Tim Let me put it simply. A true copy is as good as a link. Not better, but the same. But a true copy is difficult to ensure.
The odds are significant that the "copy" is not true. · Let us assume that the copying process makes a copy of the file to which the link refers. · It may not collect the actual full-text but to a route on the way. In this case the copy is a serious distraction and its links may well be broken and fail. · It may be a partial copy to the full text or a copy of something completely different. For example suppose the link is to an html article. First Monday is an example Gold refereed journal which uses this style. The copy process retrieves the raw text file (hopefully, if it is not in a frame, otherwise it retrieves just the external frame code) complete with hyperlinks to all the images, diagrams, charts, audio, or video. Pretty likely the links in the copy are broken. The reader is presented with a stupid and unsatisfying copy. He or she wants the real thing, so either uses the link in the metadata (if there is one which I assert should be mandatory) or has to search for the article using a search engine. Or forget about it. · In other cases of which I am aware, the end document is constructed of a table of contents, with dependent hyperlinked separate chapter files. The copy process now retrieves the table of contents, with probably broken links. · In yet other cases the end article is protected by a cover page with a CAPTCHA. The automated copy process fails to make a copy. · I defer talking about ensuring that you are trying to make a copy of an actual thing. To forestall Stevan’s inevitable response that this is all unnecessary because the author can just deposit the accepted manuscript, what he has failed to take into account is: · The author may not have an AM. Quite often a journal will ask me to not integrate everything into a Word file, but to provide the raw text (preferably unformatted and looking pretty crummy) with markers where I would like the figures and diagrams to go, and to provide the non-text items as a set of separate files. This means that a requirement to deposit the AM may involve the author in extra work, to suit a Stevan-inspired mandate. (Or if he/she does not do the work, the reader gets a collection of files, poorly presented.) · If the author has chosen to publish in an open access outlet and wear the cost (if any, quite often none), they will want people to see the Version of Record, not a preliminary version. The situation is totally dissimilar to that of publication in a subscription journal where the VoR is hidden behind a price barrier, and access even to a preliminary version copy (the AM) is better than none. In the already OA case the compromise does not have to be made and shouldn’t. I return to the point: copying an already open access document is a waste of the author's time if they have to disassemble a complete version to put into the repository. It is possible that it may not even be feasible to incorporate into a single file. Believe me, I have met this situation. It is not hypothetical. We are also likely to waste the reader's time as well as the author's. And we introduce the possibility of error and the temptation to fraud into open access. I don’t think we need either. I think that you are falling in the trap of thinking that every journal article is produced as a Word file or a pdf. They aren't. Even less will they be as Internet usage continues to expand into academic territory. Best wishes Arthur -----Original Message----- From: goal-boun...@eprints.org [mailto:goal-boun...@eprints.org] On Behalf Of Tim Brody Sent: Thursday, 21 March 2013 8:35 PM To: Global Open Access List (Successor of AmSci) Subject: [GOAL] Re: Harnad Comments on Proposed HEFCE/REF Green Open Access Mandate Hi Arthur, I don't understand how a link is more useful than a copy (although obviously having both is preferable)? Let us say that either a) an author imports a record from a publisher with link or b) pastes a link into the repository. Either way, that link tells us nothing about the state of the item on the publisher's web site (gold or green). As you say, you could hit a jump-off page, robots challenge or otherwise. In order to say for certain that the link given is as the metadata describes, and that the item is available under the correct license, someone will have to visit the publisher's site (from a public IP) and set a flag to say 'verified'. By comparison, taking a copy is little extra effort and the institution can say unambiguously that they have an open access copy. It also has the added benefit of guaranteeing long-term access to that work. After all, that is what libraries have been doing for hundreds of years with paper. -- All the best, Tim. On Wed, 2013-03-20 at 08:54 +1100, Arthur Sale wrote: > Thanks Tim. No I don't think I missed the point. > > I agree that certification of all *repository* documents for REF (or > in our case ERA) is the same, whether the source is from a > subscription or an open access source. The point is that repository > documents are divorced from the source, and are therefore suspect. > Researchers are as human as everyone else, whether by error or fraud. > However, Gold is slightly easier to certify (see next para), even > leaving aside the probability that the institution may not subscribe > to all (non-OA) journals or conference proceedings. > > One of the reasons I argue that the ARC policy of requiring a link to > OA (aka Gold) journal articles (rather than taking a copy) is that one > compliance step is removed. The link provides access to the VoR at its > canonical source, and there can be no argument about that. Taking a > copy inserts the necessity of verifying that the copy is in fact what > it purports to be, and relying on the institution's certification. > > May I strongly urge that EPrints, if given a URL to an off-site > journal article, at the very least *inserts* the URL (or other > identifier) into a "canonic source link" piece of metadata, whether or > not it bothers about making a copy (which function should be able to > be suppressed by the repository administrator as a repository-wide > option). > > One of the problems that the "take-a-copy" crowd ignore, is that the > link to a Gold article might in fact not be direct to the actual VoR, > but to a guardian "cover page". This cover page might contain > publisher advertising or licence information before the actual link, > or it might require one to comply with free registration maybe even > with a CAPTCHA. It may be protected with a robots.txt file. No matter, > the article is still open access, even though repository software may > not be able to access it. (Drawn to my attention by private > correspondence from Petr Knoth.) > > Arthur Sale > > -----Original Message----- > From: <mailto:goal-boun...@eprints.org> goal-boun...@eprints.org [ > <mailto:goal-boun...@eprints.org> mailto:goal-boun...@eprints.org] On > Behalf Of Tim Brody > Sent: Tuesday, 19 March 2013 9:19 PM > To: Global Open Access List (Successor of AmSci) > Subject: [GOAL] Re: Harnad Comments on Proposed HEFCE/REF Green Open > Access Mandate > > Hi Arthur, > > I think you missed the point I was trying to make. The statement I was > responding to was that gold includes everything you need to audit against > (UK) funder compliance and "the same can not be said for Green". > > I have no wish to debate the merits of gold vs. green, beyond pointing out > that publisher-provided open access is no easier to audit than > institution-provided open access. Indeed, if institutions are doing the > reporting (as they will in the UK) an OA copy in the repository is easier to > report on than a copy held only at the publisher. > > I don't know where Graham got the idea that gold will make auditing easier. > Whether the publisher provides an OA copy or the author, all the points you > make apply equally. > > -- > All the best, > Tim. > > On Tue, 2013-03-19 at 08:40 +1100, Arthur Sale wrote: > > Tim, you oversimplify the auditing of green. Try this instead, which is > > more realistic. > > For green, an institution needs to: > > > > 1) Require the author uploads a file. Timestamp the instant of upload. > > > > (1A) Check that the file gives a citation of a journal or conference > > published article, and that the author is indeed listed as a co-author. You > > might assume this, but not for auditing. EPrints can check this. > > > > (1B) Check that the refereeing policy of the journal or conference complies > > with the funder policy. This is absolutely essential. There are > > non-compliant examples of journals and conferences. More difficult to do > > with EPrints, but possible for most. > > > > (1C) Check that the file is a version (AM or VoR) of the cited published > > article. This requires as a bare minimum checking the author list and the > > title from the website metadata, but for rigorous compliance the > > institution needs to be able to download the VoR for comparison (ie have a > > subscription or equivalent database access). [In Australia we do spot > > checks, as adequate to minimize fraud. Somewhat like a police radar speed > > gun.] [Google Scholar does similar checks on pdfs it finds.] EPrints > > probably can't help. > > > > 2) Make it public after embargo. In other words enforce a compulsory upper > > limit on embargos, starting from the date of upload of uncertain provenance > > (see 3). EPrints can do this. > > > > 3) Depending on the importance of dates, check that the upload date of the > > file is no later than the publication date. The acceptance date is > > unknowable by the institution (usually printed on publication in the VoR, > > but not always), and then requires step 1C to determine after the event. > > Doubtful that EPrints can do this. > > > > 4) Require every potential author to certify that they have uploaded every > > REF-relevant publication they have produced. Outside EPrints > > responsibility, apart from producing lists on demand for certification. > > > > I just adapted this from your constraints on gold, and common Australian > > practice in the ERA and HERDC, which have long been audited. > > > > Arthur Sale > > > > -----Original Message----- > > From: <mailto:goal-boun...@eprints.org> goal-boun...@eprints.org [ > > <mailto:goal-boun...@eprints.org> mailto:goal-boun...@eprints.org] On > > Behalf Of Tim Brody > > Sent: Monday, 18 March 2013 8:45 PM > > To: Global Open Access List (Successor of AmSci) > > Subject: [GOAL] Re: Harnad Comments on Proposed HEFCE/REF Green Open > > Access Mandate > > > > On Sat, 2013-03-16 at 08:05 -0400, Stevan Harnad wrote: > > > On Sat, Mar 16, 2013 at 5:14 AM, Graham Triggs > > > < <mailto:grahamtri...@gmail.com> grahamtri...@gmail.com> wrote: > > > > > > > > > > > 2) By definition, everything that you require to audit Gold is > > > open, baked into the publication process, and independent of > > > who is being audited. The same can not be said for Green. > > > > RCUK and HEFCE will require institutions to report on, respectively, the > > APC fund and REF return. > > > > For gold, an institution needs to: > > > > 1) Determine whether the journal policy complies with the funder policy. > > > > 2) Run an internal financial process to budget for and pay out the APC. > > > > 3) Check whether the item was (i) published (ii) published under the > > correct license. > > > > 4) (For REF) take a copy of the published version. > > > > For green, an institution needs to: > > > > 1) Require the author uploads a version. > > > > 2) Make it public after embargo. > > > > > > So, actually I think green is easier to audit than gold. Even if it were as > > you say, it will still be the institution that is tasked with auditing. For > > most institutions that will be done through their repository (or > > cris-equivalent). It therefore follows that green (Do I have a public > > copy?) will be no more difficult than gold (Do I have a publisher CC-BY > > copy?). > > > > (Commercial interest - as EPrints we have built tools to make the > > REF return and are working on systems to audit gold and green for > > RCUK > > compliance.) > > > > -- > > All the best, > > Tim > > > > > > > > _______________________________________________ > > GOAL mailing list > > <mailto:GOAL@eprints.org> GOAL@eprints.org > > <http://mailman.ecs.soton.ac.uk/mailman/listinfo/goal> > > http://mailman.ecs.soton.ac.uk/mailman/listinfo/goal > > > > > _______________________________________________ > GOAL mailing list > <mailto:GOAL@eprints.org> GOAL@eprints.org > <http://mailman.ecs.soton.ac.uk/mailman/listinfo/goal> > http://mailman.ecs.soton.ac.uk/mailman/listinfo/goal
_______________________________________________ GOAL mailing list GOAL@eprints.org http://mailman.ecs.soton.ac.uk/mailman/listinfo/goal
