RE: straw poll: "parent" vs. "in-reply-to"

["Bob Wyman" <bobwyman@xxxxxxxxxx>]

Ken MacLeod wrote:
>>"Dare Obasanjo" <dareo@xxxxxxxxxxxxx> writes:
>> to trace conversations across weblogs which is very different...
> It is no different, the intent is exactly the same as if the 
> reply were made in the "comments" of the hosted entry
	It's about the Entries!
	One entry can be declare or assert an "in-reply-to" relationship
with another entry by including an "in-reply-to" element. All that
matters is that one entry is related to another. If there are any
mechanical difficulties that arise because the entries are "in"
different feeds, then we should accept that such difficulties probably
indicate a weakness in the system. The relationship between the entries
will exist whether or not it can be expressed "in Atom." The interesting
question is whether Atom can or should make it easy to capture the
relationships that exist? If so, should it be using a new mechanism, or
should we rely on past learning from things like email, news, and
Annotea? (Also, what is the URL for an entry?)

> The In-Reply-To of email and References of email and 
> UseNet are the semantics being modeled.  Variously, 
> In-Reply-To and References have been used (abused) as an 
> M:N relation, but practically have only interoperated as 
> a 1:N relation.
	I've been in discussions of "in-reply-to" for something like 23
years now and I while I've seen regular implementation and use of 1:n
systems I don't think I've seen anyone yet come up with a decent M:N
discussion management system. Many M:N systems have been proposed and
there are devotees of all sorts of specialized systems. The M:N problem
is, as Dare suggests, different from the 1:n problem -- it might even be
more "interesting." However, we've seen on numerous occasions that
simply solving the 1:n problem is sufficient to provide tremendous value
for an exceptionally broad range of uses. If we can provide good 1:n
support, we shouldn't be too bothered about not having solved the M:N
problem.
	Note: I don't believe that a decent M:N solution can be had
without providing access to the full graph of inter-entry connections.
While Waypath, Technorati, PubSub, Microsoft, HP, IBM and numerous
others all capture large portions of the graph, I think it best to avoid
the problems of sharing that graph at this time. (It is massive...)
Let's walk before we can run.

		bob wyman