Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!gem.mps.ohio-state.edu!ginosko!uunet!philmtl!ray From: ray@philmtl.philips.ca (Raymond Dunn) Newsgroups: news.admin Subject: Re: Tired of bogus subject lines? Message-ID: <699@philmtl.philips.ca> Date: 12 Sep 89 20:54:22 GMT References: <1650@unocss.UUCP> <112@blekko.UUCP> <6120@ficc.uu.net> Reply-To: ray@philmtl.philips.ca (Raymond Dunn) Organization: Philips Electronics Ltd. - St. Laurent P.Q., Canada Lines: 54 In article <6120@ficc.uu.net> peter@ficc.uu.net (Peter da Silva) writes: >In article <112@blekko.UUCP>, skrenta@blekko.UUCP (Rich Skrenta) writes: >> It's not rn-bigotry. It's rn, notes, or any newsreader that tries to follow >> threads based on the subject line. > >Following threads based on subject line is wrong. Whether or not it's "wrong", it's certainly no justification for arbitrarily making stylistic ad-hoc "improvements" to an agreed standard that causes much of the installed base to beak. Anyway, hasn't all that been covered before? Isn't the fix action already under way? > Subject lines are user- >edited, and change continually. They very rarely have much to do with the >subject, outside of moderated groups. A newsreader that depends in them >is broken. Look, the main problem with the Subject line is *not* that posters change it unnecessarily, but that they *dont* change it when they go off at a tangent. To say that they *rarely* have much to with the subject is just plain silly. In fact most of the time, even though there can be several sub-threads intertwining, the subject line is very relevant. Following the references line rather than the subject line is no panacea. It is not adequate if threads intertwine and wander back onto their original course as they often do, and even then it depends on posters not modifying the references line, an action specifically *allowed* in the spec. I've been mentally playing with an algorithm that would follow references, and frankly, it seems to always require much ad-hoc branching on little more than guesses. It is not clear whether depth first or breadth first scanning of the references tree is generally appropriate. In particular, branches will re-join without any immediately connecting references, as posters reply to several responses at the previous "level" together. Thus the thread must be followed not by the last reference, but by some earlier reference in the chain. Following the Subject line is just an extreme example of that, and is equivalent to following the original article's reference. When looking at the whole tree in hindsight it is fairly clear which paths should have been followed, but only because of their *content*, not their references, nor their subject, nor their originator!! Has anyone done any definitive work on this? Have any thread following algorithms already been suggested and tried out? -- Ray Dunn. | UUCP: ..!uunet!philmtl!ray Philips Electronics Ltd. | TEL : (514) 744-8200 Ext: 2347 600 Dr Frederik Philips Blvd | FAX : (514) 744-6455 St Laurent. Quebec. H4M 2S9 | TLX : 05-824090