Personally I'm a big fan of blog posts: they're written for a broader audience in understandable language. I don't have any trouble reading English, but papers are almost universally written in a way that is more complicated than necessary. And in blog posts, if someone made a "novel new algorithm", at least you can be sure there's code, raw test data, screenshots, or whatever else is needed to DIY. And someone on the web is talking about it, perhaps even with comments below the post itself.
I edited the link to the more "lightweight" version (i.e. just the page, without loading his entire homepage as well, which is kind of heavy), so please consider trying again.
I agree with you RE blog posts, and the point of me linking to Bret Victor's work is to give an example of a form that's better-suited to the media we operate with today. Papers still live in the paper age (pun not intended).
Since Bret has liberated the scientific article from paper, we may as well take advantage of the more flexible layout possible with digital media. Interactive diagrams are so nice for learning that I think it's worth pursuing Victor's idea further. Fred Akalin's blog has a couple great examples! https://www.akalin.com/quintic-unsolvability
The purpose of a paper is to be rigorous and exact and move the field of study forward. That tends to make them dense and use the fields jargon - which is short hand for massive amounts of information.
The problem isn't the jargon itself, it's the fact that most labs have built up decades of institutional knowledge that they hoard for fear of their research being "scooped." I'm just thankful I don't work in downstream medical research anymore, where the number of replicable papers goes from one in two to one in ten, at best.
Your comment about reagents from one manufacturer failing while another succeeded brings back fond memories of trying to make our own botulinum neurotoxin to study neurons- and the rest about hoarding institutional knowledge reminds me why I tried to get out of that. I was one of those ambitious undergrads, but I permitted myself to wilt under the insanity of a system in which I KNEW that a partnered laboratory knew exactly what to do, but they were forbidden to share the knowledge because of politics. By the time I had cut through the wasteful uselessness, I discovered that the only sane person there, one who had been willing to share knowledge, she had actually DIED.
And so I merrily skipped over to the bioinformatics department, which has issues as well but in which I was able to manage them fairly effectively. :D
Unfortunately it's the same for software and algorithms (at least in bioinformatics, which is my field). Everything is treated like closely guarded secrets, at least in the smaller environments.
Thankfully the tendency is slowly being reversed and there are even things developed out in the open: the "problem" with those is that being in the open, it's harder to get publications out.
But imagine if the incentives of scholars somehow shift so that they look less to funding committees and more to the audiences of blogs and magazine articles. I think such a shift would do more good than harm.
Scholars would still write serious papers to establish their reputation amongst each-other, but insofar as they rush to publish, I'd rather they pump out accessible blog style articles than the kind of gibberish that is made today in order to inflate the citation metrics.
you mean something like https://terrytao.wordpress.com/
or
https://blogs.princeton.edu/imabandit/
Then yes I think these are a very valuable and increasingly important part of the scientific process. I think you might benefit from labelling the blog as 'math blog' which is quite distinct from the original link.
This is a function of two things. The first is that venues often have strict page limits for their submissions. Authors typically have a lot to say and not a lot of space to say it, so they pick language that is precise, dense, and typically colorless. The second is that every domain has a lot of style and nomenclature that the layman won't know. It can be learned, but you must remember scientists write for each other, not for the layman.
The second: I was talking about reading papers from my own field. That I don't understand a Biology paper (which indeed I don't -- I've tried) is perfectly understandable, but infosec papers shouldn't (and don't) contain any unknown lingo to me. It's just a very convoluted way of writing.
The whole thing is 283 kB and half of that is due to three small images that are already compressed as much as is possible (but that are actually part of the comment about the paper formatting, not the paper itself).
That seems very reasonable for a scientific paper which would have been about the same size as a PDF.
http://worrydream.com/#!/ScientificCommunicationAsSequential...
which loads the article... plus the entire homepage of Bret Victor, which is kind of heavy.
The primary shortcoming of current scientific papers is that they're too short. They're written to physically fit into a paper journal, but their brevity massively impairs efforts to validate and replicate research. We're taught in high school science class that a paper should include all the information needed to replicate a study, but in practice that simply isn't true. Our methods of publication haven't kept pace with the complexity of modern research and analysis methods. Failure to replicate and research misconduct is eroding the foundations of science.
In 2017, the appendices of a paper should include everything - the full dataset, complete lab notes, full source code, the works.
(Disclaimer: Posting in a personal capacity.)
If you could get the Windows source code with some paid subscription program, so that it's not pirated, it still wouldn't be open source. It would be more like "open access".
Open source means we can modify it and share our modified version, and even charge for it.
Most users of scientific papers aren't looking to extend those papers and redistribute modified copies; they just read them privately and make citations (which nobody can tell whether they were from a licensed or pirated copy).
And this is exactly why Richard Stallman doesn't like the term. https://www.gnu.org/philosophy/open-source-misses-the-point....
> The official definition of “open source software” (which is published by the Open Source Initiative and is too long to include here) was derived indirectly from our criteria for free software. [...]
> However, the obvious meaning for the expression “open source software”—and the one most people seem to think it means—is “You can look at the source code.” That criterion is much weaker than the free software definition, much weaker also than the official definition of open source. It includes many programs that are neither free nor open source.
> Since the obvious meaning for “open source” is not the meaning that its advocates intend, the result is that most people misunderstand the term.
That is, Stallman is criticizing "open source" for being, among other things, easy to misunderstand as not requiring free software licensing — even though it does, in fact, require free software licensing.
It's saying that one of the problems with "open source" is that you can make the same mistake you just made (thinking that being able to see the source code makes it "open source").
Richard Stallman's primary issue with "open source" is that they intentionally avoid talking about issues of software freedom, and have consistently muddied the waters by co-opting free software as "open source".
Looks like they've gone in the opposite direction. Access to the torrent repository is no longer available. The only way to download articles now is by clicking through ads.
If the torrent repository were gone permanently, it would be extremely sad, since it's (AFAIK) the only way of having automated and/or bulk download of papers, for example for data mining — sci-hub.io is understandably not really suited for this.
No I tried to pose the same question recently on the forum but it's locked down for new users. At best it has something to do with recent litigation. At worst they're being hypocritical and seeking that sweet ad dollar. I agree that the current situation is only marginally better than the status quo.
See the Distill initiative, for one direction:
HN discussion:
HTML (in some reasonable form) is my favourite format for reading text.
It's plain text so you can throw parsers at it to extract text, data, and formulae. It can be automatically converted into HTML for online display, PDF for downloading, etc. It's not pretty, of course, but we already have plenty of experience parsing and extracting useful information from another widely used format that mixes semantics and presentation all over the place, so why not?
PDF's are a danged-sight easier to organize for research as well, IMO.
[1] because it's Bret Victor, of course
(1) MOOCs - student does not need to pay exorbitant tuition fees
(2) Sci-Hub - student does not need access to a university library
(3) ? universal accreditation - this piece is missing
===
idea for start-up: aggregate MOOCs, package instructions on how to retrieve information online, connect instructors and tutors to students, arrange accreditation, offer BA, MA, PhD programs, we have start-ups in every industry but higher level education, could it be possible? …
This will turn out to be harder than it appears. The accreditation process is complex, expensive, and bureaucratic - possibly by design.
PhDs accreditation would be particularly complicated. PhDs are partly a form of academic hazing and lineage development and very much not just about the research/content.
Unfortunately education is more of a political problem than an information redistribution problem.
I want to go off a tangent here, though. Now that open access (whether arXiv or SciHub style) is becoming the norm, I wonder what can be done to improve the format of scientific papers? Like e.g. making them more like this:
http://worrydream.com/ScientificCommunicationAsSequentialArt...
instead of regular PDFs?