[R] [OT] Is data copyrightable?

[hadley wickham]

Dear Brian, Peter, Spencer,

Thanks for your comments, which have cleared things up a little for
me.  The thing I find most confusing about copyright is that it is
emergent, not atomic - ie. if you split a copyrighted work into small
enough pieces (eg. letters, pixels) those pieces are no longer
copyrightable.  It is the combination of those small pieces into a
specific form that is important, and the definition of derivative
works seems to help define what rearrangement of those pieces is still
covered under copyright.

The specific case that I am interested in creating new data sets from
publically available data (itself stored in copyrightable works) - in
my case to produce interesting data sets to use in class.  For
example, each individual page on ebay is copyrightable, but if I
extract the price, name and category from (say) 200 pages, does the
copyright of that dataset belong to ebay?  I'm quite comfortable using
that data personally, or for a class, but if I want to publish it (ie.
in jse) do I need to get permission?  Similarly, if I take a few mp3's
and calculate some summary statistics for them, would that constitute
a derivative work?

Hadley