From: utzoo!decvax!ucbvax!ARPAVAX:C70:editor-people
Newsgroups: fa.editor-p
Title: usage statistics
Article-I.D.: ucb.2199
Posted: Mon Oct 25 03:28:52 1982
Received: Mon Oct 25 06:41:32 1982

>From Admin.JQJ@SU-SCORE Mon Oct 25 03:24:37 1982
A number of contributors to this list have collected and submitted
statistics on editor usage.  I would be interested in hearing from
other members of this list just what statistics are most useful to
you, and for what purposes.  For example, are interested in seeing
more breakdowns of editor usage by keystroke frequency?  By command
frequency?  By how much cpu time the editor is spending in various
tasks (GC time, paging, text input, etc.)?

One statistic that I would like to see gathered is transition
probabilities on keystroke pairs, and higher-order transition rates
between types of commands.  Thus, I am most interested in knowing
whether people tend to type strings of "forward-character" commands
(if they do, that lends credence to the proponents of autorepeating
arrow keys).  I am interested in knowing how frequently people in
Twenex Emacs type ^X^S^X^Z -- if very frequently, then either ^U^X^Z
needs more publicity, or a separate "save and exit" command might be
a good idea.  Do people tend to type long sequences of cursor motion
commands, or do they intersperse cursor motion with printable-
character insertion?  I want to know what pairs of functions tend to
occur together so I can design auxiliary keyboards that minimize hand
movement.  And so on.  These questions require that people keep more
than just running totals of keystroke frequency; I need ordinal data
also.  There have been some attempts to collect such data (notably by
Michael Good at DEC); I'd like to see such data collected for a
variety of different editors, in a variety of environments.
-------