I have had a couple of experiences recently which have made me wonder whether the mathematics community should try to establish and maintain something like the Wayback Machine, but specifically focused on mathematics.

In a paper that I wrote back in 1999, I cited a webpage with the title "Is elementary?" and provided the following URL:


As you might guess, that URL is no longer functioning. I tried the Wayback Machine and I was lucky that it had indeed archived a copy of the relevant part of the page. Unfortunately, crucial parts of the page were not preserved; the original page used LaTeX2HTML, whose results were not satisfactorily archived.

The above example is not so important from a mathematical point of view because there is now a much better reference available for the mathematical fact in question, but it is a good illustration of the sort of thing I am concerned about—there's a considerable amount of material on the web that is of mathematical value but that is disappearing because it is not being formally published or archived.

My second example is more (ahem) consequential, and is something that the mathematical community might be able to do something about if we act now. For years there has been a useful web resource at Purdue for researching Consequences of the Axiom of Choice. Unfortunately, the page is no longer functioning, as you will quickly discover if you try submitting a form number. The URLs have changed. I suspect that Purdue redesigned its website at some point, changing the URLs, and that since Herman Rubin died a couple of years ago, there is now nobody responsible for maintaining the Axiom of Choice page. I tried emailing a couple of random people in the Purdue mathematics department to find out if something could be done to revive the page, but have received no response.

The ideal solution for the Axiom of Choice page may be for some researchers with an active interest in the area to create a wiki, whose survival will not depend on the survival of a single person. That is the route that the OEIS took and it seems to have worked out well. One would like to have not just a snapshot of the contents frozen at a single point in time, but a dynamic resource that is continually updated. Failing that, though, a snapshot would be better than nothing. However, according to my understanding, the Wayback Machine is not well designed for something like this where you're supposed to access the content by querying a form.

These two examples are of course only the tip of an iceberg. Scattered across the Internet are all kinds of lecture notes, computer code, databases, blog posts, etc., that are of long-term mathematical interest but that are at risk of disappearing when people retire or die. Even something like MathOverflow should perhaps be archived from time to time in some independent location in case something goes awry with the corporation in charge of it.

Would it be feasible to set up something like the Wayback Machine but specifically targeted at mathematics, so that we could ensure higher quality preservation than the actual Wayback Machine is able to provide? If so, which organizations are best equipped to create and maintain such a resource?

  • 2
    $\begingroup$ Lecture notes could of course be put on ArXiv, but I think people are a bit loathe to do that as it feels like it is being "published" and they may not be comfortable with that. Perhaps some sort of analog that is focused just on hosting content (and providing stable links) so instead of putting your notes on your personal webpage you upload them to this site and put the link on your page...would require some social and promotion acceptance (but ArXiv is pretty standard now so it might not be too hard to do). $\endgroup$
    – RBega2
    Commented Oct 9, 2020 at 13:09
  • 2
    $\begingroup$ Just a comment: A lot of web sites that may appear to be gone have not really disappeared, but have just been reorganized. That is, somebody changed the directory structure and the URLs with it. Universities seem to be particularly prone to doing that. It would be nice if somebody could get it through to the powers that be that hundreds of links become outdated every time they do that, but it is probably a hopeless quest. $\endgroup$ Commented Oct 9, 2020 at 13:34
  • 8
    $\begingroup$ Internet Archive has an awful lot of curated archival projects beyond the automatic Wayback Machine. They might well be interested in supporting something like this. They have the storage and infrastructure, if some mathematician or group were to do the curation. I have some secondhand knowledge of a project along those lines (not in math) and might be able to find out more. $\endgroup$ Commented Oct 9, 2020 at 13:37
  • 4
    $\begingroup$ I wonder if the part about the CoAoC Project might do better in its own post, as people may have some more specific ideas about it. $\endgroup$ Commented Oct 9, 2020 at 13:57
  • 3
    $\begingroup$ @NateEldredge Since you've mentioned Consequence of AC, here is a relevant post on Mathematics: What is current status of Consequences of the Axiom of Choice website? (Asaf Karagila mentioned some information in his answer - perhaps he might be able to update them if there is something new to say.) $\endgroup$ Commented Oct 10, 2020 at 16:31

2 Answers 2


In general, I love the Wayback Machine, so when I read your question, I was very in-favor of it. But, we should also think about unintended consequences. If every random mathematical note one uploads to a personal webpage (including notes for students) gets archived somewhere, then that might have a chilling effect on peoples' willingness to upload such notes. I know that in grad school I uploaded a lot of lectures notes for talks I was giving, and when I looked back later I realized they had errors, incorrect citations, etc. I was glad to have the option to take them down.

A good middle ground, between archiving nothing and archiving everything, would be to archive things that have been cited. This is part of the reason I joined the editorial board of the Graduate Journal of Mathematics, because I think this journal can be a good place for useful notes to get published. When I find an unpublished note that has citations, I often write to the author to encourage them to submit it to our journal. If there was a "Wayback Machine for math" then maybe it could reach out to the authors of such notes and archive the pdf unless the author requests that they not do so. I think once something has been cited, it's important for the community that it be accessible.

  • 3
    $\begingroup$ I do think the comment about an arXiv-like resource specifically for notes is a worthwhile idea. Personally, I would not post anything to the arXiv that's not of "roughly publishable" quality. $\endgroup$ Commented Oct 9, 2020 at 13:52
  • 6
    $\begingroup$ You are greatly overestimating the damage (to you or to readers) from having a low-quality note archived (particularly when the information that the note has been removed is also available). $\endgroup$ Commented Oct 9, 2020 at 14:17
  • 6
    $\begingroup$ @darijgrinberg : The estimate of the damage done by archiving a low-quality note seems less important than the estimate of the damage done by people refusing to make their notes available because of their fear (justified or not) of archival. Though my suspicion is that the "chilling effect" will not be significant. $\endgroup$ Commented Oct 9, 2020 at 16:39
  • 9
    $\begingroup$ The last decade has brought a generation into academia for whom "scared" is a natural state. I've recently had to convince one of the most talented grad students I know that cold-emailing a well-known professor is okay. I don't know what the best way is to fight that scare, but surely confirming its symptoms can't be it. $\endgroup$ Commented Oct 9, 2020 at 21:02
  • 4
    $\begingroup$ Perhaps a retraction mechanism would obviate worries about any chilling effect? \\ Also, although the difficulty of making accessible PDFs suggests any PDF-based changes might be a while in coming, it is easy to imagine metadata added to PDFs to signify a desire that they not be archived, similar to robots.txt that the Wayback Machine already obeys. $\endgroup$
    – LSpice
    Commented Oct 10, 2020 at 16:42

In my previous answer, I focused on unintended consequences of creating a Wayback Machine for math. But I'm worried that answer is preventing us from answering the actual question.

Would it be feasible to set up something like the Wayback Machine but specifically targeted at mathematics, so that we could ensure higher quality preservation than the actual Wayback Machine is able to provide? If so, which organizations are best equipped to create and maintain such a resource?

The Wayback Machine works via web crawlers, and it would be somewhat difficult to replicate what they do just for math. You'd have to help the algorithm decide which pages to archive, and I don't see any clean way to do that. For example, if you only crawled pages with a .edu endpoint, and with math in the URL, you'd miss personal web pages. You could crawl pages where the index page has mathematician in the text, but you'd miss pages about software, paper repositories not tied to a personal web page, etc.

I think a better solution, in the short term, would be to get journals to keep a copy of any web page that is cited when a paper is published, just in case that page goes down in the future. In case the journal can't be convinced, authors could also keep copies of web pages they cite. For every paper I've published, I have a folder with the tex file. I can just make a subfolder where I keep a copy of web pages I cited as of the day I submitted the paper. This would also be a good idea for citing online lecture notes (e.g., Stefan Schwede's notes on symmetric spectra or equivariant homotopy theory) which have many different version and where the theorem numbering might eventually change from what it was when you cited it.

Keeping these folders is very little extra work for us (or for the journals) and feels important to the enterprise of mathematics. To me, it's analogous to how other scientific fields are combating the Replication Crisis, by keeping track of what they did to their data, keeping a copy of the data as it was when the paper was written, and keeping their R code, so that a future researcher could replicate what they had done. I want people to be able to see the same documents I saw when I wrote my paper, so for any that lack a permanent home, I should keep a copy just in case they vanish. If they do vanish, at least I'll have a copy I can share in case someone asks, or potentially host if I have permission (like the Wayback Machine does).

  • $\begingroup$ Could you be more precise about the copy of the webpage ? A shallow copy looses a lot. A deep copy might be very heavy. I am not sure what a realistic middle ground could be. $\endgroup$ Commented Oct 14, 2020 at 11:13
  • $\begingroup$ I suppose it depends on the kind of research you're doing. When I cite something unpublished, it's usually a PDF file. So, that's very light. What sort of heavy stuff are you thinking about citing? By the way, a tool I like for making deep copies of websites is SiteSucker: ricks-apps.com/osx/sitesucker/index.html $\endgroup$ Commented Oct 14, 2020 at 12:29
  • $\begingroup$ I only meant that making a deep copy could be heavier than the first level page, and it is hard to define where to stop ( loops are not a problem). Say university A refered to university B and C that refers .. $\endgroup$ Commented Oct 15, 2020 at 14:56

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.