On Tue, 19 Sep 1995, John Chadwick wrote:

> At 4:23 PM 9/17/95, Uri Bruck wrote:
> >One thing we should all be wary of are all those providers who promise to
> >tell us the number of 'hits' a web page gets.
> >The numbers only indicate how many times a document was ordered from the
> >server, but once the document has been to the end-user the server no longer
> >has any information on how the page is used, how long is it being viewed, or
> > [snip]
> I couldn't agree more! The number of hits is only the number of files being
> transferred. I would still like to find an effective way to get a handle on
> the visitor experience beyond the nice notes I get several times a week.
>
> --john chadwick

I can try to bounce off a suggestion here. Perhaps we should demand more
precise information from providers.
While we can't get real stats, we can still get some information.
Now, when someone logs on through a SLIP connection, and most readers of
web pages would, their machine gets a temporary name from the provider.
If the same person reads more than one page from the same server during a
short period of time, we might assume that they could be reading the
pages in succession. Factoring in approximate download times we could build
these tiny user profiles, compare these to the numebr of hits, and get a
better picture of how people read our pages. Naturaaly, we would to tkae
into account that if we encounter the same temporary machine name with an
hour's difference, it's probably a different person, and we still can't
tell how many of the different visitors are the same people, unless we
specifically ask them to sign a guest book, but I think the service we
should demand from providers should be something along these lines. Have
some proffessionals work out the details, and get some real information
about web-museum interaction.
Uri