On Mean time of Survival of URLs -Reply

[Dan Lester]

>>> "JQ Johnson" <jqj at darkwing.uoregon.edu> 07/29/97
07:28pm >>>
By the way, as the age of the web as a whole increases,
"mean" will become
an increasingly bad statistic to use for page lifetimes, since
---------
I agree with your distinctions between the two studies.  They
are indeed very different statistics.  
=============
the mean
will be dominated by that small subset of pages that have
been around from
the beginning.  Median lifetime and median age are much
more useful
statistics for most purposes.
----------------------
Again, I agree the medians will be useful, but means will also.
 It is always important to look at both, as well as the graph of
the distribution.  My bet is that the domination will NOT be by
the "ancient pages", but by the NEW pages.  The ancient
ones will be a very small bump on the left end of the
distribution curve.  The new ones will be a very large bump on
the right end.  This will become even more the case as more
and more pages become dynamic, and thus ALWAYS new,
and NEVER old.  For example, CNN pages are mostly new,
and will always be so.  No matter how much they choose to
archive, there will be a vast quantity of new pages.

cheers

dan


Dan Lester, Network Information Coordinator
Boise State University Library, Boise, Idaho, 83725 USA
voice: 208-385-1235   fax:  208-385-1394
dlester at bsu.idbsu.edu     OR    alileste at idbsu.idbsu.edu
Cyclops' Internet Toolbox:    http://cyclops.idbsu.edu
"How can one fool make another wise?"   Kansas, 1979.