[h2o-discuss] Re: meta-discussions

[j proctor <jproctor@oit.umass.edu>]

>> I'd have time to work on this software-- but I'd need help. Maybe at
>> the very least h2o could work to define some standards, and then demand
>> them from ThirdVoice and other potential vendors.

I'm game.  I've never used ThirdVoice (no Netscape plug-in yet, and I
refuse to install IE for anyone).  What kind of standards, though?  To be
considered an interactive document, your server must offer feature X?  
That can't be right.  Please elaborate a little more.  And if we "demand"
anything, what's the penalty for not complying?  Microsoft hasn't complied
with W3C's HTML standards ever, and Netscape hasn't yet (although they're
working on it), and both of them seem to be doing just fine.  If I recall,
the W3C actually trailed the browsers for awhile, to make the standard
include new elements from the software makers (that didn't contradict some
other well-justified standard) that were already in common practice.

Like open source in general, this is a public relations and education
campaign.  People have to know about and *want* a standard before vendors
will react and provide it.  And that standard must include the features
the people already expect and the room for manufacturers to improve, or
they'll rebel and make their own de facto standard without, perhaps, as
much thought for general use as a group like ours might put into one.


> Hyperlinking currently allows us to make direct references among existing
> works, or, if we have a full-fledged idea (and a webserver), to create a
> new work linking back to its sources.  Annotation, as I see it, could be an
> intermediate step -- allowing people to add brief ideas to an ongoing
> discussion, scholarly criticism around an otherwise static piece, or
> commentary on a work-in-progress.  With an open source annotation protocol,
> anyone could set up an annotation server with its own database to hold
> these conversations in public or private. 

I think it's important to remember that simple annotation is *only* an
intermediate step towards the kind of discussion that we want.  Just as an
example, I've taken Wendy's letter and chopped it into pieces, commenting
on the parts that I think are significant.  It's likely that different
people will react to different parts of my notes (*ahem*), or want to add
their own take on things regardless of what I've said.  In fact, it's
possible that's occuring while I type.  It stretches the limits of
"annotation" to describe or represent things efficiently.

What we end up with is a branching, multi-threaded discussion.  In
email, we prepend >'s to indicate history, and consider it etiquette to
remove sections to which we're not responding.  On a busy mailing list,
though, we just keep replying and don't always modify the subject line to
indicate we've focused on a particular point or two, so everyone sees all
the discussion about everything, regardless of whether they care.  On the
Web, though, we haven't been creating and quoting new copies of pages then
inserting our own text, then linking them in chronological series to the
end of the original.  That is, in essence, the same treatment that email
gets, although it somehow seems ridiculous when applied to a different
medium.

In fact, I'd like to start a new thread right here.  Below, I'll continue
to discuss the BCIS Annotation Engine, because I agree that it's not a
step in the wrong direction.  But I don't know whether it's the *right*
direction, or whether it's a big enough step to make a difference to the
art or science of meta-discussion, and if not, how we can jump directly
into representing exactly this kind of branching discussion using the Web.
There you go.  A spark for a meta-meta-discussion (are notes the best
thing we can talk about?), and a call for new ideas, preferably with
examples, about the information architecture (including infrastructure,
interface design, and any other "i" words you can think of) necessary to
get the Web caught up, at least, to what we can do with email.  Someone
quote some lines out of this paragraph and respond.  :)


> I had set up an early, and slow, prototype (pre ThirdVoice!) of an
> annotation proxy, thinking that online courses could use such a tool for
> assigned reading and marginal notes.  The proxy and its perl scripts are
> online at <http://cyber.law.harvard.edu/projects/annotate.html> .  A
> browser plugin would probably be faster, if less portable.  It would be
> great to start a full-fledged h2o project -- and before Microsoft captures
> the field with proprietary "interactive document publishing" in Office 2000.

It might be informative if you set up a test document that this list could
actually use the engine on, just so those of us that are interested (but
too busy/lazy to read through the source code right now) can get a feel
for the current state of things.


> As to standards, the most immediate need I can see is a better way of
> describing position in HTML documents.  Pegging comments to text phrases is
> slow, and can be ambiguous for longer documents. 

Unfortunately, I don't think there is much of a position, especially in a
dynamically annotated document, other than "right here".  Byte offsets,
line counts, etc., all depend on content that's not changing while you're
composing your thoughts. 

I can imagine that keeping track of text in paragraphs might be some
improvement.  The engine could assign a unique identifier to each
paragraph (or "block-level element", for a little more control and
consistence with W3C language and intent), and present the text with some
mark (a pilcrow* maybe) that's actually a hyperlink to the form where
the user can enter his notes and specify an exact word-boundary in a much
narrower field than the entire essay.  From what I can infer from the
sample, this would be a fairly significant change to Wendy's existing work
(a new major version number, in software terms), but it wouldn't be
terribly difficult to add.


j proctor
Web Site Coordinator
UMass Office of Information Technologies


*  A pilcrow, for those of you who aren't also typography geeks, is
the stylized "P" we use as the editor's mark for paragraph.