[cc-community] Download data from various sources

Mike Linksvayer ml at creativecommons.org
Tue Jun 10 17:22:00 EDT 2008


On Tue, Jun 10, 2008 at 12:31 PM, Gavin Baker <gavin at gavinbaker.com> wrote:
> The JISC Information Environment Team posted a question about harvesting
> usage data for the same content from multiple sources -- namely, an
> author doesn't want multiple copies of his paper on the Web, because he
> wants "accurate" download statistics for it:
>
> http://www.earlham.edu/~peters/fos/2008/06/sharing-repository-traffic-data.html

He who hath does care more to believe falsely that he know the true
integer value of download count than that more people shall read their
work shall be cast down in the valley of salt to perish, iff the
lord's viper does not take mercy and cause an explosion in their sever
room firstly.

> The author suggests using "fuzzy matching to identify copies that were
> probably the same paper".

There are a number of companies working on this problem, one of them
attributor.com.

> It occurs to me there might be a role here for
> watermarking as well.

Watermarking brings lots of problems, but sure, whatever, it could
help a tiny amount.

But neither of these help at all with obtaining a download count.
This requires any site that might host a paper to report the count (as
the post suggests), presumably via a feed or API such that counts can
be aggregated and our small thinking sinner can believe he knows the
true integer value of his download count. Then there's the little tiny
problem of accurate reporting.

> Since this is likely to be an on-going argument against open content
> (despite the obvious efficiencies of not having to provide all of the
> bandwidth from a single point, and despite the sage advice by Cory
> Doctorow to "think like a dandelion"[1]) -- and since the CC team is so
> well-equipped in the technical department -- I thought I'd suggest that
> heads get together and gin up some ideas on this. The solution(s) may
> already be well-established, but if so, it'd be great to have (say) a
> blog post or white paper explaining what the options are and how to use
> them.

The solution for anyone who sees a problem is to get back to writing
your next contribution and peel away from bogus stats.

Mike


More information about the cc-community mailing list