Wow, lots of great ideas coming in on this. I wonder if fleshing out a
test case would be useful for the class of materials that might have
trouble finding a home in standard publications. For example, Roy's
MODS<->MPEG DIDL shootout. Let's say that Roy put a digitized book on the
code4lib server (maybe one of his own :-), and enough descriptive metadata
to kick things off. Roy then challenges the world to do something
impressive with the results, maybe directly contacting some MODS and DIDL
folks. Those of you who have ever seen Sun's blueprints know that this is
sort of the intent behind their blueprint series, define what seems to be
a agreed-upon problem and let the "best of breed" solutions emerge.

So far, any of this could happen in either the lead-up to a traditional
article (groundwork for "a comparison of MODS and MPEG DIDL") or this kind
of task could be described and offered out on any blog. Now a group like
the folks working on Evergreen do a heck of a lot with MODS and maybe they
decide to take up the banner for it. This leads to the first of a couple
of TPMDPs (Traditional Publishing Model Departure Point):

TPMDP #1: Some sort of step by step capturing mechanism kicks in for
recording how the Evergreen folks approach the task and start to build a
solution. Maybe something as low barrier as a link to their development
blog. This needs to be unobtrusive but is an important layer. It always
looks like divine design after a solution is presented, and Roy's right
that no one ever wants to talk about the wrong turns they make in the path
to a solution, but these are often the most instructive parts.

TPMDP #2: A level of comments is made available for each step of the
process. Peter's noted that comments are really hard to do well, and it
would be great if this could be done in some sort of threaded RSS model.
Jon Udell once wrote about a researcher trying to model network events as
chirping sounds so that someone monitoring hundreds of systems would
instantly notice when something was astray, and be able to instantly grasp
which system it was by the chirp that had changed. So maybe  we need a
comment chirp plugin or something, but the smart money is still on an RSS
comments feed for now. Still, it would be nice to see if comments could be
layered in more elegantly than they usually are.

TPMDP #3: An IRC bridge is made available. I once wrote a hook for IRC to
snag insults from a supybot in cocoon so that a dynamic bit of content
could appear in an S5 slide. I didn't realize until afterwards that
supybots often have incredibly foul mouths and didn't do much with the
results, but I am convinced there is something in the conversational model
in IRC that could do a lot in other spaces. The koha folks do town halls
with IRC, what if an article, or story posting, or whatever this thing is,
shows who is online in code4lib and has some sort of entry point into the
channel with a nick like "reader_wants_to_talk_about_shootout" or
something similar? Or maybe the posting itself is some sort of bot that
can be asked to record parts of discussion? Maybe there could be some sort
of scheduling mechanism, so that someone could say "discuss this solution
with the creators at 10 am on Tuesday" and a log of the discussion could
be kept with the posting. Panizzi, Ed Summers' bot, could probably be
convinced to do something like this.

TPMDP #4: Some sort of presentation tools are always an option. Most
projects have a point where you want to get some feedback before you
proceed further or you suspect a quick reality check might save you grief
later on. Again, this would have to be extremely low barrier, but it would
be invaluable in understanding the information flow and final direction
within a project.

TPMDP #5: A spin-off "traditional" publication would still be possible
downstream. A lot of people only want to find out more about the innards
of a project after seeing its final results, so nothing precludes a more
neatly packaged version at some point in the process and, in fact, this is
where a lot of this would lead. But they key to me is that the messy
"bits" would also be available if there was interest.