Re: First Class Links Make Second Strike

[Marc Stiegler <marcs>]

Dean, in reading markm's discussion of embedded links, has the 
following comment:

One of my most vivid arguments for embedded links is tracking version
in a comment (or refutation).  While reading this I had a glimpse of
reviewing a document.  Rather than making separate links to comments and
refutations, I would put all my review links into a review document.
The document might contain all reviews I made to several version of
the document.  Or it might be versions in parallel with the document
under review to reflect the comments and such appropriate to each
version.

------

I can't help noting that one can get at least similar results
with first class links, using mechanisms that will be available
even in our very first frontend. The underlying mechanism is to 
vcopy the comments made on one version together into a comments
document upon which you maintain version control: in the parlance
of the InfoFactory,

1) attach to the document a series of comments, where each comment
is a separate document with a separate link to the part of the
document it addresses;

2) Perform a "Create Inclusion List from the Link List", which 
collects all the comment documents together into a single
inclusion list. Now you have a single comment document on that
version of original document. 

Make this a containment-style document so that you have version 
control over it.

3) When a new version of the document comes round, and you make
new notes and new versions of old notes, build another inclusion
list out of the new link list, and you've grabbed the version of
comments attached to the new version. Compose a second-tier inclusion
list out of the first-tier inclusion lists, and you have both 
parallel and all-subsuming views of the reviews of the original
document. 

I have one comment on markm's philosophy of embedded links, a particular
point that comes closest to explaining why I am troubled by 
embedded links, and least when viewed through the analogy to 
sentences. Markm states:

we
   can (I claim) frequently reason about the proper place of an
   individual link of a document by making an analogy with the more
   familiar individual sentence in a document.
----------
What's troubling here is the following: I seriously doubt that there
was ever a time in the history of writing, even in its very earliest
moments of invention, when people had the least doubt about which 
document into which they should put a particular sentence. The sentence
always clearly belonged in the document they were thinking about
at the time when they composed the sentence for it.

The closest analogy for links is, the link clearly belongs to the 
group of documents being linked together by the link. Links step
beyond the modular integrity of documents as we now understand them.
Indeed, the very purpose of links might be said to be modularity
breaking! (of course, in a good, no doubt object-oriented, way :-)

Anyway, just as it is intuitively obvious which document a sentence
should go in, it is pretty intuitively obvious which group of
documents a link belongs with. Picking one rather than the other
as being "home" only seems benign as long as one doesn't count
on the relationship for anything special. Dean's review document 
is interesting, yet, frankly, if I were interested in reviewing
the original document and the comments made on it, I would perform
backfollows to find ALL the links, regardless of whether they had
been embedded in Dean's review document.

Well, I've said too much already; I've already said we need more 
bandwidth; I'll stop here.

--marcs