"Lex Spoon" <lex at cc.gatech.edu> wrote:
	If you wanted to actually rub out a character, you couldn't!  However,
	on a *punch card* you sort of could: back up (with the BS key,
	probably), and hit DEL, which has all the bits turned on.  Now your
	punchcard has an entry that has been filled with ones, which if I've
	heard correctly most punchcard readers would ignore.
	
This might have been true for IBM's 96-column cards (which used the
ASCII code; basically they were folded short bits of paper tape).
It wasn't true for the 029 keypunch.  There was a backspace key.
It didn't punch any holes.  It moved the card back.  There wasn't any
DEL key.  If you overpunched and made an invalid character, the card
reader would spit your card out and sulk.  The way you fixed mistakes
on a punched card was by duplicating the correct prefix onto a fresh
card, throwing away the old one, and continuing on the new card.  You
learned fast not to make too many mistakes...

DEL was for *paper tape* (and communications protocols).  According to
the ASCII specification, text (such as you might find in strings) may
not contain NUL or DEL characters.  (So C's strings _are_ kosher...)
On paper tape, "there is a NUL here" and "there is no character here"
looked very very similar:  no holes.  Add to that the way that paper
tape readers (especially the ones on Model 33 Teletypes) just _loved_
to crumple and tear your data.  And we thing _we_ have problems?

The most famous use of backspace may be APL.
For example, you logged out by typing O<bs>U<bs>T.

It's worth noting that while the ASCII standard explicitly says that
characters may be composed using BS and CR tricks (and uses characters
that were _designed_ to do double duty as accents:
    " = double quote AND umlaut
    , = comma AND cedilla
    ' = apostrophe AND acute accent
    / = solidus AND slash-for-O
as well as ` ~ and ^ which were mainly intended as accents),
the ISO 8859 family of standards explictly forbids this usage.
(You thought Latin-1 was a compatible extension of ASCII?  Not really.)
I think it's fair to say that standards-wise, BS-for-overstrike is dead.