Re: Interractive Response Time

From: Hall, Philip <phall@spss.com>
> Also found is that there is a big difference between machine response time
(as measured by timers in the code) and user perceived 'wall time' (as
measured by looking at the clock on the wall)
>

perception is king. When I wrote the software that is (was?) behind
the telephone company's Directory Assistence in New York City,
the contract defined response time as the time from the user's
last keystroke to the time the first character of the response
appeared on the screen. At the time, the system used async 56Kb
lines so the transmission time was noticable. We first implemented
the system to begin showing the response as soon as some data
had been received (to kind of comply with the contract). Then
just for fun (actually to test a theory of mine) I buffered up the data
inside the terminal until all the data had been received and then
FLASHED the response on the screen in one go. Although the
result showed an average of 0.5 second later, ten minutes after
the new method had been downloaded, I got a phone call from the
field (there were 2000 terminals out there) and a voice asked me
in a "frantic" way: "WHAT DID YOU DO". Preparing myself for
some halfbaked excuse, I heard her next utterance: "everything
is SO much faster now". Needless to say that was the way we left
it.