I didn't worry about whether you can be said to "print" on a screen. Wrt to ethymology (but I have to check this), I think "print" goes back to a word for forming characters or drawings with a pen, so then both CRT's and laser printers would be more in line with the original meaning that what a Gutenberg-style printing press is.
Anyway, the reason I think of "preprint archive" as a contradiction in terms is that "preprint" traditionally meant a version of the article which was circulated during the period of waiting before the article appeared in real print in the journal. By this definition it is intrinsically for temporary use, so why does it have to be archived?
I agree about the drawback of the FPA term, but I think it's good enough and couldn't think of anything better.
No disagreement here: the described criteria were mentioned as examples, and as you observe, the paper was written with a European frame of reference in mind. I agree with your last paragraph ("Another possibility..."), but we also need some criteria which can be used during the initial period and before such a system of "associates" has been set up.
As I see it, there are two problems that come into play here: (1) that the scientific literature is enormously large nowadays; (2) that FPA:s are likely to be organized along geographical lines rather than topic lines. (More about that below). Therefore, each FPA will publish articles over a very broad spectrum of topics. Somebody or something has to crank a sorting mill where papers from all the FPA:s are rearranged according to topic area.
Everything considered, I think one must expect the authors to help a little bit in that process. If someone publishes an article on causality and the frame problem in the FPA of the Cybernetics Institute in Minsk, which is a good place, but he doesn't tell anyone about the existence of the article, then authors elsewhere may rightfully complain that it was pretty tough to know that it existed. In principle, there are other alternatives, for example having a on-line classification system based on author-provided keywords, but I think it's fairly natural and has many advantages if the author herself sends a notice about the paper to one or more News Journals.
Notice that this shouldn't affect the date of appearance: it's the date of appearance in the FPA that counts for priority, but only on condition that the article has been sent to a major News Journal without undue delay. ("Major" may be defined e.g. along the lines of the ECCAI/AAAI/PRICAI trisection of the world).
I agree that "electronic" is of course a misnomer. But so what - a "post office" is no longer "along" (the road), we talk about "jobs" in UNIX although that term is really only appropriate in batch operating systems, we have "random access files" although a file should by definition be a set of things which become available sequentially, and what pressing does MIT Press do? The term "electronic" is so widely used that it's abbreviated ("E-mail"). I'd agree with ACM that the issue has already been lost.
The proposal is not to get rid of confidential reviewing altogether. It is to have a period where open reviews are invited (6 months, in the case of the ETAI) and to combine it with confidential reviewing leading up to "certification" (the term of the present paper) or "acceptance" (the term finally chosen for the ETAI). The confidential peer reviewers will of course take the open debate into account when they decide on acceptance, but they must also think by themselves.
I agree that one must have ways of dealing with this, but still I think it's important to have some outlet for authors who feel they haven't been properly understood - there may be some occasional Einsteins among them. First, of course, the electronic medium makes it possible to organize contributions with greater or smaller visibility, and a News Journal editor would not be advised to intersperse a large number of these complaint papers along with other materials. Maybe one would put them on a separate web page altogether. It might also be a good idea to have one "crank meta-article" per year, where someone goes through all the complaints of that year and picks out 5 percent of them that he'd encourage the rest of us to take a second look at. A write-in scheme might also be useful.
See above. But it doesn't have to have inspired any spontaneous reviews; all that's needed is that it has been posted, so that colleagues have had a fair chance to see it.
This policy is still in a formative stage; we have to listen to what people want. For the time being, I would be in favor of giving the author the last word as to whether the review is published or not, but to encourage the practice where at least the review summary is made public. The review summary should then either say that the paper is accepted or, if it is rejected, specify one or more of the possible reasons for rejection, namely:
Case 2 of the rejection criteria would also open the way for write-in by people who have concrete reasons why the paper is of significant value after all.
I don't see how that could happen. Such mini-papers or micro-papers would never reach acceptance because they're not considered of sufficient value. On the other hand, what would be the advantage for the author of even trying the scheme? Just counting the number of publication is falling out of fashion anyway. At the same time, the proposed system does not discourage long papers which integrate many results, as long as there are results. In particular, there is no reason why the long abstract (summary) can't contain a list of a number of results. Of course, only those results that make sense to the world are meaningful in the abstract - no reason to quote lemmas.
Another thing though is that discussion papers which analyze an issue but without giving crisp results don't fit well into this formal review scheme with its emphasis on identifying crisp results in the long abstract. Probably the conventional type of peer review is better adapted for them.
The date of publication in the FPA, for the reasons and with the constraint specified above.
I agree with you that the Transactions has no business telling people how to structure their article. The issues addressed in section 6.1 arise because of technical and practical issues having to do with the availability of the publications.
I think there are at two clear availability issues:
One effect of the separation between the long abstract and the main text is that then the long abstract can be conveyed in paper form, without undue costs, to researchers out of reach of the Internet. That will given them a grasp of what's going on, and a chance to request the full text of what they are really interested in. Furthermore, this ties in with the multimedia aspect - after getting the full text of the article, they will in principle have had access to the full result.
With respect to the "context" section (discussion about related work), the idea is not to exclude references to earlier work in the main "content" part of the paper. I just observe that many authors prefer to have a section towards the end of their paper where they discuss other people's work in the context of their own. I agree that this can be criticized as being "post festum", but people do it. Now that they (we) do, why not have an arrangement that that final "related work" section of the paper is more easily revisable than the other ones. Or, putting it differently, once you have a paper, it's going to be fairly stable over time, but you may also wish to attach to it a discussion about its ramification which is intended to change over time and to be updated from time to time. Once you make that decision, maybe some of the stuff that traditionally you would have put in the final "related work" section of the main paper can now better be placed in that time-variable "context" part.
Apart from the multi-media part, the three other proposed sections therefore have three different levels of permanence. The abstract is not supposed to change at all; if you change it then you have a new paper and a new timestamp. The "content" part is likely to change once or twice as the result of feedback from the reviewing processes (both open and confidential), but you would be suspicious if the author keeps changing it many times. Also, it would be important to timestamp and persistency-protect each generation of the "content" part, for obvious reasons. The "context" part, on the other hand, should be seen as a living object, and frequent updates would be encouraged rather than frowned upon.
I agree about libraries. However, in the immediate future many people will want to have a paper copy for many things; we should be able to offer them that service. Print-on-demand would be most appropriate.
No disagreement. Of course the main version of the article must be published by a separate entity and out of reach of the author; that's what First Publication Archives are for. However, it would be unreasonable to forbid the author to put his/her own paper on his/her own webpage structure, duplicating the copy in the FPA. The item in 6.3 is just intended to stipulate that in doing so, the author must not post a modified version of the article and claim that it is equal to the original one.
I interpret you as saying that the plea for politicians to be concerned about basic research, and not merely about hidden industry subsidies, applies elsewhere besides in the EU. That's fine, but it's your debate, not mine.
More generally, although some of our problems are the same, I think there are also important differences between the US and European situations. In particular, the AAAI has a very much stronger economy than the ECCAI, so it is more capable and likely to set up a continent-wide FPA than what the ECCAI is.
This memo was written for a more library-oriented audience, and one of first problems there (at least in our country) is that current definitions of "document" or "publication" are geared to paper-based publications. For example, when setting up Linköping University Electronic Press, one of the (apparently) nontrivial questions on the formal side was whether the E-Press was going to publish paper articles or electronic articles. In particular, we were told that we had to make that choice before applying for ISSN numbers, since one and the same ISSN number could not refer both to paper and electronic publications. This meant, in turn, that paper copies (printouts) of the electronic document would formally speaking be another work than the original, and would count as republication...
We decided to solve that by saying that formally, the E-Press publishes paper articles in a miniscule number of copies. Furthermore, it is set up to maintain an electronic copy of the paper article on-line, and to guarantee that the on-line copy is permanently equal to the paper copy in the archive. ISSN numbers are only obtained for the paper copy, not the electronic one, but presumably every printout on paper of the on-line file will be a copy of the paper original in the basement, so the ISSN number refers to it as well... It was in an attempt to proceed beyond this somewhat contorted situation that I wrote the paragraph you refer to.
Because there is already a tradition for nation-states to maintain an archive of what has been printed in the country, and this archive could easily be generalized to the electronic case. In Sweden, for example, the requirement that every printed work must be delivered in a few copies to the national archive dates back to the mid-17'th century. During the 30 years war, the government had realized the importance of using printing for religious propaganda, so they wanted to gain full control of this important new medium. Napoleon introduced the same tradition in France, and I think most European countries have a similar system. Doesn't the Library of Congress do the same function in the U.S.?
It might be argued that if a country wants to do this for electronic publications, it's their business but it doesn't affect us. However, I think we would in fact benefit from it, because if it's done in the right way it would provide one more way of checking that FPA published articles are not tampered with, so it would add to the credibility of the whole system.
I agree that it would be foolish to have some kind of control system for this rule. It would rather be a recommendation to authors and expectation from the community.
Note that this ties in with the proposed rule that only those results count that are specified in the (long) abstract. If that rule is adopted, it means that any change in the body of the text that fortifies the case for the abstract is allowed and meaningful, but changes that add more results besides those stated in the abstract are pointless since they don't count anyway.
Of course not. The formal reviewers (confidential peer review) must make up their own minds, and there is no reason why they should have to await the end of the exchange. What is important to say has probably been said after the first few rounds, anyway.
See my answer to your point re p. 5 in this paper.