. Prof., UCLA AI Lab
Flowers@CS.UCLA.EDU [or Flowers@UCLA-CS for old host tables]
...!{ucbvax|ihnp4}!ucla-cs!flowers (uucp)