Interestingly, Mendeley [1] tried to be (one of many) social platform with the specific aim to find more easily relevant papers for you. Unfortuantely, (or fortuantely for the founders?) it was bought recently by Elsevier, so I'm curious to see what will happen to it. Finding good relevant papers is still a difficult task imo.
I want to work on a website for all sorts of publishers, invite / interview process to post as to have no shitposting. Users will be able to register and set a monthly reading budget, which will be divided equally amongst all the authors you read in X time period.
I want this service to have an API so you can easily push content to other platforms like Tumblr, facebook, or your own personal website, housing all your content on my site.
I would use this. I believe in paying people for their hard work and I want the money to go directly into their pockets. No overhead.
I see a bigger problem: the fact that things like citation counts are being used as a metric. We should be judging researchers by whether or not they are solving problems (or if they are helping other researchers solve problems -- like developing new research tools and techniques, or publishing negative results that tell other researchers what to avoid). It makes no difference whether a paper is cited one time or a hundred times. It makes no difference if a research publishes one big paper or ten smaller papers. What matters is whether or not progress is being made.
But this avoids the central issue - how can we know if a paper helps people make progress on solving problems? Citation counts are just a proxy for exactly that - to move away from them you have to come up with a better proxy or evaluation method.
Conferences are a good starting point: researchers meet each other, share work, judge others' work, and discuss research. I am not saying it is a complete solution, just that citation counts are a terrible solution that encourage the wrong things (and that we need something better).
In the UK universities' research output are assessed by HEFCE using the "Research Excellence Framework" (which replaced the Research Assessment Exercise) [1]. Sadly, it suffers the same problem of being another metric [2]. It's a time sink for academics and it fosters a cutthroat atmosphere in the run-up. I heard some unconfirmed rumours that academics not submitting case studies to the REF will have their contracts re-assessed.
I think this exposes the bigger problem of metrics in general. It's hard to determine how much impact research has, particularly in the short term. I'd be interested to see what the outcome of the REF is, but I remain sceptical and apathetic towards it, as does everyone else I've spoken to.
I agree with you that using citation counts as a metric is flawed.But I was primarily suggesting that we need "HN for science research papers ",not a particular way that it should be implemented.
I've never noticed a correlation between school prestige and paper citation in my field, of course that doesn't mean there isn't one. I'm curious if you have anything to back up this statement or if its just anecdotal?
I've come across quite a bit of anecdotal evidence as blog posts (Damn it, can't find the links now) and scientists will often confess to being biased about prestige while reading articles. It's also probably the reason why mathematical papers do not print author affiliations at the front of the paper (they are listed at the back).
It's simply a matter of choosing where to invest your scarce time. A paper from prestigious authors/affiliation presumably has a better chance of being worth your time. So such papers get more attention and hence more citations, even if they might not have been the most novel or the most comprehensive. The same also applies to papers in high impact factor journals where citations have some kind of a positive feedback loop.
It's just anecdotal, but it is a lot of anecdotes. Both of my parents are professors, one at a much more prestigious school than the other. I've worked for a regular publisher, an academic publisher and I went to graduate school myself, so I've had a chat about this exact situation many times, though I admit I have no hard data.
I would argue that this "ruinous competition" is a direct implication of primary, undergraduate, and graduate institutions.
I've said this before on HN (there was another thread about science about a week ago), but I believe that schools have failed in doing what they are supposed to be doing: teaching. Instead, they are slowly turning into Olympiads - into fierce competitions that pit students against other students solely for the purpose of producing some kind of taxonomy (he has a 3.2, she has a 3.8 - she's clearly better/smarter). It's only natural that the same kind of competitive spirit would make its way into professional academia.
I'm 27 and finishing my undergraduate degree at a somewhat elite institution (top-20 worldwide according to THE), and even though after a brief stint of working "in the real world" as a software engineer and doing some start-upping, I was considering either law school or academia (once I finished my degree), right now my only thought is: fuck no! It's not that I don't have the grades or that I may not be smart enough, it's just that I can't stand the intellectual dishonesty and political machinations.
I'd rather eat Ramen for a whole year while coddling my newest start-up rather than partake in the mentally-masturbatory hallowed halls of most academic institutions. I'd rather build (and sell) new and interesting things that might actually have an effect on people instead of working harder trying to publish my papers than writing them.
I think academia has a lot to learn from internet-based free open source projects.
I find much easier to learn from RFCs, git-hosted documentation projects, and wikis. (Going from more professional/less open to less professional/more open - not that these are necessarily exclusive, but it seems costly to have a system that satisfies both.)
For documentation, I really love sphinx[1] and rst[2]. I know some people love markdown and it's good for light manuals/READMEs but IMO not adequate for full documentation.
I really disagree. Writing actual papers and having them peer reviewed is a useful model for actually doing science, and you can learn a ton from reading good academic papers.
I don't know why you bring up documentation, the current tools used to write academic papers seem fine.
As the OP states, the problem is the incentive system, not the particular set of tools and practices.
(However, I don't think his suggestions are fundamental enough to change things, and I am skeptical about his claimed expertise on the stock market and copyright.)
Writing down your thoughts and having them read by people is indeed pretty critical for science.
What does the current publishing model do for us that a blog wouldn't? What does academia-speak accomplish? From the few CS papers I've read, generally speaking the harder the paper is to understand, the less profound the insights contained. What does the formalism get us, besides formalism itself?
Peer review doesn't demand arcane terminology or difficult-to-understand writing. Ideally, peer review is just someone writing down some thoughts and other people who have some expertise in the field providing comments on those thoughts before they are published.
The problem with a blog is that there is no way to screen for quality or obvious mistakes. Of course the current method of doing this might be imperfect, or even completely broken depending on the field, but the solution isn't to just tell academics to start blogs and call it a day. It's a more complex problem than that.
I guess I'm not seeing a categorical difference. Someone posts a blog to HN that's full of garbage and bad numbers and it gets debunked. We don't have journal subscriptions or piles of bureaucratic BS, just some people who are knowledgeable in their area talking shop.
I don't see where all the ceremony creates enough value to pay for itself.
That's true, and it works well for subjective or popular topics. However, it works much less well for highly specialized, technical topics. I have noticed that quite a few very technical, or highly specialized articles end up with just a small handful of comments. This is a problem in research because just about everything is technical and specialized.
Another problem is recruiting enough people to actually review all the research produced. HN has a handful of links per day. There are tens of thousands of researchers around the world doing work. Peer reviewers aren't generally compensated monetarily, but they have a strong moral incentive to help out and review articles, especially if they themselves have published in the journal in question.
Again, I don't want to come across as defending the entire system. There are certainly things to be improved. All I'm saying is that I don't think the answer is to just tell everyone to start a blog. It's an appealing idea from a populist, egalitarian, and perhaps "hacker" perspective (just get the information out there!), but the current system provides a bit more than a simple communications medium.
I think blogs can complement the peer review system, but not replace it. Peer review gets you a lot. Anyway, the person I was responding to wasn't advocating blogs, but wikis, RFCs and documentation projects. Those things can also complement the peer review system, but not in a major way. Academics do already use those things where appropriate.
The things I talked about all involve heavy peer-review, from a much wider audience instead of a few academic referees. Academic papers are definitely not good learning materials, unless one is already familiar with the subject area. You would be hugely irresponsible if your taught a class simply by pointing your students to academic papers.
By contrast, for the sorts of projects I listed, there is more structure on the background for understanding a particular document, and even without explicit structure, hyperlinks make the entire mass of content easier to navigate and teach yourself with, instead of having to spend minutes digging out a reference. This is from personal experience of reading both academic papers and working with those sorts of projects, though granted I have more in the latter.
Good tools lower the cost of creating good documentation, meaning that more positive incentives can have higher priority. If current tools are "fine" then the OP would have used them to publish a paper instead of writing a series of blog posts.
> The things I talked about all involve heavy peer-review, from a much wider audience instead of a few academic referees.
The academic peer review process basically curates the literature by filtering out the majority of submitted papers, which often are low quality, cover something already done, etc. So, it's somewhat a different beast. More importantly, though,
> Academic papers are definitely not good learning materials, unless one is already familiar with the subject area.
That's right, because a lot of the cutting-edge research is really hard stuff that takes years of study to really understand. The point isn't to educate, it's to advance the knowledge of a small group (other researchers) that specilizes in a narrow area. Hopefully, those people (or their students) will eventually distill that knowledge into something usable by others (e.g. software). The point _can't_ be to educate (in general), because the stuff is just too hard. That's just the way research is. (If it's good research.)
I don't think the OP's experiences are typical: He needed to publish blogs because his work had immediate impact. That is rare overall in academia, including in CS. Most of the good work that is done will _eventually_ end up in a product, but it will take time.
But it's worth noting that the OP _also_ publishes papers. Really don't think there is a tool problem there, though there is always room for improvement.
> The academic peer review process basically curates the literature
I don't know why you keep talking about the academic peer review process. The things I mentioned have their own peer review processes too and it's for the same purpose as you said.
> The point isn't to educate, it's to advance the knowledge of a small group (other researchers) that specilizes in a narrow area.
This is a poor excuse, and from experience, quite insincere. Nearly everyone links to academic papers in non-academic settings as "references" to communicate with a wider audience, and nearly no-one re-writes an academic paper afterwards for it to be more understandable.
> the stuff is just too hard
Nothing is "too hard". I can understand RFCs perfectly fine, even an RFC that I have completely no background in. I cannot say the same thing for academic papers.
> Most of the good work that is done will _eventually_ end up in a product, but it will take time.
This is also just wrong.
You seem to be blindly asserting that academic papers somehow magically get turned into documentation that is more easily understood by a wider audience. This requires either academics to re-write their own papers (this nearly never happens), or for someone non-academic (like me) to come along, understand the paper, then re-write it. My entire topic in this thread has been that the latter is grossly ineffective, and that other avenues are much more effective, but unfortunately not used in the academic world.
I produce academic papers for a living, and when you call me "insincere" for claiming that the audience of these papers is other experts in an extremely narrow field, you're just proving that you're not listening to me.
> Nothing is "too hard".
Well, lots of things are too hard for me. I don't read papers in quantum physics, biochemistry, or pretty much any specialized field of science. And I don't even try to read papers outside of my subfield. Because I don't have the years of expertise necessary to understand them. That's just the way the world works. Things are hard.
> You seem to be blindly asserting that academic papers somehow magically get turned into documentation
No, I didn't say that they get turned into documentation. Go look at what I said.
It is quite easy to check the correctness of code excerpts, and implementing a piece of code can mostly be considered the analog of a theoretical calculation (we might make mistakes, but in principle you just have to turn the crank and do it right).
Otoh, with scientific experiments, it is critical to ensure validity and correctness so that further research may be built on a solid footing. Unlike code, it is not easy to "fork" and verify someone's work. So some form of peer review and verification are crucial.
It's not just about openness (scientists have arXiv, PLOS, etc) or about formats/systems (LaTeX is as good as anything else). It's fundamentally about the need to ensure correctness, to establish a foundation for future work.
I was not thinking primarily about code, but design documents, which are peer-reviewed for qualities beyond "correctness" - for example clarity and completeness, which are important for communicating your ideas to many people. And even code has a subjective quality beyond "correctness", so your first claim is false and has no bearing on real-world software projects.
OP was not talking about correctness and I am not talking about correctness, but the economics of collaboration and publication in academia.
(no connection, I just heard of them in this context). These kind of metrics would have addressed the OP's "blogger" problem.
Another strand, which at least my small scientific field is pursuing, is to facilitate citations of datasets, or dataset+software combinations (think of it as making a directory structure citeable). This (in theory - it's a recent development) rewards people who take their time to clean up and give open access to their datasets, instead of doing the minimum to extract a couple of papers out of it and move on.
reply