[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

Re: new semantics: Stamping out PartKeys

Endorsements claiming support ("I believe/assert this"), must
    degrade when the work is vcopied, to prevent out-of-context quotation
    and piecewise-editing into something the author would never have
    endorsed.  Endorsements claiming authorship ("I/we composed this text")
    need to propagate with virtual copy.   (Perhaps, again, in a degraded
    form.)  The mileage of endorsements making other claims may vary.
    At a minimum, endosements must record the original context.
    If the context is a region of an inclusion history tree, all the better.
    Hmmm... Degrading, but not eliminating, authorship endorsement during
    virtual copy might be the right mapping of authorship in collections.

I believe this is easier suggested than done.  The ability to "degrade"
an endorsement implies that the endorsement has a "grade", or some
measure of quality, in the first place.  Various calculi spring to
mind:  there could be just a few different grades (three or four), with
the highest corresponding to the authorship endorsement of the original
and true author of an unchanged document, and the lowest perhaps to a
completely fraudulent claim of authorship.  Or there could be a
continuum, which, in the case of authorship-endorsement, might simply
reflect the percentage of the bytes in the document which were the work
of the claimed author.  But we see that attempts at such precision lead
inexorably to a quality measure in a much more complex coordinate
space, one ultimately capable of mapping the entire backend semantics,
so that the measure of quality precisely reflected the editing changes
and included documents which diluted the authorship claim.  If that is
the implementation desired, it begs the question entirely: rather than
the authorship endorsments carrying around such a baroque measure of
their validity, the measure could probably be more easily evaluated at
the time it is needed, from the actual link structure and version
history of the document.  And if a simpler measure is adopted, there
remains the problem of developing an arithmetic which will keep the
endorsement's quality number in line with the subjective validity of
the authorship claim.