Search feedback (was: Re: Slow HTTP Responses)

M. Strata Rose (strata@fenchurch.MIT.EDU)
Thu, 13 Jan 94 19:48:05 EST

I think many people would be happy with (as a minimum) some combination
of "I found N hits" and "I know about X amount of data and am Y the way
through it". This conveys progress while not attempting to solve the halting
problem, especially if Y is simply a number and not a percentage. This
method can peacefully coexist with discovery-style searches, where the user
might see the following status info over time:

Data pool: 25 Mb in 18 files
Searched: 14 Mb in 11 files
Hits: 3
Current data: http:/foo.bar.baz:splat (3.2Mb)
Timestamp: Thu Jan 13 16:37:50 PST 1994

...

Data pool: 43 Mb in 25 files
Searched: 31 Mb in 18 files
Hits: 8
Current data: http:/foo.bar.waldo:roadkill (12Mb)
Timestamp: Thu Jan 13 16:57:20 PST 1994

...

Data pool: 2011 Mb in 62 files
Searched: 87 Mb in 43 files
Hits: 17
Current data: ftp:/splat.foo.woo-woo:bigdata (1029 Mb)
Timestamp: Thu Jan 13 17:12:20 PST 1994

...

And so on. Additional features of the search could be a [STOP]
button that can be hit to bypass a particular document, such as
the mythical "bigdata" in the last example. Users could have more
control still with a "Current Hits:" status line that updates hits
from that particular document and then adds them to the "Hits" display
only when that document is finished. Our hypothetical user might
change his mind about "bigdata" if he/she could see that 7 of those
17 hits alone came from that document and we still weren't done
searching it.

This is more UI stuff then Web-specific, but I consider it an example
of features people are likely to want, and things which are desirable
and could be relatively easy to provide if we support an infrastructure
for them.

Comments? UI-specific comments might go to me and not the list.

_Strata

M. Strata Rose
Unix & Network Consultant, SysAdmin & Internet Information
Virtual City Network (tm)
strata@virtual.net | strata@hybrid.com | strata@fenchurch.mit.edu