Newsgroups: comp.human-factors
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!m.cs.uiuc.edu!roundup.crhc.uiuc.edu!ux1.cso.uiuc.edu!uxa.cso.uiuc.edu!lnk10562
From: lnk10562@uxa.cso.uiuc.edu (Louis Koziarz)
Subject: Re: Audio feedback from GUI's
Message-ID: <1991Jun14.154343.7665@ux1.cso.uiuc.edu>
Sender: usenet@ux1.cso.uiuc.edu (News)
Organization: University of Illinois at Urbana
References: <1991Jun12.171211.2716@cs.umn.edu> <31228@hydra.gatech.EDU> <1991Jun12.202741.16629@ux1.cso.uiuc.edu> <1991Jun12.215523.7379@cs.umn.edu> <1991Jun13.003210.23083@ux1.cso.uiuc.edu> <2090@anaxagoras.ils.nwu.edu>
Distribution: na
Date: Fri, 14 Jun 1991 15:43:43 GMT
Lines: 35

Pete Welter writes:

>There was a presentation at CHI '91 about using sounds (in this case musical
>notes and instruments) to monitor the processors in a parallel machine.

I had to miss CHI this year...sigh...

>One example assigned each processor a note, and increased the amplitude the
>proportional to the length that the processor had been idle.  The other
>example used both instruments (one for each processor) and note to monitor
>when messages were sent and received between pairs of processors.  To my
>mind, it sounding surprisingly pleasant, when I expected cacophony.  The
>people who did it are at U of Southwestern Louisiana (Albright, Francioni,
>and Jackson).  The real measure of such an auralization is whether any
>information can be gleaned from it, and although they said that their
>parallel computing expert benefitted, there are no hard numbers to back
>this up yet (this was advertised as work in progress).

My expectations matched yours here. At first read, I was thinking major noise
pollution in a case like that, but if the selection of sounds are right, it
probably would sound pretty good.  The true advantage I find in a system like
that is that experienced users can derive a whole lot of information 
holistically from listening to the sounds.  And you do mention that the PP
expert benefitted.  There is a lot of work being done in integrated status
display indicators, where information from a lot of sources is crunched into
a single display indicator (say, an octogon).  When certain factors change,
the shape will change (for example, one vertex may draw closer to the center).
An experienced user can draw a lot from that small change.  This already seems
to exist in a lot of places, such as the PC fan that slows when being accessed
on a network.  These are subtle cues, but do convey valid information...

Louis Koziarz
University of Illinois
koziarz@uiuc.edu

