ECCC as a zero-formatting “publisher” for CCC proceedings?

Background: After going solo, the CCC conference is using LIPIcs as a “publisher” for the papers accepted to the conference. This involves a non-trivial amount of formatting (to put the papers in their format) and also some monetary costs.

I would like to use the opportunity that CCC is going solo to move to a model where the “publishing” involves *zero* effort from authors. This could be a selling point for the conference, and maybe set an example for others.

Specifically, in the vein of previous posts, I propose that authors of accepted papers simply send the .pdf of their paper in whatever format they like. The CCC people take care of placing a stamp “CCC 20xx camera-ready” and putting the paper on the ECCC. Papers with indecent formatting are treated exactly as papers with indecent introductions.

Disclaimer: although I am on the reviewing board of ECCC I had no discussions with the ECCC people about this.

The main benefits of ECCC are:

– Submission is painless: just send the .pdf! Again, authors can write their paper in whatever format they like.

– Indexed by DBLP

– It’s run by “us”, it’s about computational complexity and in fact “Under the auspices of the Computational Complexity Foundation (CCF)”

– It has an ISSN number (1433-8092). I am told this is important for some institutions, though I don’t know if some insist on ISBN over ISSN. If they do, perhaps there’s a way to get that too?

– They do various nice things already, like archiving papers in CD’s etc. In fact, going back to the ISBN issue, couldn’t we simply assign an ISBN to the reports from each year?

– It has no cost (given that ECCC already exists).

Another option is to use arxiv or an arxiv overlay. This would also be better than using LIPIcs, I think, but it does not enjoy many of the benefits above.

Advertisements

78 thoughts on “ECCC as a zero-formatting “publisher” for CCC proceedings?

  1. That sounds enticing, but… sending the .tex source has another effect, I reckon: ensuring the durability of the papers. I am not too aware of how proprietary the pdf format is, but definitely less than a text-based .tex file than we can (hopefully) still compile or read at any point in the future. Archiving the sources may still be a good idea, in that regard.

    1. Thanks for the comment. Yes keeping the source file might be a good idea, that’s what arxiv does too, but good luck recompiling it without forcing the authors to change formats. Also some people are moving away from latex. Personally I feel that the saving that comes from not imposing formats on authors outweighs the chance that in the future we won’t be able to read old formats.

  2. When we (Christoph Meinel and I) discussed the “ECCC idea” in early 1994, we first asked many complexity people “what site do we actually need” (if any). The feedback expanded from “journal” to “repository”. At the end we landed at Tech. Reports. Well, even TRs have some “format”. So, we made one. But, as I remember, it was Noam Nisan who said something like this (very imprecise after > 30 years): “We want see results, not just publish them. So, let’s make the authors life as easy as possible”. And so was this.

    To “TeX fans”: ArXiv thinks these preprints will be relevant perhaps 50-100 years ahead, or even longer. Show me please a single preprint, which was relevant > 10-20 years ahead (without being “stamped” by some journal). TCS conferences give us platforms to quickly spread up newest findings. They require formatting. ECCC also spreads “hottest news” without any formatting requirements. Readers decide …

    The “ISSN and ISBN issue”: I think ECCC needs none of them. Papers appearing here are neither “real papers” (only technical reports) nor their year-collections can pretend to “single books”. Hiring committees treat ECCC publications as only tech. reps., and this is good so: ECCC was made for working researchers, not for bureaucrats.

  3. A technical problem (which is orthogonal to your main point) is that ECCC is not very good. It is not nearly as robust and reliable as the arxiv, and because of its desire to “brand” every paper with the ECCC logo in a rather unsophisticated way, it converts every PDF file into an unwieldy mess. I have often lamented the existence of ECCC because sometimes authors post there in lieu of the arxiv.

    1. I think certain aspects of the interface could be improved, yes. But overall I think ECCC is good enough. I haven’t thought about the problem you mentioned, maybe it’s something that can be fixed without too much effort. (But why do you care? Does it make the files too big?) The main thing that I don’t like about the arxiv is that submission is more painful (i.e., if you are using latex you are required to get the source to compile on their machine). (Is there a way to trick arxiv into accepting .pdf generated from latex without submitting a source code? Even if there is presumably it wouldn’t be a stable trick.)

      But as I mentioned in the post arxiv would still be better than LIPIcs, I think. (Because the standard is much more basic and widespread.)

    2. (Also, I clearly should have been more specific: IMHO, ECCC “is not very good” at hosting papers. The editorial board and content is great.)

  4. Dear James, could you elaborate the problem with “ECCC files” a bit more? Namely, why “it converts every PDF file into an unwieldy mess” (I can read any of these papers without any problems). And also, why “ECCC is not nearly as robust and reliable as the arxiv”. If this is a (real) problem (which ECCC local board is not aware of), things will be improved. For this – we need witnesses.

    B.t.w. I agree that the way “ECCC stamps” are put onto submisions is rather unsophisticated: it is done by a robot created 30 years ago … any new ideas to improve the process would be more than welcome. The main idea of ECCC (procedure) is: submit -> let people red it, instead of: formate -> submit -> wait for a “stamp” (referee reports) -> read it. The difference between ECCC and ArXiv is “boutique” vs. “shopping center”.

    1. Arxiv is more like an online store than a shopping center – if you don’t know what you want, it can be hard to find things if they were posted a while ago. A lot of this is the volume, but it also is the uncurated junk mixed in. The Theory Blog Aggregator has improved the situation for the daily upates.

  5. A significant portion of the cost for LipiCS is the cost of DOI numbers. (LipiCS still has to pay for the DOI service – much less than ACM or a commercial publisher would – and I think that the amount CCC pays LipiCS is actually comparable to the amount they pay out for DOI numbers.)

  6. I agree with James.

    @Stasys: if you want concrete problems, consider the following.

    1) The ECCC submission process mangles hyperlinks in pdfs. I’ve found that no hyperlink in an ECCC pdf will work if it is on the first page (where the ECCC stamp is), and hyperlinks afterwards work, but not always. The “suggested latex” for addressing this problem was not helpful the last time I tried addressing this problem (I believe it required a postscript intermediary compilation step, which I find undesirable).

    2) I’ve never got the ECCC submission process to give me a preview of my submission (the arXiv has this). I want the preview to see how badly my hyperlinks are broken.

    3) Revisions to papers do not generate new RSS feed items (which is how I consume the ECCC). There was one paper I recall whose initial posting was somewhat interesting, but upon revision was vastly improved. I missed this revision because a new RSS item was not posted. In contrast, the arXiv yields new posts for updated papers.

    4) The “view older reports” only lists 3 papers at a time, why not list more? Or make the # per page adjustable?

    5) My understanding is that ECCC does not allow to update papers with a revision that includes more authors. I heard this from Swastik Kopparty, in reference to why https://eccc.weizmann.ac.il/report/2012/106/ has to be supplanted by the second paper https://eccc.weizmann.ac.il/report/2012/149/ .

    6) Authors get their own ID’s, but various authors (eg, Swastik Kopparty, https://eccc.weizmann.ac.il/author/541/, https://eccc.weizmann.ac.il/author/147/) have two. I do not know the reason for this, but can we please merge these duplicates?

    7) The HTTP headers on ECCC downloads are somehow mangled. If you click on a PDF-link on ECCC it will not open in the browser-based PDF viewer even if that is the default setting for PDF’s. This is contrast to every other website I’ve seen, and nothing I’ve done with my browser fixes this.

    8) The search function is badly broken. Consider the search for ‘swastik kopparty’ (https://eccc.weizmann.ac.il/search/?search=swastik+kopparty) which yields nothing. In contrast, if you search ‘kopparty’ (https://eccc.weizmann.ac.il/search/?search=kopparty) you get a list of Swastik’s papers, but they are not sorted by date (nor sorted by any logic I see).

    9) From what I recall I cannot update my name from ‘Michael Forbes’ to ‘Michael A. Forbes’, as user profiles do not support updating your name.

    While the ECCC may be “good enough” for some, I feel that the above problems demonstrate that insufficient attention is being paid to its technical development.

    1. This is a very useful list! Hopefully many of these things can be fixed soon. It seems the conversation has somewhat shifted from “ECCC as a publisher of CCC proceedings” to “ECCC”. It may be good to compare LIPIcs and ECCC.

      1. Just a clarification for Problem 1): those hyperlinks that doesn’t work outside the first page, are precisely the hyperlinks that link to the last page of the paper, where another ECCC stamp is put.
        Hence, the problem is with the stamps. Adding a title page instead of stamping pages will solve the hyperlinks problem, assuming one follows the specific PDF compilation guide on the submission page (which by itself is cumbersome, admittedly).

  7. @Michael: thanks for pointing to these issues! Well, hyperlinks and similar is not so important (for a real work) – the main goal is to see the result and its proof. But some issues you mentioned are indeed important. In particular:

    Issue 2: no preview. Should be definitely realized.

    Item 4: view older reports. This is disturbing also to me. In the embryo stage of ECCC (when I also was “at a button”) *all* reports of the current year were listed on one page.

    Item 8: search engine should be definitely improved.

    I only want to remember: ECCC was made to “see the happening in complexity”, not to have “well technically prepared presentations”. Here is a difference.

    Recently, ECCC went to Weizmann, Oded Goldreich has taken care about it. He already has a long “to do” list. We (Emanuele and I) will probably add your items to his list.

    P.S. I am extremely disturbed by one thing: does CCC must pay money for their papers being published at LIPIcs? If so – this would make all my comments above superfluous …

    1. I think hyperlinks should and can be fixed. Who cares about the logo anyway? It’s useless and apparently damaging. Re paying: yes lipics charge also because they get doi and isbn as Paul said.

  8. @Paul: yes, indeed, it’s the volume. When I come to “axiv cs computational complexity” – I can only access “new”, “recent” and “current month”. Everything 1 month earlier is hidden. So, go to https://arxiv.org/find and try your luck …

    Actually, the more sites we have (to “publish” our findings) – the better. I am only in concern when somebody wants this being paid (from our tax payers). We already run our communication of results “for free”. Why we then need Elsevier, Springer and other “fat guys” to communicate our results within our community?

  9. @Manu: it is not just the logo forcing to pre-process the file at ECCC: it is mainly to give the report number/year stamp. Hyperlinks can disappear during the pdf -> ps -> pdf transition, but Iddo Tzameret gave a quick and simple fix here:

    http://andysresearch.blogspot.de/2010/11/eccc-what-authors-should-know.html

    1. Use the following packages and options in the preamble of the tex file:
    \usepackage{color}
    \usepackage[dvips,pdfstartview=FitH,pdfpagemode=None,
    colorlinks=true,citecolor=blue,linkcolor=blue]{hyperref}
    2. Compile “name.tex” into a dvi file “name.dvi”.
    3. Run > dvips “name.dvi” -z -P pdf
    4. Run > ps2pdf “name.ps”

    (Of course, you can change the “blue” in the hyperref package options to a different color. I usually prefer black.) From his explanation: “As far as I understand, the point is in making sure that the ps file contains the hyper-link data. For this you use the “dvips” option in the hyperref package. And second, you compile the dvi into ps, with the “-z” option which tells the compiler to keep the hyperlinks data in the ps file. The “-P PDF” option makes sure that fonts are embedded in the ps file, so that the conversion from the ps to the pdf file would be good (but it’s possibly not always mandatory).”

    Unfortunately, I couldn’t find this fix being described at ECCC. If so, we should tell Oded.

    After this fix, ECCC will still remain a “zero formatting” site, albeit 0.001 preprocessing site …

    B.t.w. as I understand, Issue 2 (preview) is already implemented for at least 5-6 years. Is this still an issue?

    Re paying: well, if this is only for ISBN and doi numbers – its not a real paying for services, just buying (by CCC) external services . So, I misunderstood the situation.

    1. Still I don’t see why in 2017 we should bother changing the .pdf, especially given that it creates some problems. I think the ECCC website provides as good a timestamp as we can get. What would you think of simply putting online the .pdf the author sends, without any modification?

      1. Actually, this small “guide” explaining how to compile tex files into PDFs while retaining the hyperlinks, does appear on the ECCC submission website.

        One thing is that when I wrote this guide (and specifically when I first used it on ECCC), it adequately preserved all hyperlinks. But as of now, it messes the hyperlinks on the first and last pages of each manuscript, as noted by Michael. Hyperlinks on the first page are many times the most important ones (e.g., those appearing in the abstract).
        For some reason I can’t recover my initial success to overcome this problem. But, still, it is very helpful that hyperlinks are present in (most pages in) the PDFs.

        For me, the first-page hyperlinks problem is the main limitation of ECCC. I have not encountered the other problems Michael specified, but they may also be important to fix.

        On the other hand, in my opinion, one of LIPIcs format’s main disadvantages is its slightly dubious aesthetics: the use of Arial titles with Time New Roman body text, and the imposing yellow rectangles, look very awkward to me. But this is just my personal view.

        So overall, I think Emanuele’s suggestion is a good one.

      2. @Manu: I view the happening from readers side. So, CCC has
        published the list of accepted papers. (It is not seen at
        ECCC.) Where then a reader can find out that this
        (particular) paper was not just “screened at ECCC but also
        was refereed by 1-2 colleagues?

        We should remember: ECCC is a “just screened site” while CCC
        is a very solid conference (rejection at about 60%). Were I
        at CCC PC, I would never agree to “fuse” accepted papers
        among all this “ECCC screened stuff”. By no means.

      3. I am not sure I view this as a serious problem, because for example I am not sure how accurate this information is on the arxiv. But it is true that on the arxiv many papers have a comment indicating conference/journal publications, while this is rare on ECCC. For starters, it would be easy for CCC to add a comment to each paper noting that this is a CCC version, and maybe we can encourage others to follow suit.

        Best, Emanuele

  10. 1) Iddo’s fix is described in the ECCC website during the submission process. When I first started posting to ECCC I believe I tried this and found that hyperrefs were still somewhat broken, but I don’t recall for which paper I did this. At least one of Iddo’s more recent papers (https://eccc.weizmann.ac.il/report/2013/070/) still has hyperref problems.

    More at the issue is that this requires a dvi->ps->pdf compilation, which is outdated (and I believe(?) is not even available to some people using alternate compilation procedures, such as textmate). In particular, this makes its difficult to either embed PDF graphics or to use the full power of TikZ diagrams.

    2) There is a “preview” of the ECCC HTML page, but not a preview of the actual PDF posting. That is, the preview ECCC gives you has a download button, but it is not functional.

    3) I think it is important to have the ECCC name and report number, as well as the logo, on each paper. The crypto eprint server does not have such information, and if you print a paper from there you then need to remember which version you printed (and they often have many revisions). With the on-paper report number, it makes it clear where the paper was downloaded.

    4) Putting on the ECCC logo/number should not be this painful. If you just want to watermark the first page (there is also something on the last page, I’ll ignore that for now), you can use the pdftk tool (see for example https://stackoverflow.com/questions/501723/overlay-one-pdf-or-ps-file-on-top-of-another/509891#509891 ). I tried this out and it seems acceptable to me. On a sample document which is 125pg pdf that is 891KB, the resulting file is 1.2MB. Ideally there would be no blow-up, but this doesn’t seem so bad to me.

    That said, I feel it is worth remembering that the arXiv can do this watermarking with essentially no blow-up, precisely because they have the latex source.

    5) I personally support paying money to LIPIcs, as I view the DOI system as a valuable part of the archival process.

    1. Thanks for the comments! Regarding (3), papers usually have a date written already, so this is only useful in case you want to differentiate between that and the “ECCC date”. I can imagine this could actually arise in disputes, but I think those can be resolved just by checking the website. Having said that, I certainly don’t mind if an “ECCC date” is added, and yes, we should use a watermarking trick as you suggested. So I think we need a little program that creates a watermarking picture with the current date, and then stick it onto the pdf.

  11. @Manu: well, in your post you also wrote “The CCC people take care of placing a stamp “CCC 20xx camera-ready” and putting the paper on the ECCC”. So, you also agree that some “stamps” should be.

    I have a feeling, we have cardinally shifted away from your original question “LIPIcs or still ECCC?” to “technical problems of ECCC”. But the feedbacks show: “zero-formatting” must be improved.

    Re “why in 2017 we should bother changing the .pdf”: actually, in the retrospective, I am no more sure in the necessity these “stamps” being made by an “armed intervention of ECCC robot” into the PDF files themselves. This was (perhaps) useful when readers *printed* the files to read (to identify the source, within tons of other printed papers/manuscripts on the table). But now most of us read papers on a display. So, perhaps it would indeed suffice that ECCC just *adds* the first “title page” to the pdf file with all this information: authors(s), title, key words, tech. report number, year, ISSN, etc. (just like JSTOR is doing). All this can be automatically generated from the submission information. The rest then would be the original authors paper – no intervention.

    B.t.w. were accepted at CCC papers published in ECCC, these could not go as just “simple ECCC reports”: these were not just “screened” by the ECCC board, much more people were involved in the selection. So, in this case, *separate issues* of ECCC reports should be made. I would be happy to see CCC papers *separated* from “ECCC reports”. I mean: separate volumes for each CCC conference.

    1. Michael, yes, as I wrote above, the fix doesn’t seem to save the first and last pages’ hyperlinks, where ECCC’s logos occur. (When I used it initially, it did retain all hyperlinks, but I can’t recover this success now for some reason; probably because I updated some drivers.)

      As for your points 2, 3 and 4, I complete;y agree.

    2. I think we can go either with no stamp or with a stamp as long as it does not alter whatever the author sent beyond an unobtrusive watermark. Re: CCC proceedings on ECCC, I wouldn’t complicate things. I think the CCC website will already publish a list of accepted papers; this information doesn’t have to be repeated on ECCC.

  12. @Manu: actually, I am not very happy to having me again involved in this “no formatting” discussion. Wrote I 10-12 papers per year – this could be an issue. But wait! What if one (like me) writes 1-2 (max 3) papers per year? After spending months and month on the proofs – why is then so critical to spend 1-2 hours (in 1-2 days) on formatting? B.t.w. after this “formal formatting”, I often have seen my story (told to readers) in a slightly different light. And caught some mistakes during this process.

    Actually, I don’t like all this current “publishing culture”. In the former USSR it was enough to give a talk to colleagues. Formal publications not counted much. Just the fact “they know” was most important. Is our goal to produce tons of “papers”, or is it to DISCUSS our resuts?

    P.S. since “your comment is waiting for moderation” time on this blog is even longer than the screening at ECCC, I am sending also this comment (in a pocket).

    1. Stasys, I apologize about the wait. I was traveling, but now I changed the setting on the blog so that comments of trusted people don’t require moderation anymore, so it’s not going to be a problem.

  13. @Michael: I couldn’t test your Issue 2 (preview). [Haven’t submitted anything in the last time.] I rely on ECCC robot showing the ENTIRE submission (bibliographic information AND the entire submission). If this not the case – this must be corrected.

    Issue 5: paying for what? Can you give me any evidence that some 10-20 years old preprint was not “not in the system” (=journals ruled by fat guys). DOIs for tech. reports? Would be happy my doubts being wished up …

    1. Stasys, yes, adding a first title page AND ALSO moving the watermark of the final page to this title page, would solve most problems vis a vis hyperlinks. Though I would think that Michael’s suggestion above in https://emanueleviola.wordpress.com/2017/06/22/eccc-as-a-zero-formatting-publisher-for-ccc-proceedings/#comment-1119 is better, since it will work for every PDF, and not just those PDFs resulting from a specific compilation procedure, and would also look nicer, and save on number of pages.

  14. @Iddo: I am an absolute zero (not even an epsilon) concerning all these technical things. But I cannot understand, how an addition of one “title page” could depend on the way the original PDF was produced? Say, in Adobe Acrobat we have a possibility to insert pages. So, this could be also possible to be done purely automatically.

    I am very skeptical on using “TikZ and friends”: these tools are “too hidden” for normal users. I am looking for “most primitive” solutions.

    P.S. I am still in doubt whether the “formatting issue” is a “real issue” (for people writing only 1-3 papers per year).

    1. In the grand scheme of things “formatting” is not an issue, of course. But in everyday life I think it’s significant, especially in our community. Note that even if you write only 1-3 papers a year, each of those papers will require many submissions and reformatting. A typical life of a paper can involve this: Arxiv/ECCC/eprint submission, conference submission, conference re-submission, conference camera-ready, arxiv/ECCC/eprint update, journal submission, journal re-submission, journal camera ready, arxiv/eccc/eprint re-update. In many cases, things are even worse. If each of these steps requires a new formatting, this is a significant overhead even if you write 1 paper a year. If you multiply this by the entire community, I think we are wasting significant resources on something which gives us absolutely nothing.

  15. @Stasys, actually you are probably correct. If you only add a title page, without altering anything, or watermarking other pages, or doing some conversion that as of now seems to hinder the functionality of submitted PDFs, this should not be a problem.
    On the other hand, currently there seems to be some conversion done to the PDFs that breaks all hyperlinks, unless the pdf is compiled in the way I described in the “guide”.

    I think formatting issues become a problem not only for those authors actually doing the non-trivial formatting, but also for those authors who submit to ECCC without making the appropriate format changes, causing readers of manuscripts to be unable to use PDF hyperlinks (which, at least for me, is a considerable advantage, enabling more efficiently readable manuscripts).

  16. One issue with Emanuele’s suggestion is that ECCC requires full papers with full proofs, while CCC does not. Though practically most people do submit essentially full versions to CCC, this may still be something to consider.

    1. From the CCC ’17 call for papers:
      “Submissions should include proofs of all central claims.”
      In general I think every conference requires full proofs.

  17. @Manu: instead of a “real issue” I meant more a “critical issue”. It definitely *is* an issue when this “formatting” is extremely “complicated” (say, one indeed needs “good nerves” to make Wiley latex styles to work …). But my experience with Cambridge press, SIAM, Springer or Elsevier was not so “disastrous”: just take a template, adopt slightly the preamble (mainly theorem environments and the title page information), and you see the result. About 10-20 minutes. But, of course, if an author takes the *entire* road map you described, with each requiring re-formatting – then this is not in order.

    @Iddo: I (as Emanuele) could not quickly realize why this (requiring full proofs) could be a problem? The old obstacle (space requirements) are no more relevant for ECCC/ArXiv etc. Readers are “hunting” for proofs, not for results “out of the blue”. Perhaps you meant that ECCC could have an option of “telling ideas” to colleagues (key ideas are already there, detailed proofs should come). Actually, I am also missing this aspect of spreading ideas among colleagues. Some section of “announcements, extended abstracts” could perhaps be in order. Like it was/is in Doklady Akad. Nauk SSSR. The only difference is: these announcements in Doklady should undergo serious presentations in seminars (with full proofs of key results). ECCC doesn’t have this option.

    B.t.w. many (really many) years ago, there was a discussion within the ECCC board to implement an “open problems” section. There even were some attempts made. It’s a pity that this idea somehow “dead” …

    P.S. Actually, LIPIcs is a nice idea (just as Dagstuhl itself is). Its main advantage: it is free for readers AND for authors. People there are doing a very useful job. Not being mixed with all this “free publication” madness run currently by businessmans. But with “look nice, colorful” (instead of “look ascetic, but serious”) they have somewhat exaggerated, perhaps …

  18. @Manu and @Stasys, I was not implying there’s a specific problem in requiring full papers, I was merely observing, what, as far as I understand, is a small difference between CCC and ECCC: people can in principle do submit “Extended Abstracts” to CCC, while supposedly not to ECCC. Indeed, as Manu mentioned they would still need to “supplement” their initial CCC submission with full proofs of the “main results”, but maybe not of all results and also these proofs would not necessarily become public, unlike ECCC. But of course, I don’t imply that this should by itself detract from Emanuele’s proposal.

    1. Stasys. As I mentioned authors are required to submit full proofs to CCC. I never said authors need to supplement their initial CCC submission with full proofs. If you are referring to the “camera-ready” version, that’s not supposed to have extra proofs. If any of CCC or ECCC is required to have more proofs, that is *definitely* CCC, which is screened much more thoroughly than ECCC.

  19. @Iddo: can I share my experience after writing the “boolean complexity” book. The main problem with older FOCS/STOC/CCC proceedings (and with Doklady and other sources in Russian) was: “do it yourself “… So, I would wish the authors do their job, not the readers (authors job). Well, I cannot say that trying to prove things by myself was not a pleasure: in most cases completely reversed. Just like solving “exercises”. But nowadays we are flooded with tons of information. So, some focus on “authors: please do you job well” is perhaps not so irrelevant? Shortly: I am for FULL proofs. Be they not the most elegant, not for “THE Book” (as yet) – but they are done by the AUTHORS.

  20. I do not like the idea of a title page, as this seems inelegant. It makes the paper longer (I do still print some papers), adds little in of itself as the title page would be mostly blank, and creates a disconnect between the page numbers in the document and the page number of the pdf. The pdftk watermarking solution I gave above seems to be a pareto improvement on existing ECCC methods: it requires no work from authors and produces better watermarking than current methods. While this watermarking may be redundant for some papers that give date/etc information, the point is that we should not require more of the authors; ECCC should perform this task automatically for *all* papers.

    Has anyone pinged the ECCC maintainers regarding this discussion? I would like the problems highlighted to be fixed.

    1. I completely agree with you. I was planning to get in touch with the ECCC about this, but I was waiting a little to see if there’s additional feedback.

  21. Well, somewhat strange: nobody was happy with ECCC doing pdf->ps->pdf transition. And now should TikZ (a completely hidden software) “rescue” us? My suggestion: hands off TikZ and friends. Why? Just because they relay on many many things (not)existing on your pc … “Primitive” is actually = “genial”. TikZ is an opposite of “simple”. And cannot pretend to “genial” (by this reason). I am still advocating for PRIMITIVE solutions.

    1. @Stasys, perhaps I have not followed your intent precisely, but I believe Michael’s idea is not for the authors (users) to use TiKZ or anything else. My understanding is that he suggests that *ECCC* will automatically use the apparently simple solution (simpler than the current one I would guess) of invoking the pdftk command. This solution sounds quite simple to me.

      For instance, this is the example Michael linked to:

      Using the command line tool pdftk and using the stamp or background option you can do:
      e.g.
      > pdftk file1.pdf background file2.pdf output combinedfile.pdf

  22. @Iddo: I am sorry (also to Michael) , this was my fault – “pdftk” has indeed nothing to do with “tikz”. From the description, this seems to be exactly what could help to easily implement the “title page stamp”.

  23. @Iddo: “Note that for this to work, no title page is needed.” It would be nice if somebody could test this: can pdftk just put a piece of information (TR number, ISSS number) somewhere in (best on the top of) the submitted pdf. “Watermarks” are like something made *across” the page(s). I would also be happy, if we could avoid this “adding title page”. This would change the “look” of ECCC reports, and should therefore be discussed within the entire ECCC board. If you have an idea on how to avoid this (“big change” = adding title page) and just to use another tool (“pdftk”) instead of our old pdf->ps->pdf procedure – then please tell this (best – Emanuele). Any ideas on how to implement all this are also more than welcome (the “local office of ECCC has only person understanding how to program).

    1. I have tested pdftk, though not to the full extent that you desire, and am confident it can do what is needed. That is, there are the following steps.

      1) create a two page pdf called `eccc-stamp.pdf’. The second page is blank. The first page contains the “stamp” one desires. This pdf can be dynamically generated (say, from latex and appropriate scripts) to contain the TR number, etc. (I imagine ECCC already does something like this.)

      2) run `pdftk draft-submission.pdf multibackground eccc-stamp.pdf output eccc-submission.pdf’. This will stamp the first page of draft-submission.pdf with the first page of eccc-stamp.pdf, and will stamp the rest of draft-submission.pdf with the second page of eccc-stamp.pdf (which is blank).

      This replicates the ECCC look, except for the final page. (I should note that this puts eccc-stamp.pdf *below* the content of draft-submission.pdf. One can also put it above if desired). The final page could also be achieved by making eccc-stamp.pdf equal in the number of pages to draft-submission.pdf, and making the last page of eccc-stamp.pdf be the desired stamp.

      I would need access to the actual ECCC stamp files to do a more thorough test, and to see how this changes the size of the resulting PDF.

  24. @Iddo: thanks a lot for even tried this! Only one thing: the
    “stamps” of submissions during their “screening procedure”
    are not so important (they can appear in any place).

    Important is only: how ECCC reports will look after their
    “publication”. I mean: report number, year (and, eventually,
    issn number). That is, it would be good to get TR NR. and
    year on the *top* of the page (as is the case in all
    journals).

    I’ve also tried this. And what I got is just a mess on the title of the submission. Things (ECCC number, year) are written here, but these overwrite the title. So, perhaps, this “pdfpk” is still not an “escape”? How to tell “pdfpk” to “write on the top”?

    1. @Stasys: I’m not sure why it didn’t work for you. What pdf’s are you using?

      Perhaps I can be more specific. Download the ECCC stamp I made:

      http://www.mit.edu/~miforbes/public/eccc-stamp.pdf

      (This hard-codes the report-number/year, but this can be addressed). Then download a candidate paper, lets say:

      https://arxiv.org/pdf/1706.08786

      Now run

      pdftk 1706.08786.pdf multibackground eccc-stamp.pdf output result.pdf

      The file `result.pdf’ should now have the desired eccc stamp. To compare, the result I get is:

      http://www.mit.edu/~miforbes/public/result.pdf

      Does this work for you?

      1. Yes, it worked. But to have actual TR numbers, one should probably have a template “stamp.tex” file, which should be automatically changed and converted into pdf for every accepted submission. For me (as a layman) this seems to be doable.

  25. @Iddo (continued): well, it seems pdftk does not move anything – it just puts one page over another one. So, one can avoid the overapping by adding, say, \voffset=-3cm in eccc-stamp.tex. But this shows a potential problem: how the stamp is be placed will depend on the layout used by the authors. So, pdftk is also not a “panacea” but is perhaps still worth to try.

      1. If you attempt to place “background.pdf” as a background for “submission.pdf”, and they match in number of pages (or if background.pdf has more pages), then page N of submission.pdf will get page N of background.pdf as background. If background.pdf has fewer pages than submission.pdf, then the final page of background.pdf is used to stamp all further pages in submission.pdf. Thus, to only stamp the first page of a document, the background.pdf should have the first page as needed, and then a second page should be blank so this second blank-page is stamping the rest of submission.pdf. One could imagine avoiding this second page, but that is not a current feature of pdftk.

  26. Thank you all for the discussion! I have just contacted ECCC and I gave them this list of issues:

    1. Hyperlinks: avoid pdf->ps->pdf. Main issue. Specific fix on the website.

    2. Preview option (currently no possibility of previewing .pdf)

    3. RSS notification (add notification for updates)

    4. Search (doesn’t work well, specific example on website)

    5. Better lists of ECCC reports on one page. When clicking on “older reports” the page shows the same three that are shown on the first page. Might be better to just show all reports in one page.

    6. (Not mentioned on website, but my personal experience) the keywords selection is buggy, i.e., when entering a non-existing keyword the website draws some strange squares.

    7. Another thing that was mentioned is “changing authors names”. Currently ECCC doesn’t support that, unlike arxiv.

  27. Here is the reply by Oded Goldreich:

    Thanks guys. Esp. thanks for the digest on the last post, which I just received.

    In general, please send me any issue you have with ECCC, and CC also the system admin (Amir) if it is a system issue.

    Let me just say that fixing the system is not as easy as you may think, since we are still not in full command of its mysteries and we rather fix things only if we are sure that we know what we are doing.

    Items 4–7 have been on my TODO list for several years, even before ECCC re-located and have been on our TODO list ever since. But see above…

    Needless to say, in light of the above, I am not the right person to admin ECCC (even indirectly). But the previous managers wanted to free themselves from the task and nobody else on the editorial board volunteered to do so. If somebody would like to take it over, I will be happy to give it up, but do note that things are not as trivial as you think. It is a complex system that does a lot of things, some of which (i.e., a very effective assignment of screening) is not visible to you.

  28. The most I like in Oded’s answer is: “we rather fix things only if we are sure that we know what we are doing”. Current “eccc robot” (created in 1994 by Jochen Bern, then the system admin of our small chair at Uni Trier) is perhaps not the “most modern” now – but it works. For > 20 years. This – being able to get rid of this “dinosaur” in such a short time by Oded’s team – deserves great respect.

    This said, it would be nice if we (ECCC readers) could give Amir a possibly specific suggestions on what could be probed (up to “batch” files). By keeping in mind that everything must run automatically. By also keeping in mind that (as I admit) “eccc robot” still works under Linux, not Windows. Here I mainly have in mind Issue 1 (all others do not seem for me to be too hard to implement/improved). Say, “pdftk” seems to be a “simple” re-solution of the “hyperlinks problem”. But only if done manually. Trying to make the process automatic could be a more complicated task.

  29. I guess Stasys had a typo. We did NOT get rid of the existing old robot, but rather re-located it. I think Amir knows it well enough to operate and do small local changes in its source, but still does not feel comfortable enough to do more than that. I hope this will change soon. I did ask Amir to handle issue 1 at top priority. Again, I’d appreciate if requests are sent to me rather than posted on some blog, since I don’t really follow blogs and am not even sure now if I’ll succeed to post this answer (rather than have to bother Emanuele again).

    1. Dear Oded, my English is rather poor. Under “get rid” I’ve meant more “understand what these scripts are actually doing”. (I, say, couldn’t understand anything what Jochen told us during short “tee pauses” in 1994.)

      And I completely agree: if (and when) we have collected concrete suggestions in a blog discussion – these should be sent concentrated to your team (without referring to suggestions spread over a long blog discussion). This is why I am for sending Amir and you very concrete, ideally – already tested, suggestions (up to peaces of program codes).

      1. Wow, I’ve just realized that Issue 5 of your todo list (all reports of a year being listed on one page, not just “3-reports per page”) is already realized – great!

  30. … I mean when one goes to “Reports” -> particular year. Having this option, I am now in doubt whether we should change the current “3-reports per page” procedure for “Latest Reports” – this way we have more focus on “hottest events”, one by one. Anyway, issue 5 disappeared.

    1. I don’t think the “reports/2017” addresses this issue fully. Why not have “view older reports” just list the last 10 reports? It is a very prominent link as compared to “reports/2017”.

      1. Well, it would be of course nice to have an option (like in shopping centers) to have an option how many reports/products I want see on one page. My only sorrow only is: let’s solve “big problems” first (hyperlinks). Then do the rest. In this context, issue 5 is no more an issue. (It may remain to be an issue of “using eccc”.) Remember ArXiv: after 1 month we see nothing … And this in a supported site (ECCC is not such).

  31. Hi guys. PLEASE PLEASE send me email rather counting on me to read this blog. I just did by accident…. Like Michael, I have a different idea of what should be done wrt issue 5: I think the info on the middle col of the home page can be condensed into a list at the bottom of the leftmost col, and the middle col can be used to list the last 20 records or so (wo abstracts). I have a detailed plan re this and re what the additional pages (accessed by bot of the rightmost col) should look like. But it needs to be implemented.

    Again, if you want anything to come to my attention, please write me email. It is actually good to have a copy posted on blogs, so that others in the community know, but I cannot undertake to scan blogs for comments and requests re ECCC.

    1. Yes, we (at least I) don’t count on you following this thread. We will summarize the suggestions and send them to you or whoever else is concerned in batches (as I have already done).

  32. @Manu: re should papers accepted at CCC appear as “special issues” or as “simple eccc reports”?. I still think this an important question (in the context of the main question you raised in this post: “why LIPIcs, not ECCC”?. You argument (CCC already publishes lists of accepted papers, and this needs not be repeated at ECCC) does not fully convince me. Just quick thoughts:

    (1) How should the reference then appear is the section “References” in some subsequent paper? “TR Nr. xx of ECCC”, or still “in Proc. of xx-th CCC”? I would prefer the latter being followed by where the paper can be found (as ECCC TR).

    (2) Will these “sent from CCC” papers be treated by hiring committees as “tech. reports” or as “conference publications”? (Important for young people.)

    (3) Journals have “special issues” for this purpose. Why not also ECCC? LIPIcs is here on the “sun side”: they publish Proceedings separately.

    I admittedly ignore yet another (very important) aspect: to be accepted at CCC, a paper must undergo much higher “evaluation procedure”.

    1. I was thinking of simple eccc reports. I definitely thought of the reference “in Proc. of xx-th CCC” and of the papers “counting” as usual, not as technical reports. I don’t mean of ECCC as giving any “special status” to the papers. All the status is given by CCC. ECCC is just a convenient way to have papers indexed in DBLP and appear on the RSS feed of the people who follow ECCC, and put them somewhere which has storing capacity already, instead of duplicating these features on the CCC website. But the latter could be done too. But, we should indeed make sure that “Proc. of xx-th CCC” is also mentioned on DBLP, not sure how that’s accomplished now, maybe it’s automatic because of LIPIcs?

      Best, Emanuele

      1. But even in the CFP of ECCC we have written explicitly “Submission to ECCC does not prevent future submission to any conference or journal: the submissions that appear in the archive have the status of technical reports.” This is wipe out all “potential problems” with further journal publications of ECCC reports.

        I couldn’t quite understand what do you mean under “duplicating these features on the CCC website”? As I see, CCC webpage(s) are only giving links to LIPIcs proceedings page. They keep no “own versions”.

        Re your last sentence: I think this is so just because, at LIPIcs, these volumes are so called “Proc. of xx-th CCC”. Each conference has its own volume.

        I am afraid that, without being able to organize separate sections for “Proceedings of CCC” at ECCC, we have little chance to convince colleagues at CCC PCs to move to ECCC. Pro: no formatting work, no costs (30-60 euro pro paper). Contra: no DOI (not a big problem), no separate “proceedings section” (a problem).

        B.t.w. I observed, that many papers appearing in LIPIcs proceedings of CCC already have their versions published as ECCC reports. So, by just adding “CCC versions” (ECCC Rep. Nr. xx) with some stamps) would only confuse the readers.

        I actually cannot quite understand, why you are against “special issues” (Proceedings of CCC)? Perhaps I am overseeing some dangers. If these are potential implementation problems – then I completely agree: this could be not so simple to realize.

  33. Hi guys.

    I did not follow the entire CCC-put-in-ECCC discussion, but let me make a few comments.

    By the last couple of comments, I see some confusion between two options, which i think should be handled differently.

    Opt1: Have CCC ask authors to post on ECCC as regular TRs. This would be a mess, since some authors have already posted their works on ECCC, and we would not want to see two or more entries for the same work. Also, ECCC cannot put an “presented in CCC” label, since nobody at ECCC is going to do the necessary verification, and if authors put it, then it is not verified. Furthermore, not clear that these labels will be picked by DBLP and other indexers. I doubt it will.

    Opt2 is like Opt1, except that the PC chair will submit the report, which will be the full proceedings of CCC, and it will be published as a TR of ECCC. But then individual papers will not be indexed, only this proceedings-report. In addition, I doubt the PC chair will like this, let alone that one will have to agree on a format that will allow the PC chair to easily pack all papers into one file. If it is done at a latex level, then we have the same situation as with a publisher, except that we are the publisher and can enforce only very minimal formatting rules (e.g., not bother authors about overfull/underfull lines and/or some other latex warnings — but this requires agreement of CCC). But it can also be done at a PDF level 9and again CCC has to agree to that).

    Opt3: Have a special section on ECCC hold the CCC papers of each year. I mean sections such as those under the “book+lectures” button (5th of the left on the top of the homepage). But these, not very active…, sections are handled manually (not via the submission process for reports). We will need somebody (e.g., the PC chair) to provide us with a list of papers (+ author names), PDF files, and preferably also abstracts. Furthermore, somebody will have to create a page with the foregoing list + links to the individual papers. And at the end of the day, it is not clear if the indexers will pick this (since these are NOT the usual ECCC reports that the list). (I just checked, and it seems that DBLP does not index them, but this can be fixed with DBLP — less clear wrt other indexers.)

    In short, it seems to me that none of these options will work (wrt indexing).
    Again, an idea that sounds trivial in theory is not easily implementable…

    1. Thank you for your posting your thoughts here.

      What I had in mind was similar to your Opt2, but not exactly: The PC chair will upload each accepted paper separately to the ECCC. This solves both the problem of indexing and of formatting you mention. But I am not saying this is a panacea.

      One issue is what to do if he paper is already on ECCC. I thought that if it appears there in its final format, no paper is uploaded. If it is on ECCC but not in its final format, then perhaps we can find a way for the PC chair to submit an update (this may require non-zero intervention from ECCC).

      I was looking for something that involves the least amount of effort, and I certainly want to avoid all implementation issues as much as possible.

      To try to share my viewpoint let me also consider some options:

      (A) The status quo (using LIPIcs)

      (B) A “radical” solution: The .pdf of the accepted papers are on the CCC website, no LIPIcs, ECCC, or arxiv interaction.

      (C) What I just said above: the final versions are placed by the PC chair on ECCC.

      To me, (B) looked good already. However I thought that (C) could be only negligibly more work and have extra useful features not in (B), such as DBLP, ISSN, search, back-up CD’s shipped to people, RSS.

      My understanding is that some of the reasons why (A) is preferred by some people over (C) is: ISBN (in addition to ISSN), indexing by Scopus, DOI. Anything else?

      1. My reason to prefer (A) is that I cannot find any convincing advantages of (C) against (A). Already to enable a PC chair to prepare the entire volume (ignoring “what PC chair is ready to do this?”), some formatting job must be done by the authors themselves. Just to “merge” the PDFs of papers will not work: we need at least a TOC, page numbers, etc. Also, somebody at ECCC should make the volume with titles available via html. All this is now done by authors and people at Dagstuhl: authors submit – they form a volume. ISBN is not an issue (at least for me). But indexing at DBLP, Scopus already is an issue. About the necessity of DOI: here I am not sure. The only “disadvantage” of LIPIcs I see is that it doesn’t give an option without DOI.

        A good aside effect of comments to this post is that the “hyperlinks issue at eccc” was raised (once again). With specific suggestions (like “pdftk”). It would be great if we could sent Amir some scripts for testing.

      2. Again, I don’t see how these resolve the indexing problem. You want paper being index as appearing in the proceedings of ccc, but I think the only way to get this effect is to have the proceedings be recognized as such, with a isbn of it own or something like that. Eccc cannot do that.

      3. OK thanks. Indeed I don’t know what it takes to get papers to be indexed as appearing in the CCC proceedings; I’ll try to understand that better.

  34. I agree with Oded (without assuming he will read this). Even with some separate “Proc. of CCC” section we would have the same problem raised by Emanuele in this post: “formatting”. Only that now it should be done by a PC chair, and by ECCC “local office” (manually). This followed by the indexing problem at data bases. This followed by ECCC keeping “conference” and “TR” versions of the same papers.

    I think, the main problem with conference publications are “paywalls”. By switching to LIPIcs, CCC resolved this problem. So, the authors can (and should) invest a bit time into formatting (in my experience – not much at all), knowing that then their papers will be accessible to anyone.

    B.t.w. when submitting to a conference, the authors must undergo much stronger “formatting obstacles” to “pack” their findings into 10-12 pages. And, unlike submissions, proceedings published at LIPIcs have no space restrictions. [Yet another advantage of ECCC being thus eliminated.]

    Shortly, I don’t see any reasons why CCC should switch from LIPIcs to ECCC. Well, FOCS, STOC and many other conferences are still under paywalls. But, at least, we see that most of complexity relevant papers also of these conferences have their full versions at ECCC …

    1. Thanks again for your thoughts. I tried to clarify a bit some aspects in the reply to Oded above, in particular I don’t envision any formatting.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s