Ask HN: Best way to publish papers as a non-scholar?
Is there a way someone who is not a PhD can publish papers without being marginalized by the community?
Do you have any advice on how to write better quality articles?
Do you have any advice on how to write better quality articles?
Peer review is a highly flawed process that has an empirically small effect on paper quality [0]. Moreso, rejection of an article is a poor or completely useless predictor of the ultimate impact of the article if it is published.
Getting a paper through peer review is slow and frustrating. You may get reviewers who reject a paper for incorrect or even bizarre reasons. If your paper casts doubts on an anonymous reviewer's work, you are going to have a hard time. It's just not fun.
What are the benefits? Tech employers generally don't care about publications, just the work involved. Clients probably don't care. NGOs and government don't care either. The only people who care whether something is published are academics, in my experience.
Perhaps you are in an industry where this isn't true, in which case you might have more incentive. But don't assume people will care.
You might have a microscopic chance of your paper becoming highly cited, but without a well-know co-author, that is unlikely even if your paper has important content. Most scientific papers fail to make any kind of splash, with numbers of readers numbering in literally the low dozens. After all that work, chances are that nobody would care.
A better solution is to put it all on a blog, and/or ArXiV or one of the other preprint services for other subjects. That way your research will be available unpaywalled for anyone.
I find it especially important if the paper has any danger of being perceived as crackpot material. For example, my paper "A type of simulation which some experimental evidence suggests we don't live in" [1], recently featured on the HN frontpage, took years to prepare and finally publish, even though it's literally less than one page in the journal. With something as 'pop' as whether we live in a simulation, I wanted to be sure what I wasn't just deceiving myself or something.
Open, non-anonymous comments on preprint services hold much more promise. Peer review without anonymous gatekeepers.
I've had cases where my initial thought was "this reviewer is an idiot!!!" When I calmed down and forced myself to give them the benefit of the doubt, I ended up realizing they were right, and they improved my papers.
It effectively ends the discussion, because ACS doesn't publish nonsense and everything is nitpicked to death. It may turn out to be garbage later on, but at the time it was sufficient.
I think the primary issue with PR is its fundamental assumptions are flawed. Papers authors are NEVER anonymous and can't be while reviewers make it obvious who they are.
BTW, that you have a paper that took you years to prepare puts you in the far tail of the distribution.
Certainly the journal should catch garbage, but authors shouldn't rely on this.
This is not strictly true - if you are trying to emigrate, some visa types may be easier with a published paper (see: HSP visa for Japan).
For the racket of academic publishing to be upended, there has to be motivation. This remains the unsolved problems. Opportunity and capability already exist in forms such as blogs, pre-print servers, etc.
However, cachet is inevitably conferred by powerful monopolies. For example, some universities are trying to upset the Elsevier applecart by moving their publishing to in-house presses. But this simply shifts the power from one giant to another.
Producing an open platform that avoids monopolising but still confers favourable reputation seems very hard.
The obvious answer that you might not like very much is "write a paper and co-author it with an established group." If you share a lede with an established group, that's immediate credibility.
Journals happily accept manuscripts from first/corresponding authors who don't have PhDs. Graduate students publish regularly; depending on your program this is effectively obligatory.
Self-funding publication in a peer-reviewed journal will raise eyebrows -- so see my first suggestion, which offloads the often substantial publication costs onto a grant. (Our group's last publication cost was ~$4500, for a benchmark. That's higher than normal but it was open access with 6 color figures). Crediting a grant for the work also lends credibility; someone gave you cash to do this thing, so obviously they don't think you're totally nuts.
Publishing preprint-only on the arxiv is well established. I think you need someone to vouch for you the first time, but that shouldn't be too hard depending on the field.
Depending on the field, single-author maunscripts are either basically normal (parts of CS, much of mathematics, some theoretical physics), a giant red flag (experimental physics, biology-adjacent fields), or somewhere in between (economics?).
Do not ever use a vanity publisher for scientific articles. If the journal ever appeared in Beall's list, run, because it's a mark of shame.
Finally, don't necessarily attribute your marginalization to not having a PhD -- the politics of publication are ruthless no matter who you are.
That is _not_ always the case. In the biosciences, "conference paper" means "an undergrad could do this half-asleep, and probably did; nobody actually cares much."
I would also consider who you want your audience to be and the scope. If it is something that you want a non-expert to read, consider a blog post. On the other hand you may want to consider something like a patent if it is something you want to monetize. Lastly, consider peer reviewed journals if it is intended to be consumed by domain experts. All of this is from physics research perspective and may be different depending on your area.
And as I only want to contribute and am not necessarily looking for fame, I really do not have any problem with having co-authors.
Regarding the venue and costs, in CS, some conferences are better venues than most journals (impact wise, at least), and attending one might be a good way to know the people in the field. In any case, you'll be expected to pay for the registration and present your paper, too.
A lot of papers, most papers I read in CS, involve incremental advances of a given technique. Usually the papers involve describing the problem, the describing how it's solved so far, describing how their solution is different and only then providing particular, computational details of the method.
And my impression is this incremental style comes because most researchers do incremental research, extending the ideas and approaches of mentors or colleagues.
Which is to say, if you, an outsider like myself, have a good idea for, say, a new algorithm, it might be useful to write a blog entry explaining it's value aside from any paper you write.
And as said elsewhere, avoid "predatory" journals - they apparently don't give credibility or get attention.
worse. They taint what credibility you have. It’s like buying a degree from a degree mill “school”.
Assuming double blind review, the barriers are going to be with respect to the contribution itself, or the writing. For a paper to get accepted in my field (CS, HPC, PL, compilers), first of all it needs to present a novel and interesting contribution. I occasionally see papers that are pretty obviously coming from industry where it's pretty apparent that the authors just have no idea what a contribution entails.
This is the sort of thing that you grow to just have a gut sense of by doing a PhD. Short of that, you could get some of this sense yourself by reading papers, though it would be far easier to collaborate with someone experienced in the field you can find a person who is willing.
Nature has offered the option for two or three years; the problem is that ~nobody uses it because unmasking is trivially easy. As of last fall, 12% of submissions were double blind.
The hyperspecialization of biomedical research means that you get obligate regulatory capture, or something quite close to it. Mostly people figure that trying to paper that over with things like blinded reviewers is a waste of time.
I'm not actually convinced unmasking is "trivially easy". Anecdotally, as a reviewer for the American Journal of Epidemiology, which is double-blind, I've had 4 cases where I've gone "I totally know who did this".
In all four I've been wrong. In one case, I actually knew the authors personally.
Now, it's possible I'm just uniquely bad at this, but I'm not sure it's as easy as everyone thinks it is.
My advisor is in PL too, and he's commented on the exact same thing. Every time he's on the PC for a conference, he's able to pick out a couple papers that he's confident are written by non-academics. He says it's a combination of the lack of thorough knowledge on the subject (attempting to present an idea as novel when it's been done before, but under a different name) and the writing style (not academic enough). It seems that these are things OP should definitely look out for if they want to successfully self-publish.
But you still need to have your paper sent for review... Couldn't the editor reject it outright?
In CS, we publish in conferences. Conferences don't have editors in the sense that journals have. So there's no initial filtering step. Anyone can submit, and anything that gets submitted will be read by at least 3 or 4 reviewers (who don't get to see the author list). Anyone who does get to see the author list (e.g. the program committee chair) is not involved in any decision making about specific papers unless there is a very serious violation of some kind (which in practice happens very, very infrequently).
Note to self: open a lucrative business submitting bulk spam to CS conferences! (Just kidding)
In some fields there is a noticeable proportion of "industry papers" with valuable contribution coming from people outside of academia without advanced degrees. Sometimes it's obvious in the reviewing process by having problems that are less frequent otherwise (i.e. they lack the experience of "how to write a decent paper according to the standards in this field" that's taught and practiced during the PhD process), but those generally are fixable and so they're just pointed out doing the reviewing/correction process.
I have never seen a case where an author who does not have researcher credentials is marginalized.
However, papers with bad science are rejected and in the worst-case, the authors are blacklisted. It does not matter if the authors have stellar credentials or none to start with. So aim to write papers with good science and forever stop worrying about getting marginalized.
There are several books that explain bad science. I recommend "Craft of Research", "They Say I Say", and "Demon Haunted World" to start with.
The most effective method to get your paper accepted is to make the experimental methods explicit and the data public. This will make your paper much more scientific than those published in conferences with poor-reproducibility checks.
Professors are hungry to write good papers. Conaact a professor who works in the same community to review your paper in return for co-authorship. They will gladly agree if your paper is aligned with their interests.
All the best!
1) The work is sloppy and/or has obvious errors.
2) The work lacks clear context (i.e. citations and framing - "why should we care").
3) The structure is not idiomatic.
4) The writing is not idiomatic.
I agree having a "traditional" collaborator can help, not because of the credentials so much as avoiding issues 2-3. It also will help to read a large number of well written articles. At minimum, if you are thinking of publishing your own ideas you should have read every core/significant related paper done in the last decade or so, and as many of the other related ones you find interesting as you can. That will help you with both issue #2 and with improving quality.
If it's in an area with an active preprint server like arxiv, by all means submit there. If your idea is interesting and you've called it out well, you should get some feedback.
On writing better articles my advice is to simply start with the end in mind. Explain the idea how you would to a child without using childish language. Leave the thesaurus aside, don't use industry language just cause you have to. Don't not use it just cause you think you'll sound pretentious.
Hope that helps, hopefully I don't come off sounding like a knob. Good luck.
arXiv is a pretty good option for disseminating information, but most papers published there don't receive as much consideration as those that have gone through a peer review. Therefore, only the most obviously-groundbreaking papers end up accruing a lot of citations there.
If you want to build a reputation, you'll need to target peer-reviewed conferences, and probably those that have a double-blinded process. This will allow your work to stand on its own, although you'll need to ensure that it conforms to the structures/patterns/shibboleths of the academic community. They best way to learn these, if you don't already know them, is to read as many papers as you have time for in a domain as close to yours as possible, and then replicate those formats.
If you're looking to increase your citation count, this is difficult to do without really groundbreaking research. Some manage it by doing "citation sharing" with collaborators, but this is (A) frowned upon, and (B) difficult to get going if you're not in academia.
In any case, you'll probably want to learn LaTeX if you haven't already; this is a pretty necessary first step for publication, either at arXiv or an IEEE/ACM conference. Edit: This is because most conferences have a template that you must follow, and these templates are provided in LaTeX format.
In my PhD I put some work on biorXiv that I never bothered to put through peer review and was pleasantly surprised to see it cited by a peer-reviewed journal article.
If you're a member of a recognized academic institution (as determined by email address I think?) then you can skip the endorsement step.
The thing that surprised me the most was that people watch what gets posted on arXiv, so they someone will probably read your paper if its an active research area, you may even find people discussing your paper on Twitter without ever promoting it.
If you have any institutional affiliation (eg even big tech companies qualify), you can post to arXiv. And it should be pretty easy to get someone to endorse your posting if you don't, it's basically to stop total nonsense getting submitted.
My route to learning how to write quality papers: Find an expert in your field to tear your paper apart, and then get to work rebuilding it. Rinse and repeat 20 or so times. Additional experiments may be needed. You'll probably have a good paper at the end.
This of course is much easier to do in grad school. But can be achieved elsewhere.
One thing you can do is make sure you're still on the map. This could include, for example, making sure Crossref tracks your blog so we can record citations / references from your work to other work. Our service for tracking non-traditional publications, Event Data, is still in beta, but you could add your blog to the list:
https://github.com/crossref/event-data-enquiries
http://www.crossref.org/services/event-data
You could also publish your article on FigShare, which means it will get a DOI and thereby be more easily citable by others.
If you hadn't guessed, I'm an employee of Crossref.
1. Introduction. This not only lays out the problem you're addressing, but also locates it in the context of previous work. This is important for a few reasons, not least that the people reviewing your paper will probably expect you to have cited them there. But also, it shows that you understand your work in the context of a broader scholarly effort to advance your field. If you think your work is truly novel, you probably haven't done enough reading to find parallels in the literature. The introduction should also briefly state your results -- this isn't a mystery novel, you're not saving up for a big reveal.
2. Methods. This section describes the new thing you did or made. What it is, how it works and why. You're still going to be putting a fair number of citations in here, but they'll be a good deal more focused than in the intro. Often I see people citing their own research group's previous work here, because you're building on something the group already did.
3. Experiment. This section describes what you did to evaluate your work. This description should be detailed enough that somebody else should be able to repeat your measurements.
4. Results. This section should have the most figures and the fewest citations in it. It describes what happened when you did your experiment.
5. Conclusion. Here you explain what it all means and how it ties back to the broader scholarly effort to advance your field. Where the intro states your results, the conclusion restates them and puts them into context. The conclusion also usually talks about future research directions suggested by the work you've presented.
A lot of grad school is about learning how to write papers like this.
1. If you imagine the knowledge in your field as mountains (like the Rocky Mountains, or the Alps), then you need to describe very clearly where exactly your contribution is located. Which summit do you elevate further? What are the neighborhood summits (alternatives to your solution)? Pointing this out in the introduction shows the reader that you are part of the community. It doesn't matter then that you have no formal academic position.
2. Choose the venue wisely. There are two factors to consider: a) it makes sense to publish where the work you build on and/or compete with was already published. You know these venues if you have done step 1. Some authors of your 'neighborhood' papers will probably be your reviewers. b) if you think that you have a significant contribution, select a tier-1 venue to maximize impact; if you think you have a minor contribution, it makes more sense to go for tier 2. Seek advice by a researcher in your field which journals and/or conferences in your field belong to which tier, and ask them for their opinion which venue would be suitable.
3. When you have chosen your preferred journal/conference, identify the 10 most-cited papers from the past 5 years, and carefully analyze their structure and the used methods. Use them as role models for your own paper, together with your 'neighborhood' papers from step 1.
4. Your paper will likely be rejected initially. Read the reviews carefully. Some reviewers are crappy and just "don't like" your work, but many reviewers raise helpful and constructive points. And even the crappy reviews give you hints on how the community works, so you can revise your paper and try again somewhere else (or the next year, for a conference). Don't give up quickly, but also expect that you will have to invest significant effort on improvements.
My experience is about publishing in CS, not sure if my suggestions would also apply to other sciences.
- Many conferences use blind peer-review.
- After you've finished up with the work, you can ask a PhD to provide feedback/editing in exchange for co-authorship.
* Choose and Know your community!
When you choose a venue like a conference, you are writing an article with the intent of being read by a very specific subset of the research community. You can imagine each subgroup having a set of "interesting conversations" around a narrow set of topics or methodology that are deemed important by the community.
So ... you have to convince your audience that you are contributing something to their conversation. Go look up papers in the previous edition of the conference, dig into their references, and frame your contribution in the term of the conversation they are having. In this way it'll be easier for everybody reading the paper to understand what you are doing.
In some cases, it might be that you are telling the community that they should care about this new problem of yours, but if it is completely unrelated to their discussion, it'll be an harder sell.
Also, be aware of the style of the community: are they interested in experiments backed by strong theoretical work? Or are they more interested in practical experiences, without caring much about theory? A decent paper accepted in a theoretical conference might actually be rejected by a more practical conference (and the other way around).
The standard way is that PhD students (often non-scholars) team up with more senior, well-read researchers (i.e. experienced scholars), with the latter at a minimum explaining and framing the work and in many cases guiding and shaping it.
You get marginalized by the community for bad scholarship just as proposing crappy conference talks will lead to "marginalization" by the conference organizers: there's already so much content to consume, and scholarship and other requirements serve to shape and organize people's contributions in a way that means the community as a whole can make progress.
As for advice on how to write better: Read a lot. Get rejected a lot and learn to read between the lines. If you want a friendlier environment, aim for workshops rather than major conferences. Don't be afraid of putting in the work that's necessary for a good paper.
Regarding "how to write better quality articles?" one key part is reading a lot of high quality articles from the field. Of course, there are other things yo may want to do, but again, it will depend on the area.
Another thing to do is to hire an academic editor. Even though it might not get everything right, there is nothing like peer-review quality feedback on a paper that highlights both the strengths and weaknesses of your writing. One thing to keep in mind is that a peer reviewer will be less likely to tell you that your whole paper needs restructuring. They are often instead focused on "within-the-box" improvements.
Seeing that a paper needs to be completely restructured can stem from being a good non-fiction writer in general, rather than being a good "paper-writer." It's a general skill relating to effective and efficient communication, and it can be improved by writing blog posts, more feedback, and writer's workshops. Similarly, to publish a paper without a PhD, you are probably going to have to become, at the very least, a "lay-scholar" in the field you are thinking about anyways. Maybe you are already well on your way.
Also, I went through a phase of thinking that I was special due to what I was attempting to write and get published. I'm pretty sure that I was looked at oddly by academics during this time. Maybe this is common for those of us trying to publish a paper and who are right out of college. My point is that this phase gave me the energy to submit to peer review, which was great. At the same time, I'm very grateful to have outgrown it. I suspect this phase is something many people go through; it's common.
The key to having your work published is to know the field within which you are publishing very well, cite the other papers concisely and accurately and make a contribution to that body of work. This, in a nutshell, is the process of working to gain a PhD.
As a general rule most journal papers evolve from conference papers. Most conferences in CS/SE welcome "industry papers" aka non-academic contributions. You could join the IEEE and/or ACM and subscribe to journals that match your area of expertise.
Once you read widely you will identify who the key contributors in your field are. You could contact them with a synopsis of your proposed contribution and ask whether they might be interested or know of somebody who could be. In this way you could connect with people who will collaborate with you to increase the quality of your articles.
Science is a social endeavor. If you don't have colleagues, go meet people!
As a referee, it doesn't matter at all to me whether you have a PhD. Your institution may matter a little bit as context, but not a lot. It is the science that matters.
As with any paper, I'll read the abstract, read the introduction, look at the figures and the captions, glance at any short equation, and read the conclusion. By the end of that process, I will know whether or not the paper is worth further attention. The refereeing process is generally an investment of several days of my time, so I do it with care.
Do the work, talk with people, get feedback, and repeat.
1. Be in the process of getting a masters degree. No one really cares at the end of the day as long as you are a graduate student affiliated with a decent school (this may be field-dependent). In CS you can easily publish for example as a masters student at brown -- I have a friend who had no problem doing this with no co-authors and never went on to get a PhD, but he was at an Ivy League school.
2. Find an academic friend or a former professor/adviser in your field and be his/her co-author. This has the added benefit of potentially increasing the quality of the research if they agree to edit/help you with it. Could be combined with option 4.
2. Be a former academic and publish to ArXiv using your old academic email address (not really publishing, but at least people can cite it at that point).
3. Publish to ArXiv without an academic email address. People will be able to cite it, but you won't be "published" per say (in the deep learning community many, many citations are to preprint servers). People not familiar with academia won't know the difference and will still be impressed if you say list it on your resume under publications.
4. Form an LLC, make a good website about how your company does research in X field, and self-publish as a "white paper" (I took this approach a few years ago, but I no longer maintain the website). In some industries, much of the significant research is contained in white papers or company funded papers. It might also be possible to submit to a conference/journal under the auspice of a company -- I don't know how that works though.
5. Get in touch with the journal/conference you want to submit to and explain your situation and ask for advice. Depending on the venue they might be very accommodating.
A note on "funding". At least in the CS community, funding is often not needed to conduct groundbreaking research as it's 99% of the time just you sitting at a computer. There are exceptions, but I think in CS being self-funded isn't going to raise as many eyebrows as in other fields.
There must be a result or at least a clear contribution. Sans the result it's really a poster or a think piece, but really you don't have a publication. A publication is to demonstrate a new piece of confirmed knowledge; an observation, measurement, analysis or proof.
The quality bit is "how good is the way that you are conveying this new knowledge" is it sewn up, is every doubt closed, is the detail all there?
In many fields, if you have the resources to do the research you wouldn't be asking this question. And fields where single author, more creative (let's call it) work is acceptable also seem to be areas with a broader audience beyond readers of narrow, often paywalled journals and could be more appropriate as a book, talk, blog, or arxiv post. And if you're not trying to be an academic, you don't need publications for treading water in your career.
(I'm not being critical of the ambition. Even if the reason is just ego or getting accepted into a group of people you respect, that's good enough reason for me!)
All I would advise you is to embrace the unknown and strive to learn more everyday.
Whichever path you chose, I wish you good luck.
I suggest, write a good paper, look for appropriate journals, and submit to one of them. Maybe speed up the process a little: Send a copy of the paper (or PDF file, as they wish) to the editor in chief with a cover letter not making a formal submittal but just asking if, first glance, might this paper be of interest for their journal?
I never paid anything, no page fees, etc. to publish.
None of the papers, co-author or my sole author papers, got rejected.
Here are some hints that might help:
(1) Write the paper, especially in the abstract and the first paragraphs, like you really know technically just what the heck you are doing. E.g., I started one sole authored paper where I mentioned that a derivative I was taking was a "dual vector" -- not everyone who writes such applied math pays attention to duality.
(2) In each of my sole authored papers, some of the key topics, prerequisites, etc. were advanced and narrow enough that I'd guess that less than 10% of the editors had all the prerequisites.
(3) I suggest that write applied math, mathematical statistics, and computer science making important and appropriate use of some relatively advanced pure math.
(4) Generally I suggest that just write applied math, with nearly all the content in theorems and proofs, for mathematical statistics, computer science, machine learning, artificial intelligence, etc. The usual criteria for publication are "new, correct, and significant", and new and correct theorems and proofs are big steps forward for these criteria. If the theorems are also, in the paper, relevant to some applications or an applied field, then that can help with "significant".
(5) Know quite well just how the heck to write math. A good way to learn this is to have the equivalent of a good undergraduate major in pure math.
A good, first start on such writing is a theorem proving course in abstract algebra, one that starts with sets and foundations and, then, all based on just sets, develops groups, rings, fields, the rational, real, and complex numbers, vector spaces, linear independence, linear transformations, subspaces, null spaces, quotent spaces, duality, the adjoint transformation, eigen values and eigen vectors, the Hamilton-Cayley theorem, inner products, Hermitian and Unitary operators, maybe group representations. Then have a good course in linear algebra, e.g., from the classic P. Halmos, Finite Dimensional Vector Spaces (one of the best writers of math). Then a good course in analysis, e.g., from W. Rudin, Principles of Mathematical Analysis, with highly precise writing. With a few more such good, pure math texts, courses, etc. where the homework is essentially all theorems and proofs and graded by a good mathematician who cares, one will no doubt learn how to write math, and the learning will show.
In my case, I got a good start in a course in abstract algebra, with a good prof who did well grading my papers, and then learned the rest just by studying really good writers, learning how they wrote, without more good grading.
How to write math is no big secret: There're a LOT of beautifully written math texts on the shelves of the research libraries.
* Read a lot of papers with a related work. Unless you have read and understood at least half a dozen of these, you are likely not ready to publish new work.
* Learn technical terminology and make sure how to use it precisely.
* Identify lose, unverifiable statements. Precision and clarity is super important.
* Know the history of the work. Make sure you note folks who came before you. Describe how your approach differs from others and/or builds on other. Failing to cite is one of the top reasons for rejection even for professional researchers.
* Get some theoretical grounding of your approach. Formalize your hunches and intuition in math equations that can be manipulated.
* Design experiments that shows your approach delivers better result than other under conditions you claim.
* Study the layout of other papers, their style and emulate them. Pay attention to how figures, data tables are describes, conclusions are made, claims are stated etc.
I often say that the primary job of researcher is to read papers and stay up to date and secondary job is to make new discoveries. The way developers are supposed to spend major chunk of their time to write code, researchers are supposed to read papers. So familiarizing current state of the art and knowing what has been done before, understanding pro and cons, internalizing techniques, learning related math skills etc must be your absolute top priority for the area you want to contribute.
If you have never published before, writing academic paper is a major chore. It can easily take up 2-3 weeks of your time in ironing out details after you have good results. However publishing paper is also satisfying. You add the knowledge pool of humanity, your contribution can be built upon by others, you get noted as "first people who did X" and so on. However don't expect to build career as researcher - that strictly requires PhD.
On a side note, there are bunch of non-academic or semi-academic conferences which have less rigorous demands. So you can try to publish there as well. I've seen lot of developers publish system design related papers at places like KDD.
2) Put PDF on your domain.
3) Put your domain behind something like cloudflare.
4) Publish links to PDF. Ensure you have a method of feedback: email is generally preferred and include your name and email in the PDF.
You could publish on third party services but consider them non-authoritive and ephemeral. If you use them as primary distribution you will get burned in the future.
Thanks!
Since many journals have double-blind peer review, if the OP’s work is sound, then there is no reason he should not go for journal submission. In terms of building a reputation, that is a lot better than self-publishing, the domain of cranks. In my own field of linguistics I know several people interested in the subject who post their own PDFs on their own websites or on Academia.edu, and though they try to format those PDFs to look serious and respectable, these people creep most actual scholars out and we try to avoid those writings.
Finally, peer review is not just gatekeeping, it leads to better papers. If your work is publishable, then the peer reviews will often suggest ways you can clarify your argumentation, and they will point you to interesting citations that you might have missed. You miss out on all that by self-hosting.
Submitting to a journal is fine, but own your work on your own site.
Why you think that the people who self-publish are more worth looking at than people who publish in journals?
And while it may not be true of cutting-edge STEM fields, most of the must-cite literature in my own field is not available digitally and is in fact held at only a few libraries worldwide. People are unlikely to have access to it unless they are already closely involved in academia, and in that case they will be keen on journal publication for career advancement. Consequently, self-publication strongly correlates with not having an awareness of the standard literature.
I guess a lot of experts in their respective domain still spend a fraction of their time reading other information sources with low signal to noise ratios: news riddled with advertisement and propaganda, perhaps sports, fiction books movies or television series. EDIT: addition: If they spent a quarter of that time sorting crackpot papers the signal to noise level could be fixed, and they'd still have 3/4 of the time for their usual news binge or whatver)
I'd actually be more than happy to see some tax payer money spent on the following system: authors withoud accreditation can sign up on a government hosted site, and provide their articles (or links or p2p hashes of them). Accredited domain experts can participate and earn money by getting assigned 2 or 3 random papers in the same domain but from probable cranks. They simply sort them in credibility. This way probably crank papers get scored, but the top 1% eventually floats up. Scientists/experts inbetween jobs, or out of office hours can earn money on the side, and BS gets seperated from intesting ideas or insights. Signal to noise level near the top is good.
Edit:
I hereby hand out the idea for free, if you feel like making a startup:
Crack Crucible:
- host (openly tongue in cheek) "crackpots" and their papers, self-categorizing in domains
- domain experts can sign up, the site generates a unique string, the expert inserts the string on his/her researchgate profile, the site scrapes the expert's researchgate profile. (hereby delegating 'expertness' to researchgate), they too self-categorize in domains?
- the site randomly selects 2 or more papers for the expert's applicable domain, weighted by inverse word count of the paper (conciseness is rewarded with opportunity to be seen)
- the expert reads the 2 or more papers and is only asked to sort/rank them, without any implied support since most of the time all the papers wwere nonsense, yet the site demands sorting them by (in)credibility
- the result is interesting outsider papers by domain (without the coming and going of news cycles, since the papers are selected at random, and hence undergo brownian motion up and down)
Granted there are "gems" to be found anywhere but it's still largely far too much of a waste of time for most people.
See also: "the attention economy". What can you offer that others want more of - news, publishing, self-publishing, tweets? What's the cost and reward for each party? Who is paying for the food and bills, and how?
The real problem with this isn't that there is nothing worth reading being presented, but that the signal to noise ratio is just terrible. You can waste an inordinate amount of time looking fruitlessly - I can't blame anyone for deciding it isn't worthwhile.
1. Self-publishing the document under a CCO license (vital!) as an easily-navigable text on my own blog, snazzy PDF and datasets attached. Then I made an ~1h-long video summarizing the entire report and plopped that on YouTube (with proper timestaps that took forever but oh so worth it), linking it to my blog and back.
2. Next I shared all these on ResearchGate, Academia.edu, and, thanks to this post, Figshare (for the DOI primarily). I had it up on Medium but that just looked ugly for such a huge piece so I took that down (~50 pages without citations). I also shared it on some other niche decentralized social networks I frequent (eg: Diaspora). I would probably also share it on Twitter/Facebook/LinkedIn if I was still there.
3. The last steps I'll be doing shortly: emailing all my patients that I've treated with relevant issues over the years with the blog post (~1000 people), as well as all the main authors I'd cited (~25 folks) and all the therapists I've worked with over the years (also ~25 folks), plus the few relevant professional associations I'm a part of (at least 4, each of which has a reach of a few thousand therapists via newsletters).
I'm doing all this because, applying what I've found, I'm able to resolve muscle tension headaches in a matter of seconds instead of minutes, and I want others to learn how to do this too -- and build on it. Once conversation flourishes with all the above I might look into publishing with a journal (higher sphere peer review is still very useful!) but, as a clinician, this is a lot less important than its application. I'd already had some profs interested in helping me turn this into a journal article but it was crucial to me that this document I'd created remains 100% in my property, mostly so I could then release it under the CC0 license without penalty. Having published the official thesis on my terms like this now, I'm no longer worried about this ever being a threat.
Some more context: my MSc was from a little-known EU program in sport psych, and deals with the crossover between that field and massage, where I've been working for a decade now. And so I say what all the (mostly CS/eng/mathy) folks before have said: your mileage may vary. You can see where the rubber meets the road on https://spmx.ca/trp, and the WP plugin that creates those great collapsible headings is https://wordpress.org/plugins/olevmedia-shortcodes. And why I strongly advocate a CC0 license wherever possible: https://spmx.ca/wp-content/uploads/2018/03/00-headaches.png ; )
Regarding writing better: write more. Remember the famous pottery parable (https://kk.org/cooltools/art-fear, Ctrl+F "ceramics") and the old go saying: "lose your first 100 games quickly". If rejection is a real fear, watch https://www.youtube.com/watch?v=-vZXgApsPCQ. It really doesn't matter, especially if you're making this world a better place. Good luck!