Re: [USB] Keyboard, mouse and ergonomy

From: Alexander Leidinger <Alexander_at_Leidinger.net>
Date: Mon, 27 Sep 2010 15:34:19 +0200
Quoting Hans Petter Selasky <hselasky_at_c2i.net> (from Mon, 27 Sep 2010  
14:21:42 +0200):

> Hi,
>
> I was thinking about adding a sysctl to ukbd and ums that shows how many
> keypresses have been done and how many pixels you have moved the mouse during
> a day. These number will mostly be useful for making ergonomic arguments that
> a certain way of working is better than another one.
>
> Anyone have any comments or input on this? Are there any security concerns
> about this?

Be careful with ergonomic arguments. For example the screen corners  
are easy (and fast) mouse targets (if they are used for something in  
the software), and would give a false impression if you count  
mouse-device movement instead of mouse-pointer movement on the screen  
(which you can not measure in ums). The most easy ergonomic point for  
a mouse is the current position (pop-up menus), and this involves not  
much movement...

Regarding key-presses there's also the problem of having the keys  
directly under your fingers, or having to move back and forth with the  
hands (also depends upon the keyboard layout and type of work you do).

Regarding metrics I suggest to:
  - add a click-count
  - not hardcode the counter-reset (let a sysctl handle the counter reset
    upon request)
  - differentiate between normal key presses and shift-ed/control-ed/...
    ones (I do not think that caps-lock should count as shift-ed)
  - I do not know if it makes sense to have a hold-time (for
    shift/control/...), if this is for real ergonomic research,
    this could be helpful

Regarding the security:
  - don't make this real-time stats, add some artificial delay
    (you could interpolate what is typed just by watching the
    stats), I suggest to make the delay at least several
    seconds long (to cover people with disabilities and maybe
    people which search each character with one finger), for
    real ergonomic research this is counter-productive (but
    you stated that you want to measure on a per-day basis,
    to this should not matter in your case)
  - make it depending on a compile time knob (disabled
    by default) and issue a warning on device attach if
    compiled in

Bye,
Alexander.

-- 
An honest tale speeds best being plainly told.
		-- William Shakespeare, "Henry VI"

http://www.Leidinger.net    Alexander _at_ Leidinger.net: PGP ID = B0063FE7
http://www.FreeBSD.org       netchild _at_ FreeBSD.org  : PGP ID = 72077137
Received on Mon Sep 27 2010 - 11:34:35 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:07 UTC