Look, I don’t mind wacky fonts or formatting. But this list of references is a pain in the neck, a usability nightmare, and worse than that.
Here’s how it looks from a reader’s perspective:
I start reading how curiosity “leads us to ask questions”...
I notice the superscript 1
I notice that the 1 isn’t a hyperlink, I infer that it leads to a footnote
I scroll down to the footnotes, losing my place in the main text
I locate footnote one, which reads “Peters (1978)”
I recognize the terse-citation style
I infer that there is a list of references
I scroll down to the references, losing my place again
I use Cmd-F to find “Peters”, since the list is very long
I finally get to the actual information I want: “Effects of anxiety, curiosity”, etc.
This still isn’t a hyperlink, so I must look for it manually
If I’m really dedicated, I’ll open a new tab, copy the title, and Google it
Even once I get the hang of what you’re doing here, only the few steps that are inferences will go away; most of the rest of this unnecessary work is still left for me to do.
You know what the worst part is?
You show how to do it right in the first goddamn paragraph, the one after the quotation: hyperlinks.
Don’t use a footnote to lead indirectly to a near-useless chunk of static text which the most dedicated reader can then use to maybe find the article. Link to the fucker directly. That’s how it’s done in this bright new millenium.
If at all possible, link to a free PDF. I know that in many cases the PDF is still held hostage behind a paywall. Hopefully that will change by and by. The best homage the LW community could pay to the virtue of curiosity would be to “liberate” some of these PDFs and make them available in some sort of public pool, the way you can get many paywalled PDFs for free if you know how to use CiteSeerX, or even just Google. We need some sort of AcademiaBay.
But anyway, the basic message is: Less Wrong is not a dead tree academic publication. There is nothing to be gained by pretending it is one.
In passing, I note that you’re making the Peters article say something that doesn’t quite square with what the abstract says. Not only are you needlessly dressing up a trivial observation (“curiosity leads us to ask questions”) in the guise of stuffy academic work—but worse, you aren’t reporting the research’s conclusions faithfully (as far as I can tell by looking at the abstract).
Have you actually read the full text of Peters 1978?
ETA: don’t take this as an attack on you personally; I think Luke(prog) is to blame for introducing this broken style of citation to LW in the first place—in any case I wrote to him with basically the above complaints. But this is kind of the straw that breaks this camel’s back, prompting me to comment on it publicly.
It would be nice for the posts to not lose information when printed out, and link rot may be a problem with URLs. Might use a LaTeX style bibilography where you have [Pet78] in the text which links to the Peters 1978 entry in the bibliography which has the standard biblography details and a hyperlink to the article, where the URL is both written out in text and a hyperlink.
In the footnote style, all footnotes are unique, so you could have the footnote bodies have links back to the text where the initial footnote was, but why bother since browsers have back buttons.
Reading these complaints, I can’t help but feel a little smug about my own articles on gwern.net—not only do I have real footnotes, I have floating footnotes to make it even easier, and I usually have fulltext as well. To top it all off, I have an elaborate archiving system just to deal with linkrot.
Smugness is justified. The floating footnotes are a great solution.
The standard way to not lose your place while reading a Web page is to Cmd- or Alt-click a link, opening it in a new tab. But the other problem solved by the floating notes is that hyperlinks are generally opaque as to what’s behind them; a reader appreciates the extra context provided by a “title” attribute, your tooltips are an extension of that. What’s impressive is that they work equally well on a touchscreen device.
Since as far as I can tell this is done in Js+CSS, we could easily steal your mechanism for LW. Would you object?
ETA: never mind, I’ve found the source. I’m now feeling bad about using the clichéd phrase “usability nightmare”, but this would definitely be a good addition to LW.
Using well-supported CSS techniques, it is possible to create a document where URLs appear when printed, though it does require placing them in the document twice.
On the reader’s side, if you don’t mind monospace and no images, the text web browser Lynx can format a page with all URLs as footnotes: lynx -dump <URL>
Does the CSS make the URLs appear in the middle of text paragraphs, where there was a hyperlink in the online version, or as footnotes or references outside the main text? Nobody really wants the main text to get interrupted by noisy URLs all the time, and on the other hand having URLs written out in the open even in the online version doesn’t matter that much if it’s in the references section at the end of the document.
The Lynx thing probably isn’t good for pleasant reading, but might work for some kind of compromise scheme where you get a document that’s very human-readable and can be used for an OCR data restoration that gets you the text and references back, though not most of the formatting.
CSS allows one to make arbitrary content appear or disappear when a document is printed (given that said content is already in the document). So you can have plain hyperlinks on screen, but numbered footnotes/references in print.
Sorry about that. I’ve now added all the PDFs I found. At the moment I’m unable to host the ones that are still missing, but it might be worth investing in.
Oops, looks like I accidentally cited Peters 1978 when I meant to cite a paper that article pointed me to. Fixed now.
I have read at least abstracts of all cited articles, which the authors of the paper you link to seem to think is fine:
we adopt a much more generous view of a “reader” of a cited paper, as someone who at the very least consulted a >trusted source (e.g., the original paper or heavily-used and authenticated databases) inputting together the citation list.
Most of my remarks about form still stand, and I’m stil very uncomfortable with your updated citation (Evans 1971).
Citation form functions here as a rhetorical device. I mean this as in “dark arts” rhetorical: its intent is to make a non-academic publication look more like an academic publication. The subtext is “look how well researched my claims are”, or perhaps more generously “this is settled science”.
What happens if we rewrite your claim, erasing the academic form, and reinstating the context?
What you write expands to the following: “Curiosity leads us to ask questions, as shown by Evans’ 1971 research on ‘The Ontario Test of Intrinsic Motivation, Question Asking, and Autistic Thinking’.” The implication is that the researcher has established a causal link between increased curiosity and increased question-asking behaviour.
If we look at the abstract of the paper itself, we get a slightly different story. The author is reporting on a study of the OTIM (Ontario Test of Intrinsic Motivation) designed to distinguish between various sub-traits of the trait “curiosity”, and the study gets null results on two of the scales examined (consultation and observation), turning up only one significant correlation, that between “directed thinking” and question-asking behaviour. Moreover this isn’t an experimental protocol, but a passive study, so causal inference is not warranted here: “correlation doesn’t imply causation”.
So really it would be much more accurate to write: “The ‘directed thinking’ aspect of curiosity seems correlated with asking lots of questions, one study observed (Evans 1971).”
This isn’t really settled science: it’s one study, reporting on one sub-trait of curiosity. If you want to make the broad generalization that “curiosity is that which leads us to ask questions”, then I expect you to cite a different kind of work, such as a broad survey listing many replications of the initial study.
In the cases where you cite a singular study, it’s a good idea to have read the full text, not just the abstract, to see whether there are any major issues with the study: otherwise you may be jumping to conclusions and updating too strongly in the researchers’ direction. This is especially true of old singular studies: if you’re not aware of more recent replications of the research, is that because attempts are replication were failure, or because the study asked a question which has since then been dissolved or reframed, or simply because you don’t know of more recent research? Those questions are worth asking.
The question isn’t what I or someone else “thinks is fine”. The question is whether I can trust you and your scholarship on factual matters investigated by academic research.
If, on examining the very first citation in your text, I come across sloppy or careless handling of factual matters, I am going to revise my evaluation of your trustworthiness—downward and significantly. You shouldn’t see this as a bad thing: after all, it’s what you would expect of an editorial board of an academic publication with decent standards.
In some instances, I use citations for pointing to relevant studies, without intending to imply that this is settled science. But I now realize that it does carry that implication, and that the wording of the sentence is particularly unfortunate. I have updated the first and other footnotes to take this into account.
By “thinks is fine”, I didn’t mean some arbitrary personal standard, but precisely the kind of epistemic abilities that you mention.
I understand your revision and thank you for pointing in out, so I can keep trying harder.
Look, I don’t mind wacky fonts or formatting. But this list of references is a pain in the neck, a usability nightmare, and worse than that.
Here’s how it looks from a reader’s perspective:
I start reading how curiosity “leads us to ask questions”...
I notice the superscript 1
I notice that the 1 isn’t a hyperlink, I infer that it leads to a footnote
I scroll down to the footnotes, losing my place in the main text
I locate footnote one, which reads “Peters (1978)”
I recognize the terse-citation style
I infer that there is a list of references
I scroll down to the references, losing my place again
I use Cmd-F to find “Peters”, since the list is very long
I finally get to the actual information I want: “Effects of anxiety, curiosity”, etc.
This still isn’t a hyperlink, so I must look for it manually
If I’m really dedicated, I’ll open a new tab, copy the title, and Google it
Even once I get the hang of what you’re doing here, only the few steps that are inferences will go away; most of the rest of this unnecessary work is still left for me to do.
You know what the worst part is?
You show how to do it right in the first goddamn paragraph, the one after the quotation: hyperlinks.
Don’t use a footnote to lead indirectly to a near-useless chunk of static text which the most dedicated reader can then use to maybe find the article. Link to the fucker directly. That’s how it’s done in this bright new millenium.
If at all possible, link to a free PDF. I know that in many cases the PDF is still held hostage behind a paywall. Hopefully that will change by and by. The best homage the LW community could pay to the virtue of curiosity would be to “liberate” some of these PDFs and make them available in some sort of public pool, the way you can get many paywalled PDFs for free if you know how to use CiteSeerX, or even just Google. We need some sort of AcademiaBay.
But anyway, the basic message is: Less Wrong is not a dead tree academic publication. There is nothing to be gained by pretending it is one.
In passing, I note that you’re making the Peters article say something that doesn’t quite square with what the abstract says. Not only are you needlessly dressing up a trivial observation (“curiosity leads us to ask questions”) in the guise of stuffy academic work—but worse, you aren’t reporting the research’s conclusions faithfully (as far as I can tell by looking at the abstract).
Have you actually read the full text of Peters 1978?
ETA: don’t take this as an attack on you personally; I think Luke(prog) is to blame for introducing this broken style of citation to LW in the first place—in any case I wrote to him with basically the above complaints. But this is kind of the straw that breaks this camel’s back, prompting me to comment on it publicly.
It would be nice for the posts to not lose information when printed out, and link rot may be a problem with URLs. Might use a LaTeX style bibilography where you have [Pet78] in the text which links to the Peters 1978 entry in the bibliography which has the standard biblography details and a hyperlink to the article, where the URL is both written out in text and a hyperlink.
In the footnote style, all footnotes are unique, so you could have the footnote bodies have links back to the text where the initial footnote was, but why bother since browsers have back buttons.
Reading these complaints, I can’t help but feel a little smug about my own articles on gwern.net—not only do I have real footnotes, I have floating footnotes to make it even easier, and I usually have fulltext as well. To top it all off, I have an elaborate archiving system just to deal with linkrot.
Smugness is justified. The floating footnotes are a great solution.
The standard way to not lose your place while reading a Web page is to Cmd- or Alt-click a link, opening it in a new tab. But the other problem solved by the floating notes is that hyperlinks are generally opaque as to what’s behind them; a reader appreciates the extra context provided by a “title” attribute, your tooltips are an extension of that. What’s impressive is that they work equally well on a touchscreen device.
Since as far as I can tell this is done in Js+CSS, we could easily steal your mechanism for LW. Would you object?
ETA: never mind, I’ve found the source. I’m now feeling bad about using the clichéd phrase “usability nightmare”, but this would definitely be a good addition to LW.
Well, I didn’t write it in the first place, and I’d be happy to see it on LW.
Using well-supported CSS techniques, it is possible to create a document where URLs appear when printed, though it does require placing them in the document twice.
On the reader’s side, if you don’t mind monospace and no images, the text web browser Lynx can format a page with all URLs as footnotes:
lynx -dump <URL>Does the CSS make the URLs appear in the middle of text paragraphs, where there was a hyperlink in the online version, or as footnotes or references outside the main text? Nobody really wants the main text to get interrupted by noisy URLs all the time, and on the other hand having URLs written out in the open even in the online version doesn’t matter that much if it’s in the references section at the end of the document.
The Lynx thing probably isn’t good for pleasant reading, but might work for some kind of compromise scheme where you get a document that’s very human-readable and can be used for an OCR data restoration that gets you the text and references back, though not most of the formatting.
CSS allows one to make arbitrary content appear or disappear when a document is printed (given that said content is already in the document). So you can have plain hyperlinks on screen, but numbered footnotes/references in print.
Sorry about that. I’ve now added all the PDFs I found. At the moment I’m unable to host the ones that are still missing, but it might be worth investing in.
Oops, looks like I accidentally cited Peters 1978 when I meant to cite a paper that article pointed me to. Fixed now.
I have read at least abstracts of all cited articles, which the authors of the paper you link to seem to think is fine:
Most of my remarks about form still stand, and I’m stil very uncomfortable with your updated citation (Evans 1971).
Citation form functions here as a rhetorical device. I mean this as in “dark arts” rhetorical: its intent is to make a non-academic publication look more like an academic publication. The subtext is “look how well researched my claims are”, or perhaps more generously “this is settled science”.
What happens if we rewrite your claim, erasing the academic form, and reinstating the context?
What you write expands to the following: “Curiosity leads us to ask questions, as shown by Evans’ 1971 research on ‘The Ontario Test of Intrinsic Motivation, Question Asking, and Autistic Thinking’.” The implication is that the researcher has established a causal link between increased curiosity and increased question-asking behaviour.
If we look at the abstract of the paper itself, we get a slightly different story. The author is reporting on a study of the OTIM (Ontario Test of Intrinsic Motivation) designed to distinguish between various sub-traits of the trait “curiosity”, and the study gets null results on two of the scales examined (consultation and observation), turning up only one significant correlation, that between “directed thinking” and question-asking behaviour. Moreover this isn’t an experimental protocol, but a passive study, so causal inference is not warranted here: “correlation doesn’t imply causation”.
So really it would be much more accurate to write: “The ‘directed thinking’ aspect of curiosity seems correlated with asking lots of questions, one study observed (Evans 1971).”
This isn’t really settled science: it’s one study, reporting on one sub-trait of curiosity. If you want to make the broad generalization that “curiosity is that which leads us to ask questions”, then I expect you to cite a different kind of work, such as a broad survey listing many replications of the initial study.
In the cases where you cite a singular study, it’s a good idea to have read the full text, not just the abstract, to see whether there are any major issues with the study: otherwise you may be jumping to conclusions and updating too strongly in the researchers’ direction. This is especially true of old singular studies: if you’re not aware of more recent replications of the research, is that because attempts are replication were failure, or because the study asked a question which has since then been dissolved or reframed, or simply because you don’t know of more recent research? Those questions are worth asking.
The question isn’t what I or someone else “thinks is fine”. The question is whether I can trust you and your scholarship on factual matters investigated by academic research.
If, on examining the very first citation in your text, I come across sloppy or careless handling of factual matters, I am going to revise my evaluation of your trustworthiness—downward and significantly. You shouldn’t see this as a bad thing: after all, it’s what you would expect of an editorial board of an academic publication with decent standards.
In some instances, I use citations for pointing to relevant studies, without intending to imply that this is settled science. But I now realize that it does carry that implication, and that the wording of the sentence is particularly unfortunate. I have updated the first and other footnotes to take this into account.
By “thinks is fine”, I didn’t mean some arbitrary personal standard, but precisely the kind of epistemic abilities that you mention.
I understand your revision and thank you for pointing in out, so I can keep trying harder.