C70:editor-people (06/04/82)
>From ZIMMER@DEC-MARLBORO Fri Jun 4 03:23:12 1982
Two points:
1) Cursor movement takes up a lot of the time that people spend using
text editors. We have collected data with three diverse editors -- a
word processing system, a production screen editor, and an experimental editor
designed for rapid, consistent cursor movement using arrow keys.
(None of these editors used multiple windows, or I/O devices like mice.)
Each of these samples contained about half a million keystrokes.
Typing took up 1/2 of the users' time. Cursor movement took up 1/4.
Deletion took up 1/8. Everything else took up 1/8.
2) We performed an experiment involving a cursor positioning task and three
editors (not the same ones as above). One had unpredictable cursor speed
and lag when using the autorepeat key (30 cps, .5 second lag in the
hardware). Another moved at a steady 30 cps with the .5 second lag. Yet
another moved at a steady 60 cps with a .08 second lag.
The effect of editor type was significant; those using the steady 30 cps/.5
second lag editor did best, followed by the steady 60 cps/.08 second lag.
The unpredictable cursor speed and lag did worst.
Conclusion:
I used Twenex Emacs regularly for two years and have used VMS Emacs
regularly for the past year. I like these editors better than any other
editor available to me on either system. But I don't like the slow speed
of the cursor movement keys. One ^N or ^P is fine, but the problem is in
binding these functions to keys on the VT100 alternate keypad so that you
can autorepeat them. This is slow and unpredictable in Twenex Emacs. VMS
Emacs is better, but the screen manager will often delay some of the cursor
motion feedback in order to avoid unnecessary redisplay. I wish we could
do better without changing other aspects of the Emacs interface. Since
cursor movement takes a lot of the users' time and unpredictable
autorepeat/lag hampers cursor movement, there is some data to go beyond the
personal preference.
--------