I almost fell for it

This year, the chair of ICALP decided to play an April Fool’s prank three weeks early, and I received the following message:

“Dear author, we regret to inform you that the margins in your submission are too small, and hence we are rejecting it without review”

I was almost fooled. In my defense, the second time that I applied for a position in Italy, the hiring committee judged all my publications to be non-existent, because the (multiple) copies I had sent them had not been authenticated by a notary. So I am trained not to consider it too strange that a paper could be evaluated based on the width of its margins (or the stamps on its pages) rather than on the content of its theorem.

41 thoughts on “I almost fell for it

  1. Submissions that are too long or formatted incorrectly may be rejected immediately. The LNCS margins were deliberately designed to accommodate a proof of Fermat’s last theorem.

  2. In their defense, they did send an email informing the authors that they needed to really make sure their papers were complying with the formatting constraints. The fact that this email was sent 3+ days prior to the deadline, and thus only to those authors who had a submission already uploaded to the server at that time, may be considered worth mentioning only by 99% of the people.

  3. When you say “almost fell for it”, it implies you are now certain that it was a joke. How do you know it was a joke?

  4. I found it quite rediculous that the CFP of ICALP did not mention margins etc this year, but then were apparently expected to be unchanged. Sure, they said the submission should be in LNCS format and at most 12 pages. At other conferences though, if the margins are not mentioned then they do not mind if they are changed within reasonable bounds. Getting a mail just a couple of days before the deadline saying that margins should be untouched produced a lot of unnecessary last minute work due to additionally having to shorten the main part of the submissions.

    In fact, if you say that you almost fell for it, and they actually accepted your submission with smaller margins, then I would say that you had a (small) unfair advantage, now that everyone else who submitted early enough to be warned changed their papers to comply with untouched margins… IMHO the ICALP committee should have made a better work at informing everyone of what exactly they expect in terms of submission format beforehand.

  5. By the way, I found it quite amusing that some commenters here failed to read the irony of the post. To be clear, risking the negligible possibility that I have somehow misunderstood, Luca’s paper was automatically rejected. The ‘almost fell for it’ is an ironic disbelief at the stupidity of this decision.

  6. Jelani, thinking that this is serious would be really offensive to the intelligence and the common sense of the program chair.

  7. @tas: a notary would be certifying that the copy of the publication is in fact a copy of the publication and not a fake. Although he actually certifies that the copy is identical to an original I showed them, and the notary has no opinion on whether the original is fake.

  8. Luca, this is an orderly German-run conference. You just do not mess with the margins of all things. What’s next? Your tree growing into your neighbor’s yard? Your car tire on the curb? Your pencil not sharpened? If you don’t like it, try one of those North American conferences. I heard they’re a total free for all. They say you can even get in with the wrong font!

  9. Moritz, I see what you are saying, but I left the margins alone. But we did have the audacity to only end the paper by 12 pages, and not also the references thanks to a conscientious bibliography with almost 60 references. Apparently it is okay for references to spill onto page 13 but not onto page 14. If you settle for an approximation algorithm, it better be an optimal +1 one.

  10. I love countries that need notaries for everything, I also happen to live in one of them. In Hungary you also need to have all your documents verified by the mafia called government. This includes translating documents in obscure languages, like English, for about 100 dollars by page, and paying about 200 to have your foreign PhD accredited if it comes from Cambridge.

  11. There’s a fine margin between getting your paper accepted for review at ICALP and getting it rejected.

    (BTW, the same April Fool’s prank was played on me)

  12. I tend to disagree with the comments, though I have to say that I did not submit any paper to this year’s ICALP and so I did not check how clear the rules were. But let me assume the rules on margins (and other typesetting format) were clear.

    The point is what do you expect the PC to do with papers that have no margin at all, with 15 pages instead of 12 (say), with font size 8, etc.? I seems clear that in this ridiculous extreme, the paper should be rejected because of its form (it is unreadable, actually much longer than what is authorized, etc.). So where should one put the limit? To my mind, putting the limit to what is actually asked for seems quite reasonnable! At least, it is deterministic and I know what I can and cannot do for my paper to have a chance to be reviewed.

    Btw, and sorry for those people who love national biases, I am not German.

  13. @B., every ridiculous bureaucratic rule can be justified that way. The goal of rules is to facilitate smoother and useful procedures as a mean for having a good conference and lively scientific program. Strictly and automatically rejecting many papers because of artificial rules is surely not something that helps for this goal.

    Another thing is that I don’t see the point at all of having any limitations on margins. If authors want to have zero margins it’s their problem. The reviewer can asses fine a zero and ample margin papers just the same, with almost the same effort.
    In any case, the effort in reading an evaluating papers is not linear in their length. Why not have limitations on the “hardness of understanding the results”? E.g., a paper in a “deep” area using nontrivial mathematics should be shorter than other papers by this rationale. Maybe we should limit the strength and complexity of methods used to prove theorems in papers, because “we have limited time” to evaluate them?

    Margins, numbers of pages etc. are all artificial and formal limitations, that no one should take too strictly. You have to use common sense when applying them. It is quite obvious that in this case the decision made by the program committee is very poor, and it simply hurts ICALP’s reputation.

  14. First, let me make it clear that I never believe in page limit and would advocate for getting rid of it completely.

    But given that a conference chooses to adopt page limit rules, I kinda believe that what ICALP’s PC did makes certain senses. The page limit puts a lot of pain on many authors submitting to ICALP, and they tried very hard to meet these rules, at least I tried very hard. My paper compression hurts its readability, and I feel that it hurts the chance of acceptance.

    How would I feel if I knew that someone had an easy way to play these rules and get away with it fine? I feel that it’s unfair. The point is that there are people who take the page limit seriously (I am not implying that they are Germans, don’t get me wrong), and it is unfair to them if this rule is announced but not enforced.

    So, this is the conclusion: Either we get rid of the stupid rule completely, as in some preceding FOCS/STOCs, or we aggressively enforce it. Pretending that we have such a rule but lightly enforcing it, well, doesn’t make sense at all to me.

  15. @B.: the point is that the ICALP submission guidlines were not clear at all! All they said was LNCS style with the main part (including references) on the first 12 pages. No comment on margins, font size, or anything else. How would you infer from this that you are not allowed to tinker with the margins? Even if I agree with Ben that these limitations are ridiculous from the start, if there have to be limitations then please state them clearly so that you don’t have to automatically reject papers of people who quite naturally assumed that changing the margins slightly is accepable! Sorry, I’m venting here, but honestly I have been very disappointed by the ICALP organization so far.

  16. And by the way, if the chair is determined to enforce draconian rules like wide margins, it is a good practice to contact or send an email to all authors *after submission*, asking them to resubmit immediately before rejection. This is what was done in STOC a few years ago.

  17. Formatting rules are silly. If authors want to submit photographs of their handwritten, coffee-stained manuscript, then they should go for it, with the understanding that reviewers will take readability into account. (And the PC will make it clear that the camera-ready version needs to be better-formatted.)

    There should always be some flexibility around formatting. Most papers will have a few equations that go into the margins or references that don’t adhere to the required style. Rejecting papers for minor formatting transgressions will only reduce the quality of papers at your conference and does nothing to advance the research in your area.

    Bureaucrats of course love such rules, as they are very easy to enforce. Rather than reading and understanding a paper and judging it’s novelty, interestingness, and technical merit, it can be summarily rejected based on the use of a measuring tape. Of course, a serious academic like the ICALP PC chair would never be so lazy.

  18. On A4 paper, the standard LNCS margins do not look quite as ridiculously narrow as they do on Letter paper. I would guess that a higher proportion of North America authors got caught out because they saw them on Letter paper and because they were not so familiar with the style itself.

    A message to authors about margins was sent via easy-chair just before midnight on Feb 16, but this missed the large number of authors who did their first submissions after that time.

    The whole LNCS style is absurd and a waste of paper and space. LNCS proceedings are also vastly overpriced for libraries. (A year’s electronic subscription to LNCS proceedings costs much more than that for the complete ACM digital library and contain proceedings of loads of worthless conferences in addition to conferences like ICALP.) It will be good riddance when we no longer use LNCS or their style!

  19. It seems obvious to me that LNCS style implies that one is expected to use the LNCS style file and not mess with it (like increase the margins or reduce the font size). I agree, this could have been made clearer. But what about asking the PC chair if in doubt?

    Assuming that this is understood, nobody should complain that their paper was rejected because of deviations from the submission guidelines. It is a matter of fairness. Where do you draw the line?

    @tas: I doubt that you have ever served on a PC, where you are expected to read 30-40 papers within 4-6 weeks. Go ahead, and read 30-40 handwritten photographs and then try to take hours to decipher every single smear. If you’re at it, perhaps you could retype it yourself, so that you can give it a fair evaluation based on your idealistic criteria. And do all of that in addition to your regular 10h per day workload.

    Alternatively, you could impose some constraints that whelp PC members to be reasonably efficient. Or you could just reject every paper that you deem unreadable. And then answer 100 angry e-mails of authors whose papers got rejected and they don’t understand why.

    I guess, it is always easy to complain and to blame the PC for everything, and especially for rejected paper. It’s almost always the PC’s fault and not the author’s. (That’s a theorem.)
    Talking, complaining, blaming – that’s easy. Certainly easier than following the submission guidelines. (But how hard can that be?)

  20. “My paper compression hurts its readability”

    This is the reason that I don’t believe in technical rules such as page limits. Reviewers are always free to skip sections or stop reading at any point they feel its appropriate to do so.

    That said, if you have such rules, I don’t think its crazy to enforce them. However, since our community’s norms is that such rules are rarely enforced, its best to follow what Salil Vadhan did in STOC 2011 when he sent the following email to all authors (*after* the deadline):

    Dear authors of STOC submissions,

    It appears that a number of STOC submissions have not followed the font size and spacing guidelines in the Call for Papers, and the program committee does not wish to read such submissions.

    Thus, we are giving authors of these papers a chance to send (by reply to this email by tomorrow 11/19, 8pm EST) a version of their submission that is recompiled to meet the guidelines, namely at least 11pt font with at least 1 inch margins and ample spacing all around (including normal spacing between lines of text). The recompiled version should otherwise be *identical* to the submission, with no change in the content or the order of material in the paper. If the resulting paper body exceeds the allowed 10 pages, then material after the 10th page will be treated as part of the appendix. As stated in the call for papers, submissions that do not comply risk rejection without consideration of their merits.

    Please note that this message is being sent to all authors, and does not necessarily mean that your submission is in violation of the formatting guidelines.

    Sincerely,
    Salil Vadhan
    STOC 2011 PC Chair

  21. To be clear: I am not saying that using LNCS style is a good idea. But some constraints are necessary for efficient PC work. Given that there are submission guidelines, they should be enforced for fairness reasons.

  22. I feel it’s wrong to dignify this ridiculous episode with a serious comment but here goes.

    I have served for a lot of conferences as referee, program committee member, and program committee chair, and of the several challenges involved in performing this service, it has never occurred to me that the formatting of a paper would be a notable one, except for papers with very small fonts or otherwise hard-to-read styles (such as the Spring LNCS style). I know that some people feel that an upper page limit is helpful to reviewers, and I disagree with it, and in fact I imposed no page limit when I chaired FOCS 2010, but I accept that other people feel differently.

    Many of the rejected papers did not “significantly deviate from the guidelines” in terms of length; they were either one or two pages longer, or were in a different style.

    My coauthors and I actually compiled our paper with the Springer style, and it was roughly within the guidelines (references ended at page 13), but the Springer style hurt my eyes so much that I asked my coauthors to recompile it differently, so that it had bigger fonts, and, yes, smaller margins. (Luckily, my coauthors are not mad at me because this will just mean we will send our paper to FOCS instead.)

    But how does it serve the purpose of putting together the best possible conference to reject out of hand several good (present company excluded) contributions? How does it help the reviewers to enforce submissions to be in an unreadable style?

  23. Luca (and, from what I gather here, Rahul, Venkat and the other authors of ICALP 2015 Track A submissions that were rejected without refereeing),

    I am posting this comment with my President-of-the-EATCS hat on. Since ICALP is the flagship conference of the EATCS, I take full responsibility for the decision of the PC of ICALP Track A. The buck stops with me.

    The PC for ICALP Track A asked the leadership of the EATCS whether we would oppose rejection of papers that are too long or formatted incorrectly, and we had no objection. Of course, it is up to the PC to decide where to draw the line and how to do so. We select the PC chairs, assist them in forming their PCs and then let them do their job without interfering.

    Of course, I understand the disappointment of the authors and I appreciate all the opinions that have been aired here, which will be useful to us. They show that this is not a clear-cut issue and that there are many different viewpoints on it.

    I will limit myself to saying two things.

    1. I assure everyone that no decision related to ICALP is taken lightly. I like to think that anybody who has ever attended ICALP can testify that ICALP organizers and ICALP PCs uphold high standards that are reflected in the quality of the conference organization and of its scientific programme.

    2. The Submission Guidelines for ICALP 2015 start as follows:

    “Authors are invited to submit an extended abstract of no more than 12 pages, including references, in LNCS style presenting original research on the theory of Computer Science.”

    IMHO, a paper that changes font size, the size of margins, line spacing and so on redefines the LNCS style; I might be naive, but it never occurred to me to do so. However, one should learn from experience and we will make explicit the fact that no change is allowed to the LIPIcs style that will be used for ICALP submissions from 2016. (ICALP proceedings will be published in open-access form with LIPIcs from next year.)

    I do believe, however, that there is nothing unclear about the allowed length of the paper or the fact that one should use the LNCS style.

    I would be glad to hear any further comment or suggestion that can be useful for future ICALP conferences and you are welcome to send me an email with subject line “ICALP comment”.

    Our goal is to continue improving ICALP and your opinions are important for the EATCS.

  24. Assume the PC chair accepted to review all the papers (that do not fit the styling guidelines), and some are accepted. So, when time is to submit the final version you must submit the accepted version (taking into account the minor comments). How to do it within 12pages (which I always find ridiculous) if you exceed it? I can understand that at Springer they can arrange if you exceed by at most 1 page, but what about those who change so that they exceed by >=2 once formatted with the right llncs?

  25. Let me precise a few points to make my comment clearer:

    1. I’m against formal rules for the number of pages and so on, for several reasons including the ones mentioned in other comments such as “shorter is not necessarily easier/faster to read”, etc.

    2. The LNCS style is ugly.

    3. Given the rules, I find it normal to respect them even if I do not like them! As XYZ states it, there is also a question of fairness and his example of a paper whose quality is mindered by respecting the rules and that could be rejected while another one that does not respect the rules is accepted is an important point.

    4. Changing the font size, the margins, etc. of the paper is clearly a violation of the very simple and clearly stated rule “no more than 12 pages, including references, in LNCS style”. For me, this is cheating.

    5. The solution to give some extra time for authors who did not respect the rules to modify their submission make perfect sense. But then, for fairness, everybody should be able to modify their submission. What I would find reasonnable is the following: First deadline at date D; Very quick feedback on formatting issues and so on at date D+1; Second deadline at date D+2 or D+3 for everybody. And strucit enforcement of the rules after the second deadline.

  26. Let’s be clear about it: when you say to people to use LNCS style, it does NOT exclude the possibility to change the margins. Indeed, this is a clear cut question with almost unique and determined semantics: using a style in latex formally means using a .sty file. But it doesn’t exclude using common commands to widen the margins.

    Thus, formally there was nothing in the cfp to justify these draconian actions.

  27. @Ben: I totally disagree. Using the LNCS style file, I am able to produce a document with any kind of typesetting. It make no sense to pretend using this style while redefining its features!

  28. @B, I disagree. The fact is that when program chairs want to be specific about formatting rules they say so explicitly in their CFP. That is, they ask explicitly not to change the margins, font size and line spacing as well as using certain style files. Asking to use a specific style file (in this case LNCS) does not exclude the possibility to use other formatting commands.

    In this case, not only there was no explicit mention of margins, there was no email sent to ask authors to change their formatting before automatic rejection (an example email was shown above in Barak’s comment).

  29. (cont.) also, regarding:

    “It make no sense to pretend using this style while redefining its features!”

    I claim that this is factually wrong: if many senior and very experienced authors, including people who served as PC chairs of leading conferences (!) find that it DOES make sense, then by definition of “makes sense”, it does.

  30. Re. P. (https://lucatrevisan.wordpress.com/2015/03/10/i-almost-fell-for-it/#comment-17043):

    A paper can conform to the formal formatting guidelines and be entirely incomprehensible and, conversely, a paper can deviate from the required format and be a pleasure to read.

    PC members can decide for themselves whether a paper is readable or not. I find it perfectly acceptable that they reject a paper for being unreadable—PC members have enough to do; they are not obliged to decipher obfuscated submissions.

    So what’s the point of formal formatting rules? I find it strange when CFPs seem to imply that the reviewers are obliged to read your paper as long as it meets the guidelines.

  31. Formatting deviations are a recurring issue in conferences, and if there are some guidelines, it is perhaps reasonable for the PC to enforce it to some extent. However, there are deviations which actually aid reviewing especially for a nasty underlying format like LNCS style (like Luca’s increased font size, or in my case using a nonzero parskip), but which ended up formally violating the rules (because references didn’t also end by page 12), and there are deviations which shrink font size and other changes to make things harder for reviewers. The question is how to distinguish these two — the only uniform one seems to be to follow the CFP guidelines to the letter. But given that this is hardly a new phenomenon, the common approach has probably been to just let PC members take this into account in their review. I was on the STOC’11 PC that was mentioned above by Boaz, and indeed there were several PC members complaining about papers with 10 pt font, and some even suggesting to reject them, which after some discussions led to the (excellent) action by the PC chair.

    Doing away with page limits can get around this problem, but even if there is a page limit, it does not make sense to include the bibliography within the limit. It actually disadvantages authors who are thorough and make extensive citations. I hope future ICALP Calls for Papers will address this issue, if Luca Aceto is reading this.

    About page limit vs no page limit, I actually thought about this recently when drafting the FOCS’15 CFP. When I polled the PC for their opinions on the submission guidelines, there were basically only two universal points that came up: (i) PC members should not be expected to read beyond some limit (say 10 pages) in any paper, and (ii) submitting authors have to provide full proofs in one form or another.

    There have been two approaches to meet these objectives in STOC/FOCS/CCC:

    (i) The traditional one is a 10 page + appendices format (which of course leads to tweaking to fit things in 10 pages, and hurts readability especially if the 10 page version is formed, as is typical, at the last minute by hastily shoving things to the appendix).
    (ii) The other recent one is not having a page limit but saying the first ten pages should convey everything needed to make a decision about the paper.

    In principle (ii) is a very good solution, but not all authors immediately change their writing style and conceive the full versions this way, and indeed in some cases it may not even be best to write the full paper this way (for example if the full version is intended for a Math or other community journal, and the first 10 pages cover mostly background and notation).

    For FOCS 2015, the submission guidelines suggest a hybrid of these two. The authors should submit a 10 page extended abstract (bibliography and also explanatory figures are excluded from this limit), and can upload the full version as a separate file. While I don’t like managing multiple versions as much as anyone else, really the idea is to give authors flexibility and also encourage the writing of a clean full version (which I ultimately view as the most important version) in whatever format they feel is best, as long as it includes all proofs.

    If written with a goal like (ii) in mind, the first 10 pages of the full paper can be the extended abstract. But if the full version is already written, then a shorter version summarizing the interesting aspects of the paper, including discussion of its context and motivation, significance of the main results, and an outline of proof methods, can be submitted as the extended abstract, and there will be no need to tinker with the already done full version.

    Obviously if the extended abstract includes all proofs, there is no need to submit the full version. As added flexibility, if the full paper isn’t much longer than 10 pages, the authors can supplement the extended abstract with appendices that extend beyond the main body of 10 pages and ignore the full version (i.e., solution (i) above).

  32. I should add that the above mentioned hybrid is not exactly new; indeed it has been suggested in several recent conferences that the appendix simply be the full version. The FOCS 2015 submission format saves the authors the hassle of stitching together two files, at the cost of the reviewers not having the full version handy in the same file.

    Btw, let me take this opportunity to announce that the FOCS 2015 submission server is open and accepting submissions now. It is linked from the Call for Papers http://www.cs.cmu.edu/~venkatg/FOCS-2015-cfp.html
    where the above submission instructions are also spelled out.

  33. We all are somehow ignoring the main aspect of conferences: The call for papers of ICALP asks to submit EXTENDED ABSTRACTS, not PAPERS.

    Easy to grasp, easy to decide, easy to advertise.

    Please, don’t mix all this with normal journal submissions. (Where no fonts or margins play any role.) Just PC members have a huge load to be done in a short time. So, many things (like automatic rejections) just MUST be automatized. Easy screening, easy decision – this is the credo of conferences. Not a long and exhaustive refereeing in journals …

  34. Well Stasys, what you say does not justify what happened here. Because, similar conferences have the same constraints and they don’t haphazardly reject papers based on slightly smaller margins and without prior warnings and clear notice about formatting matters. So, no, automatic rejections of serious papers is not something that usually happens in leading conference.

  35. – While I agree that if there is a rule, it has to be enforced somehow (otherwise some people will surely abuse it), the CfP could have been a bit more clear. For example, the ESA CfP has been using much stronger language for years:

    ” Papers must be formatted in LaTeX, using the LNCS style file (available here). The maximum length of the paper (including references, but excluding the optional appendix) is 12 pages. Do not change the margin size or the font, do not make a separate title page, etc.: use the LNCS style file as given, but please do add page numbers, which can be done, for example, by adding \pagestyle{plain} just before \begin{document}. Proofs omitted due to space constraints must be placed in an appendix to be read by the program committee members at their discretion.

    These guidelines are strict: papers failing to adhere to the guidelines (by not providing the omitted proofs in an appendix, being more than 12 pages, or not being in LNCS format) will be rejected without consideration of their merits. ”

    – Maybe some people are not aware of this, but the physical dimensions of the printed LNCS proceedings are very small and the margins are actually very narrow. So there was a reason for the choice of narrow margins.

    – Another point to consider: the PC may want to see a version of the paper that is an approximation of the 12-page version eventually included in the proceedings. If the authors change the formatting of the submission, the first 12 pages may need to be substantially shortened in order to fit into the proceedings. Note that these days STOC/FOCS/SODA PC members do not really know what will appear in the 20-page proceedings version of the paper, but probably this is not an issue as the page limit is not as restrictive.

  36. IMHO, once the CFP sets some rules, it is unprofessional NOT to enforce them. For one, a lot of authors may have gone through the trouble of complying with those rules, spending time and risking clarity of the presentation by making things concise etc. Not enforcing the rules would give a slight advantage to the authors who just did not bother. In the case of ICALP, the problem may be the formatting rules set by the CFP (which I personally believe should be changed), but certainly not enforcing them.

    On the other hand, at the end of the day the final camera ready version should fully comply with the formatting requirements anyways. So why not prepare the original submission correctly to begin with? Unless you think the paper isn’t good enough to get in, or are just recycling some rejected paper without bothering to fix it for a new submission, sticking to the CFP instructions shouldn’t be much of an issue.

    By the way, if I’m not wrong LNCS uses 10pt font size (as opposed to the usual 11pt), uses a more compact style, and is optimized for being printed in book form (so the actual margin will be pretty narrow in the end). But when printed on letter/A4 the margins look too wide.

    Admittedly, we are used to strictly enforced deadlines (sometimes precise to the second) but not strictly enforced formatting rules. And the whole thing does sound a little like college homework submissions. In the end, we’d like to care about theorems more than margins.

  37. In general, the submitter should be given the benefit of the doubt — if their interpretation of the requirements is reasonable, then it should be allowed. It is not enough for the PC to just think that their own interpretation is reasonable, it must be that the alternate interpretation is unreasonable. (In this case, I think both the submitter and the chair’s interpretation of the margin rules are reasonable, hence, the submitter must be given the benefit of the doubt).

    If you google “LNCS style” you will find more detailed author instructions (see link below), which define what is entailed by this style. There is nothing said about margins. Thus, someone is perfectly reasonable in assuming that they can use llncs.sty, decrease margins, and still be following the “LNCS style”.

    http://www.springer.com/cda/content/document/cda_downloaddocument/Springer_CS_Proceedings_Author_Guidelines_Jan_2013.pdf?SGWID=0-0-45-1121537-0

  38. Pingback: How many theoreticians does it take to approximate Max 3LIN? | in theory

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