tea.mathoverflow.net - Discussion Feed (Pointing to papers on the arXiv)

Scott Morrison comments on "Pointing to papers on the arXiv" (1851)

2010-01-15T13:44:30-08:00

There's one slightly annoying problem with the Front, which causes me to hesitate to link to it.

When the new arxiv identifiers came into effect, Greg made up "new-style" identifiers for the all the old papers. These appear in some places, but not others. If you take one of these new-style identifiers for an old paper, and try to construct the corresponding URL over at the arxiv, you find yourself getting surprising 404s.

Ilya Nikokoshev comments on "Pointing to papers on the arXiv" (1836)

2010-01-15T05:08:36-08:00

Links to documents should always be to the canonical page for that document

This functionality could be replicated off site: you could write a "smart bookmark" which allows you to highlight something looking like an arXiv reference and search the arXiv for it (via the front if preferred). As this would be in the browser, it doesn't require sending even more stuff back and forth and could be used on other sites.

This is basically what I suggest; it should be a piece of javascript that would work on any page with arXiv links. However, there's nothing wrong with Mathoverflow writing, hosting and advertising such a bookmark. It's not hard, but still requires some community input and a bit of work; if there's a big interest, I'll give it a try.

Andrew Stacey comments on "Pointing to papers on the arXiv" (1827)

2010-01-15T01:46:00-08:00

@Harry: I don't see any particular difference between searching the main arxiv and searching the front, apart from the fact that you can do a full text search on the arxiv but not on the front. I suppose that some might think that cumbersome and annoying ... And via the arxiv API you can design your own customised RSS feed direct from the arxiv, can't do that from the front ...

Links to documents should always be to the canonical page for that document. In the case of the arXiv, that's the abstract page on the arxiv itself. Anyone who has a preferred other way of accessing the arxiv can do so but the person posting the link should not presume to impose one. After all, specifying the front is a bit US-centric of you! Maybe I should ensure that all my links go to the German server just for balance.

Mariano comments on "Pointing to papers on the arXiv" (1818)

2010-01-14T10:12:07-08:00

@David: I do not think there is any serious problem with linking to the Front, yet the argument Andrew makes against it is really strong. What is not good is to link to (what should be internal, hidden) URLs that arXiv presents, like those pointing to cached copies of PS and PDF files---they may well change in the future, and/or point to no longer cached files, and so on. arXiv should be smarter and never present those URIs to the user, really.

Harry Gindi comments on "Pointing to papers on the arXiv" (1816)

2010-01-14T09:15:24-08:00

Searching is just worse to use on the regular arXiv. It's cumbersome and annoying compared to the front.

Michael Lugo comments on "Pointing to papers on the arXiv" (1814)

2010-01-14T09:03:06-08:00

I've also wondered why people prefer the front. As far as I can tell it doesn't offer any features that the real arXiv doesn't. Can someone enlighten me? (Does Greg read meta.MO?)

Andrew Stacey comments on "Pointing to papers on the arXiv" (1809)

2010-01-14T03:09:20-08:00

@Ilya: I vote against. There's enough "junk" already in MO javascript without adding to it. This functionality could be replicated off site: you could write a "smart bookmark" which allows you to highlight something looking like an arXiv reference and search the arXiv for it (via the front if preferred). As this would be in the browser, it doesn't require sending even more stuff back and forth and could be used on other sites.

As for using the front versus the arXiv; yes, there's a very good reason for using the arXiv. It's the canonical URL for a given preprint. The front is ... er ... a front. If Greg decides that maintaining the front is too complicated and gives up then it stops working but the arXiv itself is still usable. On the other hand, if the arXiv stops then the front doesn't work either.

More concretely, when the arXiv changed their identifier scheme, it took Greg a little while to catch up and the front wasn't usable.

The ideal solution is to add a sort of wikilink syntax to the markdown. Thus something like [[arxiv:0908.0512]] gets converted to a link to "http://arxiv.org/abs/0908.0512" (or whatever it should be). You could also put [[arxiv:0908.0512|Title of paper]]. Then you could extend it to allow automatic links to the nlab, wikipedia, google, and whomsoever you like.

Kevin Lin comments on "Pointing to papers on the arXiv" (1808)

2010-01-14T00:39:53-08:00

@Ilya: That sounds like a good idea. Btw, I've never understood what the difference was between arxiv and the ucdavis arxiv thingy. I mean, why would you prefer one over the other? They seem to be basically the same in every way.

Ilya Nikokoshev comments on "Pointing to papers on the arXiv" (1804)

2010-01-13T21:25:12-08:00

I want to point out that it's possible to write a javascript script that would go between arXiv, pdf and ucdavis representation according to viewer's preferences (you just loop over all links and change their href accordingly).

Anyone thinks it's an interesting idea?

David Speyer comments on "Pointing to papers on the arXiv" (1803)

2010-01-13T20:59:40-08:00

I often point to the corresponding front.math.ucdavis URL, such as http://front.math.ucdavis.edu/0908.0512 . I prefer the front's search interface and it seems silly to go find the corresponding arXiv.org URL when I already have the front URL in my browser. Is there a downside to this that I am not aware of?

Mariano comments on "Pointing to papers on the arXiv" (1802)

2010-01-13T20:20:26-08:00

BTW: one thing I would emphatically suggesr is that when people provide links to thearXiv, they provide links to the actual URL, as in http://arxiv.org/abs/0912.2227, instead of the various other variants that arXiv should probably hide better (including direct links to cached PDF files, most notably)

Ilya Nikokoshev comments on "Pointing to papers on the arXiv" (1783)

2010-01-13T11:15:24-08:00

Here are my ideas.

In the 2009 paper Algebraic homotopy classes of rational functions Cazanave computes ...

 *[Algebraic homotopy classes of rational functions](http://arxiv.org/abs/0912.2227)*

Take a look at recent paper by Chris. Let's cite it again and again and again.

This is done by a reference-style link:

[recent paper by Chris][1]. Let's cite it [again][1] and [again][1] and [again][1].
  [1]: http://arxiv.org/abs/0912.2227

Noah Snyder comments on "Pointing to papers on the arXiv" (1780)

2010-01-13T10:37:08-08:00

I agree, links ftw!

David Speyer comments on "Pointing to papers on the arXiv" (1777)

2010-01-13T09:50:46-08:00

I agree with Ben. The reason to tell someone an arXiv number is so that they can type it into their browser. On the internet, it's better to use a link.

Ben Webster comments on "Pointing to papers on the arXiv" (1775)

2010-01-13T08:12:49-08:00

I think using the arXiv identifiers is silly. If you want to refer to a paper in the title of a question, put the title of the paper there, and then a link in the body of the question.

Michael Lugo comments on "Pointing to papers on the arXiv" (1774)

2010-01-13T07:16:09-08:00

Look at the question A rational point in the scheme of pointed degree n rational functions [0912.2227] . To me, all arXiv identifiers look the same; I can certainly tell that this refers to a paper from December 2009, but if I had read the paper in question I'm not sure I'd remember that I had by seeing the number alone.

However, I'm usually aware of who wrote the papers I read; I suggested to the poster (philip314) that it might make sense to put [Cazanave, 0912.2227] in the title for this reason. Philip pointed out that such a convention could get unwieldy for multiple-author papers. Does anybody have thoughts on this? Of course a convention will evolve eventually if people regularly ask questions that are tied to specific papers on the arXiv, but we should help this process along.