fascinating stuff. I was just today wondering why papers are so densely sprinkled with references. it almost feels like science writing is a game of playing lego, where the building blocks are references to other papers. I mean look at this sentence: "Given that implicit attitudes uniquely predict many everyday behavioral responses (Cameron et al. 2012; Galdi et al. 2008; Greenwald et al. 2009; McNulty et al. 2013; Perugini et al. 2010; Towles-Schwen & Fazio 2006; cf. Oswald et al. 2013), and potentially play a role in dysfunctional interpersonal relationships (see McNulty et al. 2013; Towles-Schwen & Fazio 2006), this traditional view of implicit attitudes suggests that the clinician’s role in updating maladaptive implicit affective memories seems necessary but challenging." do we really need ALL those references??
I'm still left wondering though, what incentives are keeping this system in place? if everyone agrees the entire paper writing / review process is painfully tedious, why haven't alternatives become more mainstream?
I think it's really hard to change the system because powerful and established players have a strong incentive for nothing to change, and they've been pretty competent at keeping their status. There's widespread fear that breaking the system would make us lose the few things that work well, like the norm of referring to previous work — which is probably overall good, even though I fully agree that distracting in-text references in boring "background" sections at the beginning of papers are absolutely a scourge.
Well it is and it isn't. I think humor may be one of the last human creativities that AI will have a difficult time mastering..since even we humans have difficulty knowing or recognizing it when it is deployed. And no one agrees on what is funny. Reference the remarks of the Justices in the Supreme Court in the Jack Daniels litigation.
So far as outsourcing the misery to AI, and PASTA..I have reservations.. The Pre-socratics emancipated humanity from its own view that the world was controlled by the will of the gods and only comprehensible by those same gods, since we could not know their minds. If we outsource the production of science to AI might we not find ourselves eventually in the same self-limiting mind-set and have to perform a second emancipation from the new gods we created? Assumimg they'll let us.
Oh, agreed, I'm really not sure it would be good to let a higher intelligence take care of scientific advancement in ways that may be totally incomprehensible to us. I would rather science remain a human activity that celebrates human curiosity and creativity. Since it's not happening anytime soon with respect to mainstream science publishing, though, that's a good case to applying AI.
Étienne, that's precisely the path that will surely come- a gradual, pleasant one of incremental reassignment of tasks, a blurring of boundaries, a brief period of epistemic confusion, followed by an acceptance of a new normality. We've been through this exact dance before numerous times with many of our transformative technologies...
But this time things will be different in this respect..none of our previous dances involved a partner who became sentient.
One of the major thing that people missed about papers, are that research is a live tradition. Adding more emojis is just going to confuse people 5/10/20 years down the line.
MEMES, however, would have a more standardized vocabulary, and is good for conference presentations in the worst case and be equal to "eponymous laws" in the best case.
Hmm, I'm not sure that's strictly true? I mean in the sense that writing for future people doesn't seem close to be the primary concern of (most) paper authors. Some papers are meant to be authoritative future references (e.g. literature review), but most aren't expected to be read much in the future.
In any case, I agree that making stylistic choices that are deeply embedded in current fashion might not be a good idea of you're writing for posterity, but either way those stylistic choices should be left to the authors, not the vague yet powerful conservative norms that float above the whole field.
fascinating stuff. I was just today wondering why papers are so densely sprinkled with references. it almost feels like science writing is a game of playing lego, where the building blocks are references to other papers. I mean look at this sentence: "Given that implicit attitudes uniquely predict many everyday behavioral responses (Cameron et al. 2012; Galdi et al. 2008; Greenwald et al. 2009; McNulty et al. 2013; Perugini et al. 2010; Towles-Schwen & Fazio 2006; cf. Oswald et al. 2013), and potentially play a role in dysfunctional interpersonal relationships (see McNulty et al. 2013; Towles-Schwen & Fazio 2006), this traditional view of implicit attitudes suggests that the clinician’s role in updating maladaptive implicit affective memories seems necessary but challenging." do we really need ALL those references??
I'm still left wondering though, what incentives are keeping this system in place? if everyone agrees the entire paper writing / review process is painfully tedious, why haven't alternatives become more mainstream?
Thanks! I explored these questions at some length when I wrote this essay on prestige in Nature and other journals: https://etiennefd.substack.com/p/why-is-nature-prestigious
I think it's really hard to change the system because powerful and established players have a strong incentive for nothing to change, and they've been pretty competent at keeping their status. There's widespread fear that breaking the system would make us lose the few things that work well, like the norm of referring to previous work — which is probably overall good, even though I fully agree that distracting in-text references in boring "background" sections at the beginning of papers are absolutely a scourge.
But if papers used emoji, how would we know whether the science is SeRiOuS?????
Well it is and it isn't. I think humor may be one of the last human creativities that AI will have a difficult time mastering..since even we humans have difficulty knowing or recognizing it when it is deployed. And no one agrees on what is funny. Reference the remarks of the Justices in the Supreme Court in the Jack Daniels litigation.
So far as outsourcing the misery to AI, and PASTA..I have reservations.. The Pre-socratics emancipated humanity from its own view that the world was controlled by the will of the gods and only comprehensible by those same gods, since we could not know their minds. If we outsource the production of science to AI might we not find ourselves eventually in the same self-limiting mind-set and have to perform a second emancipation from the new gods we created? Assumimg they'll let us.
Plus ça change..
Oh, agreed, I'm really not sure it would be good to let a higher intelligence take care of scientific advancement in ways that may be totally incomprehensible to us. I would rather science remain a human activity that celebrates human curiosity and creativity. Since it's not happening anytime soon with respect to mainstream science publishing, though, that's a good case to applying AI.
Étienne, that's precisely the path that will surely come- a gradual, pleasant one of incremental reassignment of tasks, a blurring of boundaries, a brief period of epistemic confusion, followed by an acceptance of a new normality. We've been through this exact dance before numerous times with many of our transformative technologies...
But this time things will be different in this respect..none of our previous dances involved a partner who became sentient.
One of the major thing that people missed about papers, are that research is a live tradition. Adding more emojis is just going to confuse people 5/10/20 years down the line.
MEMES, however, would have a more standardized vocabulary, and is good for conference presentations in the worst case and be equal to "eponymous laws" in the best case.
Side note: Most programmers understand that any code they wrote would look "new" after 6 months. For language, emoji would vibe like Xu Bing's works on emoji illegibility. https://archive.ph/pjdfn http://www.xubing.com/en/work/details/688?year=2022&type=year#688
Hmm, I'm not sure that's strictly true? I mean in the sense that writing for future people doesn't seem close to be the primary concern of (most) paper authors. Some papers are meant to be authoritative future references (e.g. literature review), but most aren't expected to be read much in the future.
In any case, I agree that making stylistic choices that are deeply embedded in current fashion might not be a good idea of you're writing for posterity, but either way those stylistic choices should be left to the authors, not the vague yet powerful conservative norms that float above the whole field.
It is kind of weird to want one version for bookkeeping and another for communication... But it is less conservative whilst maintaining practicality.