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

The meaning of authorship

Dean & I had a breakthrough of sorts yesterday.  I think we figured
out what authorship means in a Xanadu context.  First let me explain
why it is a hard problem.

The thinking to date (worked out with Drexler):

In the conventional world of discrete documents, a document can have a
set of authors.  The authors at any time (during the development of
the document) may modify the document, and add others into the set
of those invited to be authors of the document.  There may be people
who exercise write permission on the document who do not list
themselves as an author (such as a proffredder), as well as those whom
the authors would like to sign up as co-authors, but who don't
consent.  In order to be an author you both have to be invited & you
have to consent.

(An Aside: DataOrgls seem to be transmuting into DataBerts (not to be
confused with Date of Birth).  This is not a significant operational
change, it just makes aspects of the semantics more uniform (I

This works fine for DataBerts, where one is concerned with the
original authorship of actual data, but what does it mean for MapOrgls
where anyone may make a new Orgl derived from any Orgl he may read
either by correcting a typo or completely replacing the contents?

Not to keep you in suspense any longer, what we came up with is that
the owners of a Bert may invite others to "author" the document that
the Bert represents by giving them write permission on the Bert.
Anyone may "endorse" any Orgl he may read--meaning sign his name to
its contents.  This last corresponds to public-key, where you (and
only you) may sign your name to any data you may read.  The authors of
the document represented by the Bert are precisely the intersection of
those invited to write the Bert, and those which endorse the Orgl the
Bert is currently on.  This means that if one of the authors modifies
the document in some way, he cannot go and publish the new thing with
all the old names attached as authors.  Even if he just corrects a
typo, he must go back to the original authors and get them to endorse
the new Orgl, if he wants them to still be considered authors.  (Note
that good version compare support enables other authors to quickly
inspect all the differences between the current state of the document
& the last one they endored.)

Comments?  (This stuff is pretty new, please shoot it down.)