Fermat’s Library – Annotating Academic Papers Every Week
blog.ycombinator.com
blog.ycombinator.com
Rendering is slow, buggy, and results in choppy scrolling. The annotations feels out of place UI-wise, but often also content-wise.
Unfortunately, http://fermatslibrary.com/ is down at the moment so I can't comment with an informed opinion.
But there are two killer features that most annotation software fail to supply:
1. print-ability. Physical copies are important.
2. Accurate rendering (related to 1). There's a pretty high barrier to overcome before someone else's inline notes are more valuable than knowing that I'm looking at exactly what the authors intended for me to look at.
That said, the scientific PDF annotation space is ripe for disruption. IDK how meaningful that market is (directly or indirectly), but someone could definitely step in and own it with something that's even an iota better than current options.
I don't know what kind of annotation software you usually use, but I have yet to annotate a PDF and lose the ability of 1 and/or 2.
As for crowd-annotation services, I can only think of genius and medium. Genius does not preserve original format, but that is obviously unreasonable to expect (they're songs!). Medium is its own format, so rendering is accurate by definition.
I've mostly dabbled with preview, Adobe reader, Zotero and Mendeley. You can in all do whatever advance shit you'd like in 3rd party software and copy-paste.
Reading academic papers requires focus and concentration. While I like the idea of comments on papers, I found that site too full of petty distractions to serve that purpose.
"I've been working on a project to implement a feature similar to PDF native 'custom annotations,' but using an external JSON file. The idea is to allow a more hacker-friendly way to add javascript-powered interactivity to a PDF file, and allow a js-based-document-viewer to interact with other elements in a web page. Somewhat like using an external subtitles-file for a video instead of hard-coded subs, allowing you to use things like HTML5 tracks for generating events etc. (manipulating PDFs internal workings is hard work for those who create documents/content, and the JS support is old-fashioned, if I remember rightly).."
https://groups.google.com/forum/#!topic/mozilla.dev.pdf-js/2...
I think mostly I liked PDF's precise, fixed control over layout, and wanted to be able to program some js onClick callbacks on top of it in an easy way. Maybe sometime I will revisit the ideas, see if there was something there.
1) opinions masquerading as news. 2) misinterpretation of facts/data. 3) misrepresenting the facts. 4) cherry picking (not showing full picture, such that the meaning changes). 5) insufficient context / history about the topic.
All these issues with reporting could be mitigated with expert + crowd annotation — esp where the author and the subject (say people mentioned in the news) can come and add more context.
When objective function is speed and clicks, sometimes depth on a topic suffers. This is true whether news is human or machine generated. In that case, continual engagement with the original article, amendments and updates become a very important tool.
If journalism is defined as a search for truth, then news annotation could be the thing that saves it.
HN, thoughts?
Journalism isn't suffering from a scourge of fake news.
You don't need to have an increase in fake news to have an increase in the perception of fake news. Another way to get an increase in the perception of fake news is to change the structural graph through which news propagates. The two most outspoken people I know of on the topic of the problem of fake news both happen to be twitter users (Elon Musk, Donald Trump).
So basically, I disagree with the claim that fake news doesn't exist, in a subtle way.
I got that bit, it's just not a thing I argued.
However, it's recently become politically controversial to acknowledge this fact, I think because Trump uses "fake news" to disqualify criticism of himself and his administration[1]. But let's not pretend that journalism is without faults or bias just because (on average) the virtues of the fourth estate outweigh their flaws.
-----
0. https://en.wikipedia.org/wiki/Manufacturing_Consent
1. He's hardly the first person to have issues with his press coverage, but it seems like he's getting a bit more traction with his complaints. It helps that he has an observably hostile relationship with the press. But the term "fake news" was originally applied to right-wing blogspam/memes that showed up during the 2016 election, which were highly slanted, based on cherry-picked information if not outright lies.
Leaving that aside and to your actual point - one way to think about this is 'what major journalistic mishap could have been fixed/avoided by crowd annotation'. Judith Miller's reporting in the run up to the Iraq war? Wen Ho Lee's case? Jayson Blair? It doesn't seem obvious to me it would have made a whit of a difference in any of those nor many others.
It might be better to take an approach similar to "open science". Have the reporters publish their raw footage and interview transcripts to address misrepresentation/cherry-picking, and get them to commit to answering follow-up questions from readers to allow the readers to get additional information. Having more of a back-and-forth could improve the quality of the reporting, allow for spurious objections to be debunked, or ensure that the author gets taken to task if their bias taints the story.
This seems time/capital intensive and therefore impractical in the age of listicles, but on further consideration I realize that many "name brand" bloggers popular on HN essentially follow this model-- for example, Matt Levine and Brian Krebs. However it's not clear whether their model can be applied to all areas of journalistic interest instead of specific niches, or if it requires a rare kind of individual to make it work.
Annotation sites like to say, "Hey, we have this paper and we have this paper." Meanwhile, thousands of new scientific papers are published every day. Of course, the majority of those are not going to be of interest to any journalist. But even when we whittle the pile down to those that are, there is still a huge amount of work to do, and some qualified person has to do it.
And why would they do it? I'm an academic researcher. So I have the required expertise to do this kind of annotation for papers in my narrow specialty. But my life is full with my own research and publishing and teaching and advising and committees and paperwork .... And every October I have to write up a report on my achievements for the past year. Getting a paper published counts for a lot in that report. Annotating someone else's paper is not going to count for much at all. If I start substituting the latter for the former, then my career is going to go down the tubes pretty rapidly.
I think it would be great if this kind of thing could actually happen. But I don't see how to make it happen.