Some RSS readers do an rss:author to email:from mapping, so we should think carefully before allowing anything other than one author. Some weblog entries are co-written, but usually this is clear from context or content. In journalism the common practice is to have one author and zero-or-more contributors. Similarly, we might want one person to take responsibility for a weblog entry (be the author) and have the rest be credited as contributors in the metadata and probably also in the content. Some weblog entries are generated automatically, but even these can be considered to have a program as an author.
Please don't limit your brainstorming with concerns that will break an existing tool.
Place your name by the option you like best: [OpenPoll]
Zero or more authors: CraigEwert, BruceLoebrich, KenMacLeod, ClayShirky, DannyAyers, MikeWarot, StephenFarrell
Exactly one author: SamRuby, BillKearney, MarkPilgrim, AaronSw, JoeMadia, GregReinacker, TimBray, JoeGregorio, TimothyAppnel, MishaDynin, JakeSutton, JeremyGray, TomasJogin
One or more authors: JamesSnell, NickChalko, KevinMcCurley
Define author first: ShelleyPowers, AsbjornUlsberg
Standing aside (see TrueConsensus):
[MarkPilgrim] Do we have concrete examples of either of these?
[AaronSw] RaphLevien and I co-wrote a weblog entry once. However, I doubt having the format be able to describe this would be very useful since a) we clearly stated this in the content of the entry and b) neither of our content management (Advogato and Movable Type) systems can handle the concept of multiple authors for an entry.
[GaryF] Yes. My content management system can automate frontpage entries (a simple description, link and perhaps author of real content) for items added in categories not automatically listed on the frontpage. But then I would still consider the CMS as author of automated posts.
[AdinaLevin] A system as an author. An IT project blog where some entries are written by individuals discussing the project, and other entries are automatic posts by the CVS system logging code check-ins.
This page is an example. See the rationale below.
[BruceLoebrich] A significant portion of the posts on our weblog are co-authored with neither Katy or I being primarily responsible for the content. That tools force us to create an artificial Katy and Bruce persona to get around a single author per entry model is incredibly annoying.
[TedWhalen] Nearly every paper published in a scientific or academic journal has multiple authors. Echo ought to be able to syndicate online journals.
[SamRuby]: NewsGator presents RSS data in Outlook (a newsreader by MSFT). In the process, it maps author to From. I presume that HEP does likewise. IMHO, we should think very carefully about what the benefits are to having multiple distinct authors are before we close down this natural mapping.
[BillKearney]: if you need more than one author then have none and use dc:contributor for each of them. Then the readers would have to pick one. Perhaps additionally including an author entity that expresses those members as a group. In RDF this would be trivial to properly express.
[GeorgBauer]: I think this would be a sensible way. Trying to stay in the spirit of prior art and give ways to extend the format to accomplish new ideas. PyDS does create mailto links from author addresses in the aggregator. I always viewed it as "this address you send complaints to". Might be misinterpretion on my part, though.
[JoeGregorio]: Multiple authors of the same entry seems like a pretty rare occurence. Any lessons from Journalism on this?
[AaronSw] I've noticed that newspapers usually list one author and state at the bottom "Johnny X contributed to this report." I don't recall seeing a newspaper article with multiple authors. It seems like the same thing should be true of weblogs: one person should take responsibility to the post; other people can have contributed.
[JakeSutton] This seems like the perfect solution. Besides I can't imagine a CMS that allows multiple users to simultaneously create content. The creator is the author; others are contributors, I'd think.
[ClayShirky] The lessons from journalism are that ink creates no semantic harness: The NY Times supports no byline, one byline, and multiple bylines.
Perhaps more to the point, why should we limit ourselves to journalistic practice? Weblogs are a flexible medium, of which the journalistic pattern is just one piece. The web is good for writing manifestos, which are often written by groups operating under an umbrella identity (e.g. Fluxus, the Futurists, and so on.), weblogs are good for publishing papers, where the names of many contributors are listed, but where order becomes important (a paper by Gould and Lloyd is different than a paper by Lloyd and Gould), weblogs are good for publishing menus (12" pizza $9.95, 14" pizza $11.95) which have no authors at all.
Given that the semantic nature of "who wrote what" is unfixable in a medium this flexible, we can either say "One URI per author, multiple URIs OK, order should not be munged by software" and let the reciever interpret it, or we should say "0 or 1 authors only, multiple authors can be handled by making the Author URI in turn point to multiple other URIs."
[CraigEwert]: A general format for logging interchange shouldn't impose stylistic norms ("...one persone should take responsibility...") and shouldn't be bound by current CMS capabilities ("...neither ... can handle the concept of multiple authors..."). I've made posts with two coequal authors http://infinityzero.com/?ID=64 (note that it only lists me, because of software limits). Why not just let it be so; it's not so hard to specify or process (and the processor will have to be prepared for missing or duplicated 'author' anyhow, cause your feeds are untrustworthy).
[DonPark]: Does the number of authors matter when there are no means to count reliably?
[BryantDurrell]: Since email addresses aren't constrained to a 1:1 address:person mapping, I'm not entirely sure there's a dichotomy.
There are two separate issues here:
a social one, who is speaking? This should not be constrained. It can be an individual, a group, anonymous, an agent....
a technical one, how to uniquely identify the author for purposes like authenication and threading. See Identity
The same URI may addres both issues. The URI will often be a mailto: (or based on a unique hash from a mailto). But it can also be a personal web page, group web page, or even, as suggested in WikiPageAsEntry, the entry itself.
[ArveBersvendsen, RefactorOk] In a publishing system where the roles and workflow is important, you could for instance have one creator, multiple co-authors, and one person responsible for approving moving the entry from draft to publish.
[AsbjornUlsberg] We definately need multiple authors, but we first need to define what an author is. Also, we need to specify the mechanism and semantics of adding an author to both feeds and entries. I think <contributor> is the best name for the element, but if we must, we can possibly have an <author> element as well. An alternative is to specify what role the contributor had in the current entry with a "role" attribute on <contributor>. I think the latter is more clean.
[KevinMcCurley] I've put together three feeds for scientific literature. More than half of all scientific articles have multiple authors, so it's hardly a rare case. Without this functionality, Atom/RSS will be relegated to a very narrow class of written material, and to insist on a single author is simply unrealistic.
(comments regarding anonymous wiki authors and entries moved to WikiPageAsEntry)