Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!cbatt!cbosgd!mark
From: mark@cbosgd.ATT.COM (Mark Horton)
Newsgroups: net.text,net.unix
Subject: Re: Use of ``vi'' for business office word-processing
Message-ID: <2638@cbosgd.ATT.COM>
Date: Mon, 29-Sep-86 17:03:16 EDT
Article-I.D.: cbosgd.2638
Posted: Mon Sep 29 17:03:16 1986
Date-Received: Tue, 30-Sep-86 20:58:18 EDT
References: <1413@bu-cs.bu-cs.BU.EDU> <189@bsiao.UUCP>
Organization: AT&T Bell Laboratories, Columbus, Oh
Lines: 50
Xref: watmath net.text:1389 net.unix:9405
Summary: it wasn't the CACM paper

In article <189@bsiao.UUCP> uh@bsiao.UUCP (Uul Haanstra) writes:
>> For example, I was told of a paper a while back done within Bell Labs
>> (probably late 70's) where they brought in some outside researchers to
>> measure some things about UNIX and some of the editors. If I remember
>> right they gathered a few different experience groups and plotted
>> 
>I have seen such an article in the Communications of the ACM, in the

The paper in CACM isn't the one being referred to.  The Bell Labs
paper was by Merle Poller and Susan Garter in 1983.  That's an
internal version; I think it was published somewhere externally too.
And I think there were different versions.

The paper (I have it in front of me) compared experienced vi, emacs,
and ed users.  (This makes the validity questionable, the type of
person who uses ed is different from the type of person who uses vi
or emacs, especially in 1983 before everyone had vi.)  It found that
vi and emacs were much better than ed, but close to each other.  One
surprising result was that the vi group made fewer mistakes than the
emacs group.  (Apparently the emacs in question is Montgomery's.)

Other conclusions:

The vi type editor is the choice for editing from marked-up hard copy.
[This is because you're mostly typing commands, which are usually
lower case letters, often on the home row.]

An emacs type editor is the choice for free composing at the terminal.
It is also the choice for editing that primarily involves many small,
close-together insertions.  [This is because you're in "input mode" most
of the time, and not having to switch modes to fix typos you just made
is an advantage.]

Screen editors are preferable to line editors for editing from marked-up
hard copy and for composing at the terminal.  However, an ed-type
editor has some speed advantages in certain situations.  It is the
editor of choice for making a few simple editing changes to an existing
file.  [This is because ed is small and doesn't display output, so on
a heavily loaded 16 bit machine without enough RAM that swaps itself
silly, you'll get faster response.  For creating a 4 line text file,
I go one step further and use "cat > file" as my editor, unless the
system is nice and unloaded.]

The arguments about number of keystrokes, mode errors, etc tended
to pretty much be a wash.  The vi group was a bit faster and more
accurate, but they tended to be less willing to go back and fix a
typo, preferring to go back later and fix it.  Evidently it was
too much work to go into/out of input mode.

	Mark