[uf-discuss] be it resolved: irc chat logs deserve their own microformat

[Christopher St John]

I'd like to have a microformat for IRC chat logs. People publish
them on the web with alarming regularity in a wide variety of
formats, but it would be useful if they were consistently
represented so that basic info (what said what when, and who
entered and left the conversation at what point) could be more
easily extracted.

Specifically, I'd like to be able to correlate, in space and time,
many real-time events as captured on the web so that they
can be played back later or observed from afar as a consistent
set. IRC logs (presumably the backchannel of a conference
session, but could be anything) are one thing it would be
useful to capture.

One option (documented on the chat-examples/formats pages)
is to auto-generate both a human-readable version of the log
and a separate machine readable (rdf or xml) version. I won't
discuss that option further :-)

There has been some suggestion that IRC chat logs are part
of a more general idea about "conversations" that may or may
not include things like plays, skits and scripts, or possibly
non-IRC-like conversations recorded on the web. I think they
are not, and would like to see IRC-like chat logs get their own
microformat.

Plays, skits and scripts aren't records of conversations, they're
instructions to perform particular actions. Although the actions
obey a general ordering, it's expected that some of them take
place simultaneously and even in cases where the timing is
explicitly specified, it's generally aproximate ("at 6m 32s into
Act 2 Scene 3, Horatio enters" seems strange) Scripts for plays
are a bigger, more complex problem than IRC chat logs.

Non-IRC conversations[1] are potentially much closer. More
examples are probably needed, but I'll observe that there
are no timestamps, and no enter/leave messages. I suppose
it's a judgement call, but it seems to me that adding optional
elements like that ("just leave out the timestamps and some
other stuff") makes the format more complex, less specific,
and harder to parse and author.

But hey, I got a point of view.

It's certainly reasonable to continue to look for real world
examples, but it would help to agree at least aproximately
on the scope.

-cks


[1] http://kitta.net/2006/02/01/i-am-heartbroken/

--
Christopher St. John
http://artofsystems.blogspot.com