At 22:54 03/12/2002, you wrote:
I think deletion should be a (discouraged but available) option, but with a persistent tag for the deleted (null) text, as a place-holder for would-be citers who did read that draft and do want to refer to it (even against the author's request, and even backed up only by hearsay). The best corrective for this is to self-archive the updated, corrected draft too, and have the archive pointers always point to that too, even the null (deleted) ones. (But I really do think that if someone's draft is so raw that they would rather it could vanish without a trace, then perhaps it was too early a draft to post publicly in the first place, and should only have been emailed to a few trausted colleagues!)
I'm glad this is getting an airing. Can I make a proposition, then (essentially what came up earlier): 1. All archived copies are citable and users MUST be aware of that 2. (to reinforce this) Only upon a properly reasoned and authenticated application will archived copies be withdrawn from access (and only on the requested server). There will be a risk that references will fail if such copies are withdrawn. Users will then adapt their behaviour to the rules. I suggest that this will encourage the desired behaviour and disuade 'half baked' material being archived. There was some merit in the old paper-based process - time to think and revise before 'publication' i.e. uncontrolled dispersal to the total community, rather than controlled dispersal. This assumes that any archived copy is published. To allow the 'email to trusted colleagues' model you could have an archived copy that is password protected - much as the review process in e-journals. However, this does compromise the principle of an archive, which I think is that everything should be publicly accessible. It could be that the early drafts may become public after a gracetime. They may well be of interest to the history of science one day. A/
