4.0427 Nature of Computers (2/32)

Elaine Brennan & Allen Renear (EDITORS@BROWNVM.BITNET)
Mon, 27 Aug 90 17:08:33 EDT

Humanist Discussion Group, Vol. 4, No. 0427. Monday, 27 Aug 1990.


(1) Date: Mon, 27 Aug 90 17:54:45 GMT+0100 (17 lines)
From: macrakis@ri.osf.fr
Subject: 4.0421 Digital computing

(2) Date: 27 August 1990 13:35:34 CDT (15 lines)
From: "Michael Sperberg-McQueen 312 996-2477 -2981" <U35395@UICVM>
Subject: [Calculation Tradeoffs and Bases; 0/1 vs. T/F --eds]

(1) --------------------------------------------------------------------
Date: Mon, 27 Aug 90 17:54:45 GMT+0100
From: macrakis@ri.osf.fr
Subject: 4.0421 Digital computing

Electronics can perfectly well represent multiple distinct values by
multiple distinct voltage levels (or current flows, or whatever).
Binary circuits are simply by far the cheapest, fastest, smallest, and
most reliable today, and anything else can straightforwardly be
simulated using them. (By the way, most (all?) non-binary machines in
the past were actually NOT based on non-binary electronics, but simply
simulated other bases using binary circuitry.) Analog circuits (where
there are no distinct values, but rather a continuous variation) have
been used in the past for certain kinds of numerical computation, and
there is (as I said in a previous message) some interest now in
"neural nets" (but don't be seduced by the name!), but no one proposes
to use them as a substitute for digial circuits for text editing,
databases, telephone switching, banking, etc.
(2) --------------------------------------------------------------35----
Date: 27 August 1990 13:35:34 CDT
From: "Michael Sperberg-McQueen 312 996-2477 -2981" <U35395@UICVM>
Subject: [Calculation Tradeoffs and Bases; 0/1 vs. T/F --eds]

[...] Me for base-negative-two computers: handling of negative integers
is much simpler with base negative-two than with twos-complement
notation. Only drawback is that it makes hex arithmetic even more
confusing.

But I still think that the most natural interpretation of binary digital
signals is as T/F, not as 0/1. Deep down, computers don't deal with
numbers but with booleans. And for real fun with those who claim 0 and
1 are 'really' there, ask them about tape formats, many of which do
*not* have a one-to-one relationship between the magnetic fields and the
bits represented.