20

I just noticed that arXiv.org is basically running out of monthly IDs, as can be seen in this graph (source [13-12-09]):

enter image description here

They have gone over 8000 monthly submissions 3 times in last 1.5 years. With the steady increase in the number of submissions, it can be expected that it gets over 10000 in cca 2 years. However, the current ID format is restricted to 4 digits, i.e., 10000 submissions.

I haven't found any information regarding this on the webpage. So: How will the submission IDs look like if they don't fit into the scheme? It interests me because in some helper applications, I tend to rely on the yyyy.xxxx format of the IDs.

4
  • 1
    Could please the two down-voters explain me what is wrong with the question? Thanks.
    – yo'
    Dec 31, 2013 at 14:21
  • 1
    Interesting question. I see Prof. G around sometimes in the department. I will bring it up next time I have a chance to talk to him.
    – Shion
    Dec 31, 2013 at 14:27
  • 1
    @tohecz the downvotes don't matter. You've got your answer. Voting is primarily to promote or demote a question on the board, not primarily to raise or lower your cred. Dec 31, 2013 at 14:28
  • 4
    @JonathanLandrum I have 25k rep, so I know what voting is for. However, if someone downvotes a post, they should have a good reason for that. I just want to know that reason ;)
    – yo'
    Dec 31, 2013 at 14:29

1 Answer 1

28

The format presumed in the question is incorrect. The format is YYMM.NNNN, rather than YYYY.NNNN. Shortening the year to YY and including the month as MM gives a factor of 12 increase in the number of available IDs. Version numbers are appended to the ID, as v1, v2, and so on. That said, it is still a problem, but arxiv have thought about it.

NNNN is a zero-padded sequence number starting at 0001 and permitting up to 9999 submissions per month. If current growth rates continue, we expect to change the sequence number to 5-digits NNNNN in 10 to 15 years. We will do this in a uniform fashion so that, likely starting on some year boundary, all subsequent identifiers are zero-padded to 5-digits. We cannot currently anticipate extension beyond that although extension to 6-digits would be possible.

This was posted in 2007 so we are 6 or so years into their 10-15 year window. Submissions of multiple versions might grant some extra headroom, however it does appear that we could run into problems in about two years. It looks like they underestimated their growth a little bit!


Status change in 2015. arXiv.org switched to 5-digit IDs in January 2015. The format is now YYMM.NNNNN or YYMM.NNNNNvV.

7
  • 2
    Thanks. Two comments: (1) I'm aware of vV and yes, I wrote yyyy instead of yymm, my fault. (2) The numbers over 8k are new sumbissions: the largest numbers for the last 2 months are 1310.8658 and 1311.7692, so the original prediction of 10-to-15 years seems to be too long.
    – yo'
    Dec 9, 2013 at 17:17
  • 2
    Well, I think they didn't expect people to use arXiv as a repository for every publication, not mentioning having solely-arXiv-based e-journals, and not mentioning people publishing very short notes on arXiv so that others can cite them. However, in all honesty, +(put a large number here) for the changing attitude towards open-access!
    – yo'
    Dec 9, 2013 at 17:31
  • 4
    another comment: they could use hex numbers, it's pretty standard and 4 digits (64k) would be enough for a very long while...
    – yo'
    Dec 9, 2013 at 18:12
  • 5
    they could use hex numbers — Oh god, please no. ArXiv IDs are meant to be read by human beings.
    – JeffE
    Dec 9, 2013 at 22:03
  • 23
    @JeffE: But it's not clear to me that the numerical value of an arXiv ID is of any real significance to humans. Is 1312.1976 really more meaningful to you than 1312.07b8? Does the fact that the ID is 1976 and not, say, 1796, have any real significance? Dec 9, 2013 at 23:03

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .