When you have a hammer…
…all the world is a nail. And the currently popular hammers are things like Twitter and Instagram and Tinder. While some have long advocated the first two as important tools for scientists, the last has been used as a model for scanning through preprints. Lots and lots of preprints. The Science story on this says “A web application inspired by the dating app Tinder lets you make snap judgments about preprints—papers published online before peer review—simply by swiping left, right, up, or down.”
Nothing says “science” like “snap judgment”.
While GG lambasted an effort to capture social media-ish solutions as a means of post-publication peer review, how about tools to let you find what cutting edge science is appearing? That Science report on social media linked above says that is what social media is good for. Um, really?
GG studies the Sierra Nevada. Try going to Twitter and searching on #SierraNevada. Bet you didn’t think there were that many people so fascinated with taking pictures of beer bottles. Add, say, #science. Chaff winnowed some, but very little wheat. Add #tectonics. Crickets.
The idea of this new app (Papr) is that if only you were able to see lots and lots of stuff quickly, you’d find some gems to explore. Really? Students complain bitterly about a firehose approach in the classroom, and the solution here is, um, a firehose? (To be fair, it appears the app developers are not necessarily expecting great things here).
Forget that. What we want and need are tools to reduce chaff, not accelerate it.
What we need is something akin to Amazon’s suggestions tool. Imagine visiting the preprint store to get a couple of papers you know you want. One maybe is on a topic you care about–say, the Sierra Nevada. Another maybe deals with a technique, say full waveform tomography. A third uses some unusual statistical tests. You download these and the preprint store suggests a few other preprints based on the full text content of the papers you got. Why that instead of keywords? Keywords have a way of being too picky. You might call work “tectonics” and GG might call it “geodynamics” and thus the keywords searches might pass by each other. But if the text is still talking about changes in elevation, changes in lithospheric structure–those are less likely to get overlooked. If this tool is smart enough to recognize quasi-synonyms and phrases, all the better.
Such a tool grows more powerful the more you work with it. While on that first try, you will also get recommendations on papers overlapping in non-interesting ways (say, applications of the techniques in paper 1, the geographic area under study in paper 2, and the measurement types in paper 3), the more you interact with this, the better it gets.
Here’s the sad thing: the tools to make something like this have been around for decades. The best spam filters (like SpamSieve) use a form of Bayesian filtering based on message content in addition to black- and whitelists. Earth science got much of its literature into a single “preprint store” long ago in GeoScienceWorld. And yet here we are, swiping left again and again and again….