> For scientists submitting their papers to journals, there’s an all-too-familiar drill: spend hours formatting the paper to meet the journal’s guidelines; if the paper is rejected, sink more time into reformatting it for another journal; repeat.
What? The journal provides (or at least should) a latex template, you write stuff in that, then you ship it. Last one I wrote in a browser through overleaf, spent zero time on formatting.
I definitely don't know how to write latex, can muddle through anyway. Total loss to trial and error / searching bits of syntax is probably below an hour so far this lifetime.
> What? The journal provides (or at least should) a latex template, you write stuff in that, then you ship it. Last one I wrote in a browser through overleaf, spent zero time on formatting.
Then you don't have much experience writing papers.
Formatting is very time consuming. Some want 8 pages, some are unlimited, others want you to use whatever makes sense.
Every journal has a different page length. They expect to see results in different places. Some want a detailed appendix. Others are ok with a website.
The latex template doesn't help at all with any of these.
Three in the real world, some number in undergrad. Writing papers takes a fair amount of time - think how to express things, producing graphs and so forth - but reordering sections or adding things to a bibliography is yet to cause much grief. Maybe we mean different things by formatting.
That said, recognising static academic reference format, then reformatting for the other 4 or 5 standard formats seems like a python project for first year students (to this non-programmer, I'd guess there are lots of edge cases to trip on making it a great learning experience!)?
It often isn't so much the actual bibliography line item format, which are usually handled quite well be systems like bibtex or biblatex+biber, but overall citation formatting complexities. Sometimes journals want specific formats that become a challenge to implement. For example, on a recent submission, we needed our main text to have a numbered bibliography at the end, then needed our methods to have numbered bibliography at the end, but with numbers starting after the last number in the main text, and no repeats, but with references in the methods to items already in the main text bibliography referring to those numbers. This was surprisingly difficult to implement reliably in LaTeX and Biblatex.
An additional annoyance is that to save space, journals often want formats that are compact but inefficient to use, for example, formats that are heavily abbreviated and don't include titles. These, however, are annoying to use, and annoying reviewers is generally unwise. So in our case, we actually intentionally submit PDFs for review that use the 'wrong' bibliography item format, of a similar form to the journal's but less abbreviated and including titles and hyperlink backreferences to citations in the text. Then for final submission we have to replace this with the journal's preferred format.
Now that is a solved problem. The issues the grandparent post raises about rewriting the text to be shorter or longer are real, but changing to a totally different reference format is what Latex does automagically for you when you apply the publisher's template - I can't recall the last time when I even looked at what a publication's reference format is, I have the list of sources in bibtex and whatever their template outputs should be correct, and if something in the format was difficult to implement in Latex, well, that was their problem for whoever made that template back when they did it.
Nature accepts LaTeX+PDF submissions, and will send the PDF out for review, but for final submission after acceptance, they prefer mostly-unformatted Word files; if you submit a LaTeX+PDF final submission, they convert it to Word [1]. As they stress that authors should not bother working on the visual formatting of their final submissions, my expectation is that, internally, they then take the Word file and manually put it into their own typesetting system.
Publishers like Dagstuhl Publishing that can require LaTeX submissions and use a completely LaTeX publishing workflow probably have significant efficiency advantages, but then rely on potential authors being sufficiently experienced with LaTeX that a LaTeX submission requirement doesn't drive people away. This can work well for computer science and some parts of physics, but for other fields it can be a significant burden.
Yes but what about all this unnecessary gatekeeping? gatekeeping is necessary to maintain a certain level of quality. On this specific issue of formatting gatekeeping. There was no conspiracy. I imagine that the policy solidified over time. The paper wants to maintain a specific look. Authors are upset when changes get introduced while formatting. The paper then says "you can format it yourself if it meets our specification". Now you have to format it yourself.
Thankfully in our web enabled era there a ways to publish without going through a traditional publisher. However you will note that the quality of these publishers is a mixed bag, real science right next to quackery. No one has done the gatekeeping needed to separate the wheat from the chaff.
First is usually not a huge worry. There are times where you are competing with someone else in the same field and want to be first. But most of the time you are just hoping someone is interested enough to get additional funding to peruse the project. The researcher wants their results published via the most respected publisher possible. The publisher wants to increase their respect(and desirability) by picking the most respectable results. If the researcher is unable to to get published by the ideal respectable source they will have to settle with a less respectful source.
How much does having a consistent "look" lend itself to the perceived professionalism and respect a journal has?
Scattered additional thoughts on desirability of publishing platforms.
* Self publish(and hope you are famous enough to get noticed)
* Publishing company that does little to no checking of papers.
* open web based publishing.
* corporate researcher, all published results stay in corporate library.
There are a couple of journals in math that ask for the paper to be submitted using their particular latex template, and a typical latex template is incompatible with what I consider "standard" latex, i.e. either the `amsart` document class or `article` document class with certain AMS packages included.
Sometimes it's small, like whether or not `\author{my name}` needs to be included in the preamble or in the main body. These are short but annoying. Sometimes it's very annoying. I'm glad that I've never tried to publish in a two-column journal (which might exist in math? I don't know) which would require more substantial superficial edits.
I always assumed that this was a deliberate source of friction, slightly raising the bar of entry so as to discourage a certain sort of submission. I don't say I agree with this, but that's what I thought.
This sounds like the sort of problem that latex would be great at solving: changing the look of an article just by using different document classes. And sometimes it is great --- but some templates are terrible, terrible, terrible. (And sometimes these terrible templates are for great journals! I see little to no correlation there).
> I always assumed that this was a deliberate source of friction, slightly raising the bar of entry so as to discourage a certain sort of submission.
There’s no doubt in my mind that it’s intentional gatekeeping. There’s a reason why many papers are being “published” on Arxiv and foregoing all this boxticking.
The whole publishing process is fundamentally broken. Aaron Swartz (RIP) had the right idea all along. Get rid of the middlemen sucking up millions of dollars by paywalling government funded research.
It’s been proven time and time again that a shocking number of studies can’t be reproduced and that the traditional peer review process just isn’t working. It’s a good old boys’ club putting a stamp of approval on their friends’ work and keeping the outsiders away.
When it comes to Math though, I do have some sympathies. There are so many cranks that think they solved Collatz or a millennium problem and don’t even know what they don’t know. It must be exhausting to sift through all that noise in a volunteer, unpaid position. These shibboleths might be a necessary evil.
I think it’s an asymmetric problem across areas of study. Including within math — there aren’t many amateurs who are going to try their hand at algebraic geometry, but number theory tends to be like catnip, because many wicked problems are very simple to state.
Math is not so different from computer science in that it seems relatively approachable to a layperson. But there are relatively few mathematicians out there. That same level of spam is being filtered by much fewer people.
To me, it’s akin to why the Linux kernel devs operate the way they do. No GitHub issue page, no Slack, no Discord, banning topposting and web email providers. It’s all based on old school mailing lists, because there’s orders of magnitude more amateurs than kernel devs and they don’t have time for handholding the unwashed masses.
Anyway, I think there’s some middle ground to be found between this extremely burdensome process to submit a paper for publishing, and making it as easy to publish as making a forum post…
For what it's worth, as a curious computer science university student in 2013 I sent Linus Torvalds an email from my personal GMail account. He answered both my original question and a supplementary question, even though they were (in hindsight) quite stupid. Even the great and very busy Mr Torvalds has at least some time for handholding unwashed amateurs.
On one project I had someone trying to slow us down (we were making them look bad) with rules lawyering about our API docs. In the end I ended up prettying up our integration tests quite a bit, and using jcite to combine snippets of the integration tests with our inline documentation.
I couldn't quite get it to the company standard, but I got it down to under an hour worth of work for our tech writer per release.
In the end it was a week's work for me and it improved our integration tests. Not the constant draw on resources and sanity that I think someone was after.
I am a mathematician, and I always ignore requests to submit papers using a certain LaTeX template, or using any other journal-specific formatting. Judging by the papers I get to review, so do most other mathematicians.
Once (for a Springer journal) I got called out by their editorial staff, and told I had to follow their dumb rules. So I did. But I would never do this preemptively.
Publishers charge thousands of dollars for a subscription to a single journal. If they want their papers formatted in a certain way, then they can employ staff to do that work themselves. If any for-profit journal tries to insist that authors do lots of reformatting work before even submitting, I will never submit to or review for them again.
Go one level deeper and you'll find the problem with all of academia (source: I have a PhD).
Most of academic science as we know it today is structured so that the output is "publishable" and/or helps future grant applications. Incremental improvements are very publishable, but that doesn't necessarily make good science. Grants are awarded to scientists who are consistently able to deliver results, in the form of published papers. I can only really speak for my little corner of science, but from my view, the entire incentive structure of science is broken.
Incremental publication seems like a great idea. The only problem is that the publication process is so formal. But if we made publications as "blogs" or something, I think incremental units of forward progress are a reasonable way to do science. I suspect most science is done this way. I imagine there are few Wiles style FLT proofs around.
As the parent poster states, the key thing is that everything is structured about incentives conditional on delivering measurable results e.g. the quantity of published papers; but that implicitly includes a very important thing that "published" counts only if (and because) it has passed certain filters. There are no barriers to publish research outcomes as "blogs" or something, it is easy, so such "publication" doesn't imply that any reasonable research was done, and no incentives would reward such publishing, and that's why researchers often don't bother with it.
Like, the existing system already has issues with publication being too easy to game, with predatory paid journals, lax standards for reviewing, etc; that's hard to fix, but it's considered a problem - and so any changes making the barrier to publication even lower than that won't be welcome.
And from the perspective of researchers, we don't really want more publications - they're a pain to read, there's an overabundance of poor content that's salami-sliced, we do so many publications because the incentives push us toward this but we recognize that this is a bad thing and we'd be far better off if the same research would be published in fewer, better articles. But we can't, because anyone who does that will be effectively denied resources for future research.
it would also help us move away from the PDF to a web based medium, pdfs are generally suboptimal if not awful reading experiences for a scientific paper
Your argument is essentially that scientists, especially early career scientists, have very little room for failure. I would argue a central problem is that many grants for early career researchers have award rates below 10%. This is problematic because at some point why would you blow away a month of work to write a proposal that you gave a tiny chance of receiving. If the award rates were 20-30%, then you have more room for failure in the interim between grant awards.
> the authors of the analysis, which was published in BBC Medicine on 10 May, propose that journals should allow free-format submissions so researchers can spend their time and money on research instead.
Yeah right.
This is YC though.
Up-for-grabs: Untapped $260m market for reformatting scientific papers.
Perhaps this is true in the field discussed in the essay, but in my field (a natural science) formatting is taken care of with latex stylesheets, so the effort of switching to another journal format usually amounts to changing a few lines. Of course there can be small details such as whether a given journal wants keywords, but dealing with such things doesn't take much time.
The problem of switching to another journal is also overplayed in the essay, at least in my field. I don't know of many people who "shop" for journals to accept their work. Usually you know the right journal, and if the paper is rejected you just give up on it. Perhaps this is discipline-specific, though. (The essay is restricted to biomedical fields.)
As a reviewer, I've never spent any time on journal rules. If the paper gets accepted, it's up to the technical editor to impose rules.
Most authors employ the proper latex stylesheets from the first rough draft. They also use section headings that fit the journal's conventions, and so forth.
I suppose it is possible that people in biomedical fields (are there any here on HN?) tend to submit to multiple journals before their work is accepted. But I sort of doubt that reviewers reject papers based on things like italics or citation formatting, so if folks are submitting to a lot of journals, maybe their work just isn't highly regarded by their peers.
Basically there is no issue here, at least in my field. It's an unconvincing essay, and not one that encouraged me to waste time looking up the original work.
The fast pdf generation is great but I lament on their choice of markup language, why create yet another? I'm a fan of writing content in markdown and there is at least on markdown flavour (MyST markdown) that gives the essential writing features needed for papers. (citations, cross references, figures, tables)
Your account has continued to post unsubstantive and flamebait comments to HN, a lot of them, after we asked you to stop. We're going to have to ban you if you keep doing this.
Not true. IEEE, one of the largest professional orgs, accepts Latex. So does ACM, a huge comp sci org. I'm not familiar with others, but wouldn't be surprised if many others still accept Latex.
How many IEEE papers are submitted using the latex template? I wonder, because word-style formatting can be spotted in several of their papers.
Latex is a giant pain to use - and i say this as someone who still uses it to write scientific articles. Services like overleaf exist but that is sometimes not an option with proprietary work.
Depends on the specific publication/conference within the org. ACM accepts LaTeX or Word. I always write in LaTeX, but got a paper accepted to an ACM conference that wanted Word and my beautiful photo ready paper got borked in the final print. I guess there is significant interest in authoring ACM papers in Word[0].
In my CS subfield we use latex, but switching journals still tends to be a huge pain, as many use extremely brittle templates that can easily create incompatibilities and make the paper not compile if you don't spend considerable amounts of time debugging after making the change. In addition, different journals have different length limits, word limits, structural requirements, one vs. two column (which may make your figures overflow), what goes in the main text vs. supplementary material, bibliography format changes that sometimes are only partially handled by bibtex, etc., so basically what the post says. At least, ChatGPT seems to be very good at cutting text to try and fit length limits.
And no one gives up with a paper, although this might have more to do with location than discipline (the academic evaluation system in my country is heavily metrics-based, having indexed papers is what makes or breaks careers). People resubmit and resubmit and resubmit, typically to venues with a progressively lower bar, until the paper is accepted.
In CS we also have competitive conference publications and, at least in my subfield, they don't have these problems to such a great extent as journals (as at least their latex templates tend to be decent, and some of them have the same length limits). One of the reasons why I prefer conference to journal publications.
> Perhaps this is true in the field discussed in the essay, but in my field (a natural science) formatting is taken care of with latex stylesheets, so the effort of switching to another journal format usually amounts to changing a few lines.
That doesn't sound like the LaTeX I use. When you switch to another stylesheet, it often changes key dimensions such as page size and the number of columns. Then you have to reformat and possibly even redesign the figures, tables, equations, pseudocode, and so on. And it's quite likely that some packages you are using don't work properly with the new stylesheet, or even don't work at all, so you get some weird issues to debug.
But that's about the technical layout, which isn't the concern here, as the publisher is usually responsible for that. In biomedical fields, manuscripts are usually just Word documents with some basic formatting. The real issue is that different journals have different ideas about the length, content, and structure of the paper. When you switch to another journal, the paper itself often needs a major revision.
Then the switching of templates actually is one line. There is also hope to get a bit further in the publishers system (which ultimately needs JATS XML), because MyST can also export to JATS (which is used for pubmed etc.).
In my CS field, it is common to submit to different conferences at least if a paper gets rejected (because rejections can be very superficial). Fitting for a new conference template often also imposes rewrites, because of different length requirements, changes in length in the new latex template, different accessibility support (e.g., image descriptions in acm) and many more subtle changes. It was a major pain for me. Conferences especially have this problem because they lack major revisions and instead reject. But they are the goto way of publishing.
The journals I interacted with were less of a problem, as they just handed out major revisions. The only rejection I remember was with a recommendation for a more fitting journal.
While I'm not slinging manuscriots left & rught, my coauthor and I got the below linked one into Frontiers last year. The reviewers' feedback was all based on the substance. The formst we had submitted was Google Docs printed to PDF. There might have been a standard Frontiers template or font choice I started from from the outset, I can't remember but I definitely wasn't using their standard reference format in-line in the manuscript, as doing so would have made some sections and tables nigh-on unreadable for the reviewers without a lot of formatting effort.
Anyone in CS will use LaTeX with no problem. But the average biologist who is publishing a paper doesn't want to also learn some esoteric pseudo programming language just to format their papers. They'd rather write in Word.
I've spent so much time for pointless reformatting. With Springer LNCS it's often that you have to work really hard to cram everything into the page limit. Then Springer does some editing in the bibliography, and the article then often overflows by one or two bibliography entries and is published like that, making the effort of meeting the page limit a complete waste time
You're all assuming it's about making papers pretty or fit a template, but the major time sink is length limits.
The abstract has to have a specific number of chars or words (which varies by journal), and it's a critical part of the paper, so you never want to waste a single word. There's also limit for the total number of pages or words, which again needs text rewrites. Sometimes format of references is weird, and needs conversion macros or manual edits.
Reviewers always want more information, more references, more tables, so publishing is an endless puzzle to fit all of that within limit that is always too short to fit everything you have to say.
Having concise paper is of course great, but submitting to multiple places requires editing multiple variations, and that is busywork.
The length limits also forces authors to increase the complexity of their figures to cram as much data into the limited space as possible. I've seen papers with figures that have 10+ graphs all squished into a half page figure.
EDIT: This usually limits proper explanations of the data in the text too because of very limited word counts.
Yup. And then half the paper is fluff, citing similar work (and all the room those citations take up), trying to make an argument what they did is novel.
It's interesting to note the difference between papers 30 (this might vary by field) years ago compared to now. The old papers are mostly substance, and spend very little space on the introduction and political "positioning". Partly this makes sense since things were younger and work tended to be more novel by default back then.
But I also think it's a sign of a sort of petty dysfunction. Academics get graded on their "impact", and citations is part of that. So reviewers want you to cite their results, and your reviewers are likely to be folks with similar papers. So authors go citation crazy, trying to appease selfish reviewers, leaving themselves with less than three pages to describe what they actually did.
And we are living in times where productivity, tight deadlines are more and more common, inevitably people will use generative AI to fulfill idiotic and antiquated bureaucracies.
Not in my experience. The information is only so compressible. You're usually better of if the authors have an uncompressed preprint on their website. Those tend to be much clearer.
What is painful is when one venue has a length limit of 6 pages and the next has a limit of 5 pages (plus one page for references), but when you port the paper from one Latex template to the other it explodes to 8 pages because of formatting.
In all cases, what they want and what you have is "a short paper" but you're going to have to take several hours out of your life and probably make some meaningful changes to the content.
What would be best is if publishers could come together and harmonize their requirements to the extent possible.
Sure, abstracts are a good thing and we do want such compressed information about the article - however, recompressing a 'compressed' abstract from 300 words to 200 words or from 200 words to 300 words (which is usually mandatory if resubmitting elsewhere) is time consuming, lossy, and doesn't add value.
As many people here write, with Latex, most of the technical issues are already resolved. A few other technical issues might arise due to Latex, but they are usually easy to deal with.
However, what remains are rules on page limit, number of words or similar restrictions. And in most cases, you try to fit as much content as possible into the template. So when changing the template, you either have more or less space, so either you would want to add more content, or you need to remove content. Even with the same number of pages and other things being similar as well, due do small differences in the template, it will not match up. That is what takes effort.
But for the aspect on space restrictions, I have heard actually the opposite argument: By having this limitation, you save the reading time of all your peers. Often, the work you did can actually be compressed into such short format. You might need to leave out some details (if you publish the source code and make it reproducible, people could anyway check for all details though, so not really a problem), but you can focus on the actual important part. So, by wasting some extra time and effort of the author, you save a lot of time and effort for all readers by having this restriction.
If your work really does not fit the space limitations, maybe a different conference or journal would anyway be the better fit.
Papers are already structured in a way that makes reading them very efficient, by providing different layers: you read the title, if you're interested you read the abstract, if you're still interested you read the intro, conclusions and maybe charts/tables, if you're really interested you read the rest.
I have the feeling that when my interest in a paper is enough to read everything, I might as well read a couple more pages if the paper is going to feel less crammed (often compression makes understanding harder). Although I might be underestimating the issue.
It's not wasted - formatting things properly is incredibly useful as it allows for effective communication without researchers having to think of every possible ambiguity. The whole point of different journals existing is to allow people to publish to the ones that make sense for them, including the appropriate formatting. If you allow free-form submissions either you're going to need someone who is not familiar with the work to reformat it (and if it's more complicated than what a latex template can handle, it's probably not something you want a random editor doing either), or you publish things as they come in and wind up with a hodgepodge of poorly thought out formatting. Communicating is one of the most important parts of research, and researchers should be spending an appropriate amount of time on it.
All these formatting requirements accomplish the opposite though. Every journal is different, so as a reader there is zero consistency. Even journals within the same publisher like IEEE don't have a consistent format. So functionally, there isn't much difference than authors just making up their own.
Our lab (phage therapy, plasmid therapy, AMR) spends a lot of time doing this kind of thing. So we're working on throwing the full weight of LLMs to help us fit things like length limits, format our citations properly, etc.
Still a work in progress, since tiny context limit makes this so hard (I keep trying to get Claude access, including joining a Claude hackathon, but for some reason I never get access), but I'll be more than happy to share this thing for free so we can all go back to working on the science.
As my PI says: "Science is not an exercise in creative writing"
Please don't. When I don't understand what a human's written, I can often reverse-engineer it by considering what sorts of thoughts would lead to that word choice. That's next-to-impossible when reading the output of a transformer-based predictive text system.
Science is not an exercise in creative writing, so write straightforwardly.
In my opinion, science is an exercise in creative thinking, and thinking is an exercise in creative writing. Apply the >>= operator of the creative monad. QED.
In all seriousness, I think that writing ideas down, condensing them, reading the result aloud and re-editing vigorously is where the real thinking takes place. That iterative process is thinking, for anything complex. It cannot be done by a computer. If you are completely confident in your ideas before you have written them down in concrete form, they are probably not novel enough to be worth publishing.
In many fields of science your ideas don't really matter, it's about reproducible data - sure, ideas and motivation matter for choosing the experiments and interpreting them, but any reader is effectively expected to read your ideas, assume that they're wrong, worthless and discardable, take a look at your data (which is the key part of the publication, not the creative writing) and see if the reader's ideas and interpretations based on that data match yours.
In the end, your thinking is a process, not the outcome. It is a tool to prove an outcome based on data and logic; better thinking and better writing makes the argument easier to understand, but should lead to exactly the same outcome for the reader, just with a bit less effort compared to sloppy thinking and sloppy writing.
Should this be exactly where the journals have paid editors to do the busy work to match the output they want? Shouldn't there be some standard light format to slap the text and images and possibly references(hardest part) and then editors would take care of the rest?
more busy work leads to more justification for larger APCs...
+1 on a light and accessible format, like submitting markdown and notebooks, which the tools around this markdown flavor is enabling https://myst-tools.org/
> reduces the amount 10,000 people need to read the paper
Honestly, I’d wager most papers are read by a handful of people at best. Only the /most/ (top 1% or less) popular authors and papers get hundreds or thousands of real reads.
While changing and improving their content and the "form" of the article is probably time best spent -- probably we would do more to improve the experience for the audience by dumping the PDF and moving to web based articles entirely, structured data, html papers, linked content and customisable viewing experiences would all improve the reading experience massively.
KeenWrite Themes[1] are instructions that tell ConTeXt how to typeset XHTML documents (content) into PDF files (presentation). I made a tutorial[2] that shows how my FOSS desktop text editor, KeenWrite[3], allows users to write in Markdown to typeset a document against a particular theme.
Before it can be used for scientific papers, KeenWrite (by way of a flexmark-java extension) needs cross-references, which, unfortunately, aren't part of the CommonMark specification.
I posit that the vast majority of LaTeX users don't grok how to separate content from presentation. When I asked a question on TeX.SE[4] about how to adjust the line spacing between enumerated items (spanning a couple dozen enumerated lists), the vast majority of people voted for the answer of using `\itemsep0em` to tweak each list individually. The correct answer, IMO, is to fix the problem globally, and not waste time tweaking individual lists.
Even with LaTeX etc, content is not separate from presentation when you write papers. Once you have limits on pages, you have to adjust your content to the presentation to make it fit.
It's unclear from the article if that's representative of either 1) Total time spent formatting or 2) The marginal increase in time spent formatting above what it would take to follow some other common format.
#2 seems less likely, since, well, there isn't really a common format, as the article points out, so there's no default that researchers would use as a baseline.
That leaves #1. But a paper has to be formatted somehow, and that's always going to take time. So the $230M cost the article cites is going to be at least a little smaller when you take out the time it takes to write up anything at all using a consistent style and format, whether it's your own or something imposed on you.
And I can tell you based on my own profession & day to day work, format & presentation is extremely important in making it easier for your audience to consume information, and this is especially true with complex or novel information.
Undoubtedly the industry can and should do better. There should be common standards. It would smooth readability and comprehension. But it's not going to completely eliminate the rough $/hour equivalent this article is citing.
Also consider that different fields will be conveying different types of information, or have audiences with different skillsets, so there's not a one-size-fits-all approach here. Style & formatting guidelines, even made as common as possible, will still by necessity be different in different disciplines.
So many comments about fonts, subscripts, etc. No one cares about that at review time. I've never heard anyone complain about significant time on this.
Major difference in paper _style_ between Nature Brief Communication (2 figures, one page of text) and J Neuroscience full (10 pages and 8 figures). That's a big reformating. Also, Nature has a very specific abstract, which is great for a general science audience (https://www.nature.com/documents/nature-summary-paragraph.pd) but less so for specialists. Similarly, a clinical trial abstract has exact results so that it's easy to scrape. For the y-combinator audience, it seems obvious to me that a MobiCom and a ACM transactions will be very different papers.
Is it something that needs to be fixed? I surmise that Nature Brief Communication, by its name and the guidelines you gave, is about briefly communicating new results. People read it so that they can get brief summaries of new results in the natural sciences. It would do a disservice to the people that read it if they announced that they wouldn't require one page of text and two figures for new submissions, because it required too much work for authors.
If you wrote a novella, and wanted to submit it to a magazine that published short stories, I don't think it would be reasonable to complain that they didn't want to publish your work unless it was edited down to meet their guidelines. Why is it different for scientific papers?
It just seems natural to me that different publications would have different audiences and different goals with what they are publishing. They're read by different people for different reasons who want different things from the publication they're paying for. Naturally that means that if you want them to publish your work, it has to be right for their readers.
Is there still no widely adopted standardized submission format across all of science? That seems like a complete no-brainer, some structured text format that would allow the journals to render the articles as they see fit. Finally an application for XML:
JATS is gaining adoption in publishing, but is difficult for scientists to actually _author_ in. This is one of the main reasons behind tools like MyST (https://myst-tools.org) and Quarto (https://quarto.org/) - both are semantic authoring tools that allow authors to export to JATS but author in Markdown (or in JupyterNotebooks).
Researchers work a lot on their own time, so the millions mentionned in the article don't directly impact the institutions employing them. They are alone in this fight.
This is just one of many, many problems plaguing academic publishing. I've long thought a web platform with a recombination of features borrowed from Github and StackExchange could potentially solve many (maybe most) of those problems.
Last summer I was able to take 6 months off of work to build a prototype and now I'm inching it forward as a side project. It's in early beta, and I'm looking for folks willing to use it to solicit review of their pre-prints and give me feedback and direction on where it take it.
It's open source, open access, and intended to stay that way. If it gets traction I want to form a non-profit, multistakeholder cooperative (governed by its users and the team that builds it) to develop and maintain it. Though, that's a long way off.
106 comments
[ 2.9 ms ] story [ 180 ms ] threadWhat? The journal provides (or at least should) a latex template, you write stuff in that, then you ship it. Last one I wrote in a browser through overleaf, spent zero time on formatting.
edit: Nature's one appears to be https://www.overleaf.com/latex/templates/springer-nature-lat...
Millions of dollars in time wasted on learning other markup language
Then you don't have much experience writing papers.
Formatting is very time consuming. Some want 8 pages, some are unlimited, others want you to use whatever makes sense.
Every journal has a different page length. They expect to see results in different places. Some want a detailed appendix. Others are ok with a website.
The latex template doesn't help at all with any of these.
Reformatting is extremely painful and wasteful.
That said, recognising static academic reference format, then reformatting for the other 4 or 5 standard formats seems like a python project for first year students (to this non-programmer, I'd guess there are lots of edge cases to trip on making it a great learning experience!)?
An additional annoyance is that to save space, journals often want formats that are compact but inefficient to use, for example, formats that are heavily abbreviated and don't include titles. These, however, are annoying to use, and annoying reviewers is generally unwise. So in our case, we actually intentionally submit PDFs for review that use the 'wrong' bibliography item format, of a similar form to the journal's but less abbreviated and including titles and hyperlink backreferences to citations in the text. Then for final submission we have to replace this with the journal's preferred format.
Publishers like Dagstuhl Publishing that can require LaTeX submissions and use a completely LaTeX publishing workflow probably have significant efficiency advantages, but then rely on potential authors being sufficiently experienced with LaTeX that a LaTeX submission requirement doesn't drive people away. This can work well for computer science and some parts of physics, but for other fields it can be a significant burden.
[1]: https://www.nature.com/nature/for-authors/final-submission
https://gwern.net/maze
Yes but what about all this unnecessary gatekeeping? gatekeeping is necessary to maintain a certain level of quality. On this specific issue of formatting gatekeeping. There was no conspiracy. I imagine that the policy solidified over time. The paper wants to maintain a specific look. Authors are upset when changes get introduced while formatting. The paper then says "you can format it yourself if it meets our specification". Now you have to format it yourself.
Thankfully in our web enabled era there a ways to publish without going through a traditional publisher. However you will note that the quality of these publishers is a mixed bag, real science right next to quackery. No one has done the gatekeeping needed to separate the wheat from the chaff.
Which does the paper want more? To be the first with the results or to appear a particular way? Do the appearances help with "the results?"
How much does having a consistent "look" lend itself to the perceived professionalism and respect a journal has?
Scattered additional thoughts on desirability of publishing platforms.
Also, journal publication fees dwarf the implicit cost of reformatting and deliver arguably less value.
Sometimes it's small, like whether or not `\author{my name}` needs to be included in the preamble or in the main body. These are short but annoying. Sometimes it's very annoying. I'm glad that I've never tried to publish in a two-column journal (which might exist in math? I don't know) which would require more substantial superficial edits.
I always assumed that this was a deliberate source of friction, slightly raising the bar of entry so as to discourage a certain sort of submission. I don't say I agree with this, but that's what I thought.
This sounds like the sort of problem that latex would be great at solving: changing the look of an article just by using different document classes. And sometimes it is great --- but some templates are terrible, terrible, terrible. (And sometimes these terrible templates are for great journals! I see little to no correlation there).
There’s no doubt in my mind that it’s intentional gatekeeping. There’s a reason why many papers are being “published” on Arxiv and foregoing all this boxticking.
The whole publishing process is fundamentally broken. Aaron Swartz (RIP) had the right idea all along. Get rid of the middlemen sucking up millions of dollars by paywalling government funded research.
It’s been proven time and time again that a shocking number of studies can’t be reproduced and that the traditional peer review process just isn’t working. It’s a good old boys’ club putting a stamp of approval on their friends’ work and keeping the outsiders away.
When it comes to Math though, I do have some sympathies. There are so many cranks that think they solved Collatz or a millennium problem and don’t even know what they don’t know. It must be exhausting to sift through all that noise in a volunteer, unpaid position. These shibboleths might be a necessary evil.
> Math though, I do have some sympathies. There are so many cranks
Lots of other fields have their cranks, quacks, fringe theorists, and whackjobs. Why give math special treatment?
Math is not so different from computer science in that it seems relatively approachable to a layperson. But there are relatively few mathematicians out there. That same level of spam is being filtered by much fewer people.
To me, it’s akin to why the Linux kernel devs operate the way they do. No GitHub issue page, no Slack, no Discord, banning topposting and web email providers. It’s all based on old school mailing lists, because there’s orders of magnitude more amateurs than kernel devs and they don’t have time for handholding the unwashed masses.
Anyway, I think there’s some middle ground to be found between this extremely burdensome process to submit a paper for publishing, and making it as easy to publish as making a forum post…
I couldn't quite get it to the company standard, but I got it down to under an hour worth of work for our tech writer per release.
In the end it was a week's work for me and it improved our integration tests. Not the constant draw on resources and sanity that I think someone was after.
Once (for a Springer journal) I got called out by their editorial staff, and told I had to follow their dumb rules. So I did. But I would never do this preemptively.
Publishers charge thousands of dollars for a subscription to a single journal. If they want their papers formatted in a certain way, then they can employ staff to do that work themselves. If any for-profit journal tries to insist that authors do lots of reformatting work before even submitting, I will never submit to or review for them again.
Most of academic science as we know it today is structured so that the output is "publishable" and/or helps future grant applications. Incremental improvements are very publishable, but that doesn't necessarily make good science. Grants are awarded to scientists who are consistently able to deliver results, in the form of published papers. I can only really speak for my little corner of science, but from my view, the entire incentive structure of science is broken.
Like, the existing system already has issues with publication being too easy to game, with predatory paid journals, lax standards for reviewing, etc; that's hard to fix, but it's considered a problem - and so any changes making the barrier to publication even lower than that won't be welcome.
And from the perspective of researchers, we don't really want more publications - they're a pain to read, there's an overabundance of poor content that's salami-sliced, we do so many publications because the incentives push us toward this but we recognize that this is a bad thing and we'd be far better off if the same research would be published in fewer, better articles. But we can't, because anyone who does that will be effectively denied resources for future research.
Do you remember which subscripts must be italicized?
Yeah right.
This is YC though.
Up-for-grabs: Untapped $260m market for reformatting scientific papers.
The problem of switching to another journal is also overplayed in the essay, at least in my field. I don't know of many people who "shop" for journals to accept their work. Usually you know the right journal, and if the paper is rejected you just give up on it. Perhaps this is discipline-specific, though. (The essay is restricted to biomedical fields.)
As a reviewer, I've never spent any time on journal rules. If the paper gets accepted, it's up to the technical editor to impose rules.
Most authors employ the proper latex stylesheets from the first rough draft. They also use section headings that fit the journal's conventions, and so forth.
I suppose it is possible that people in biomedical fields (are there any here on HN?) tend to submit to multiple journals before their work is accepted. But I sort of doubt that reviewers reject papers based on things like italics or citation formatting, so if folks are submitting to a lot of journals, maybe their work just isn't highly regarded by their peers.
Basically there is no issue here, at least in my field. It's an unconvincing essay, and not one that encouraged me to waste time looking up the original work.
There are still missing features compared to latex, but what exists seems so much more intuitive.
That is a very broad statement. Do you have any data to back it up?
Do you suggest that it was used in the past but that has changed? If so, why?
https://news.ycombinator.com/newsguidelines.html
Not true. IEEE, one of the largest professional orgs, accepts Latex. So does ACM, a huge comp sci org. I'm not familiar with others, but wouldn't be surprised if many others still accept Latex.
Latex is a giant pain to use - and i say this as someone who still uses it to write scientific articles. Services like overleaf exist but that is sometimes not an option with proprietary work.
[0] https://www.acm.org/publications/proceedings-template
Aerospace generally does too.
Computer vision seems to.
In fact I think most tech related journals do, iirc.
And no one gives up with a paper, although this might have more to do with location than discipline (the academic evaluation system in my country is heavily metrics-based, having indexed papers is what makes or breaks careers). People resubmit and resubmit and resubmit, typically to venues with a progressively lower bar, until the paper is accepted.
In CS we also have competitive conference publications and, at least in my subfield, they don't have these problems to such a great extent as journals (as at least their latex templates tend to be decent, and some of them have the same length limits). One of the reasons why I prefer conference to journal publications.
That doesn't sound like the LaTeX I use. When you switch to another stylesheet, it often changes key dimensions such as page size and the number of columns. Then you have to reformat and possibly even redesign the figures, tables, equations, pseudocode, and so on. And it's quite likely that some packages you are using don't work properly with the new stylesheet, or even don't work at all, so you get some weird issues to debug.
But that's about the technical layout, which isn't the concern here, as the publisher is usually responsible for that. In biomedical fields, manuscripts are usually just Word documents with some basic formatting. The real issue is that different journals have different ideas about the length, content, and structure of the paper. When you switch to another journal, the paper itself often needs a major revision.
https://myst-tools.org/docs/mystjs/creating-pdf-documents
Then the switching of templates actually is one line. There is also hope to get a bit further in the publishers system (which ultimately needs JATS XML), because MyST can also export to JATS (which is used for pubmed etc.).
The journals I interacted with were less of a problem, as they just handed out major revisions. The only rejection I remember was with a recommendation for a more fitting journal.
While I'm not slinging manuscriots left & rught, my coauthor and I got the below linked one into Frontiers last year. The reviewers' feedback was all based on the substance. The formst we had submitted was Google Docs printed to PDF. There might have been a standard Frontiers template or font choice I started from from the outset, I can't remember but I definitely wasn't using their standard reference format in-line in the manuscript, as doing so would have made some sections and tables nigh-on unreadable for the reviewers without a lot of formatting effort.
https://www.frontiersin.org/articles/10.3389/fphar.2022.9099...
I hope PLOS (and whatever similar things have sprouted up) take the hint, and allow non-time-wasting submission formats.
The abstract has to have a specific number of chars or words (which varies by journal), and it's a critical part of the paper, so you never want to waste a single word. There's also limit for the total number of pages or words, which again needs text rewrites. Sometimes format of references is weird, and needs conversion macros or manual edits.
Reviewers always want more information, more references, more tables, so publishing is an endless puzzle to fit all of that within limit that is always too short to fit everything you have to say.
Having concise paper is of course great, but submitting to multiple places requires editing multiple variations, and that is busywork.
EDIT: This usually limits proper explanations of the data in the text too because of very limited word counts.
It's interesting to note the difference between papers 30 (this might vary by field) years ago compared to now. The old papers are mostly substance, and spend very little space on the introduction and political "positioning". Partly this makes sense since things were younger and work tended to be more novel by default back then.
But I also think it's a sign of a sort of petty dysfunction. Academics get graded on their "impact", and citations is part of that. So reviewers want you to cite their results, and your reviewers are likely to be folks with similar papers. So authors go citation crazy, trying to appease selfish reviewers, leaving themselves with less than three pages to describe what they actually did.
What is painful is when one venue has a length limit of 6 pages and the next has a limit of 5 pages (plus one page for references), but when you port the paper from one Latex template to the other it explodes to 8 pages because of formatting.
In all cases, what they want and what you have is "a short paper" but you're going to have to take several hours out of your life and probably make some meaningful changes to the content.
What would be best is if publishers could come together and harmonize their requirements to the extent possible.
However, what remains are rules on page limit, number of words or similar restrictions. And in most cases, you try to fit as much content as possible into the template. So when changing the template, you either have more or less space, so either you would want to add more content, or you need to remove content. Even with the same number of pages and other things being similar as well, due do small differences in the template, it will not match up. That is what takes effort.
But for the aspect on space restrictions, I have heard actually the opposite argument: By having this limitation, you save the reading time of all your peers. Often, the work you did can actually be compressed into such short format. You might need to leave out some details (if you publish the source code and make it reproducible, people could anyway check for all details though, so not really a problem), but you can focus on the actual important part. So, by wasting some extra time and effort of the author, you save a lot of time and effort for all readers by having this restriction.
If your work really does not fit the space limitations, maybe a different conference or journal would anyway be the better fit.
I have the feeling that when my interest in a paper is enough to read everything, I might as well read a couple more pages if the paper is going to feel less crammed (often compression makes understanding harder). Although I might be underestimating the issue.
Still a work in progress, since tiny context limit makes this so hard (I keep trying to get Claude access, including joining a Claude hackathon, but for some reason I never get access), but I'll be more than happy to share this thing for free so we can all go back to working on the science.
As my PI says: "Science is not an exercise in creative writing"
Science is not an exercise in creative writing, so write straightforwardly.
In all seriousness, I think that writing ideas down, condensing them, reading the result aloud and re-editing vigorously is where the real thinking takes place. That iterative process is thinking, for anything complex. It cannot be done by a computer. If you are completely confident in your ideas before you have written them down in concrete form, they are probably not novel enough to be worth publishing.
In the end, your thinking is a process, not the outcome. It is a tool to prove an outcome based on data and logic; better thinking and better writing makes the argument easier to understand, but should lead to exactly the same outcome for the reader, just with a bit less effort compared to sloppy thinking and sloppy writing.
But if they're meant to be consumed by an actual audience, it makes sense to make that part easier.
If one author spending 10 hours reduces the amount 10,000 people need to read the paper by 10 minutes, that's a savings overall that is worth it.
Honestly, I’d wager most papers are read by a handful of people at best. Only the /most/ (top 1% or less) popular authors and papers get hundreds or thousands of real reads.
Before it can be used for scientific papers, KeenWrite (by way of a flexmark-java extension) needs cross-references, which, unfortunately, aren't part of the CommonMark specification.
I posit that the vast majority of LaTeX users don't grok how to separate content from presentation. When I asked a question on TeX.SE[4] about how to adjust the line spacing between enumerated items (spanning a couple dozen enumerated lists), the vast majority of people voted for the answer of using `\itemsep0em` to tweak each list individually. The correct answer, IMO, is to fix the problem globally, and not waste time tweaking individual lists.
[1]: https://github.com/DaveJarvis/keenwrite-themes
[2]: https://www.youtube.com/watch?v=3QpX70O5S30
[3]: https://github.com/DaveJarvis/keenwrite
[4]: https://tex.stackexchange.com/questions/6081/reduce-space-be...
#2 seems less likely, since, well, there isn't really a common format, as the article points out, so there's no default that researchers would use as a baseline.
That leaves #1. But a paper has to be formatted somehow, and that's always going to take time. So the $230M cost the article cites is going to be at least a little smaller when you take out the time it takes to write up anything at all using a consistent style and format, whether it's your own or something imposed on you.
And I can tell you based on my own profession & day to day work, format & presentation is extremely important in making it easier for your audience to consume information, and this is especially true with complex or novel information.
Undoubtedly the industry can and should do better. There should be common standards. It would smooth readability and comprehension. But it's not going to completely eliminate the rough $/hour equivalent this article is citing.
Also consider that different fields will be conveying different types of information, or have audiences with different skillsets, so there's not a one-size-fits-all approach here. Style & formatting guidelines, even made as common as possible, will still by necessity be different in different disciplines.
Major difference in paper _style_ between Nature Brief Communication (2 figures, one page of text) and J Neuroscience full (10 pages and 8 figures). That's a big reformating. Also, Nature has a very specific abstract, which is great for a general science audience (https://www.nature.com/documents/nature-summary-paragraph.pd) but less so for specialists. Similarly, a clinical trial abstract has exact results so that it's easy to scrape. For the y-combinator audience, it seems obvious to me that a MobiCom and a ACM transactions will be very different papers.
Fixing the format doesn't fix this.
If you wrote a novella, and wanted to submit it to a magazine that published short stories, I don't think it would be reasonable to complain that they didn't want to publish your work unless it was edited down to meet their guidelines. Why is it different for scientific papers?
It just seems natural to me that different publications would have different audiences and different goals with what they are publishing. They're read by different people for different reasons who want different things from the publication they're paying for. Naturally that means that if you want them to publish your work, it has to be right for their readers.
Is there something I'm missing?
https://www.xml.com/articles/2018/10/12/introduction-jats/
Having semantic authoring tools also means that you can easily change the template -- saving a ton of time for the author. For example, about 400 journal templates here: https://myst-tools.org/docs/mystjs/creating-pdf-documents
Researchers work a lot on their own time, so the millions mentionned in the article don't directly impact the institutions employing them. They are alone in this fight.
The petition (https://www.change.org/p/simplify-manuscript-submissions-in-...) linked in the article barely received 100 signatures in a month.
Last summer I was able to take 6 months off of work to build a prototype and now I'm inching it forward as a side project. It's in early beta, and I'm looking for folks willing to use it to solicit review of their pre-prints and give me feedback and direction on where it take it.
It's open source, open access, and intended to stay that way. If it gets traction I want to form a non-profit, multistakeholder cooperative (governed by its users and the team that builds it) to develop and maintain it. Though, that's a long way off.
The beta is here: https://peer-review.io
The source is here: https://github.com/danielbingham/peerreview
A more in depth description of how it works is here: https://peer-review.io/about