Posted by Jerry Stuckle on 07/16/07 19:27

axlq wrote:
> In article <ctydne8X1IgZuAXbnZ2dnUVZ_gOdnZ2d@comcast.com>,
> Jerry Stuckle <jstucklex@attglobal.net> wrote:
>> axlq wrote:
>>> It is reasonable to assume that for a site like a forum, if a user
>>> logs in, that user will probably still be looking at the forum a
>>> minute after the user's most recent activity. It should be easy to
>>> keep a running database of users whose recent activity is less than,
>>> say, 5 minutes ago. The list could be displayed alphabetically or
>>> sorted by timestamp.
>> No, that is not a reasonable assumption. Some people may stay 10
>> second, others 10 hours.
>
> Yes, it is a reasonable assumption - in some cases. For a
> heavily-visited forum, statistics can be gathered about the
> likelihood of a user staying on the site after the most recent
> activity. Using these statistics, a site can get a fairly good
> idea of the approximate number of users online at any given moment,
> and could even get a good idea of the identity of the ones who are
> likely to be still logged in.
>

And after which page does he close his browser? The first? The
seventeenth? The forty second? You have no way of knowing.

>> Any other figure is a pure crap shoot.
>
> Not if you have measurements and a lot of traffic. It's far from a crap
> shoot, but it isn't precise either.
>
> -A

I agree - it's not a crap shoot. It's worse. At least with a crap
shoot you know you'll get a random roll. But your way the client
*thinks* he's getting accurate results - and they are junk.


--
==================
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex@attglobal.net
==================

[Back to original message]