Disseminating printed documents in many places on the planet is a very robust "backup" strategy for scientific literature. Printed documents (and other forms of ink-on-paper documents), or fragments of them, are known to last for millennia and with a high redundancy (many copies spread in many geographic locations), the chances of being able to reconstruct the original content is high.

All current electronic data storage require continuous catering (electrical power and servers maintenance) or frequent re-copying (i.e. of laser discs, magnetic hard drives, etc.) to last more than a few decades.

How do online-only journals and articles repositories intend to ensure very long-term archival of their content?

  • 2
    @Oxinabox looks interesting, to clarify I'm interested in the actual policy that journals have nowadays. I understand that "They have none" is a possible answer.
    – Cape Code
    Mar 10, 2016 at 13:41
  • 2
    I asked specifically about the arXiv policy here: academia.stackexchange.com/questions/47416/…
    – StrongBad
    Mar 10, 2016 at 14:35
  • 8
    I'm seriously tempted to ask the symmetric question: "Disseminating online documents in many places on the planet is a very robust strategy for scientific literature... How do paper-only journals intend to ensure very long-term archival of their content?"
    – JeffE
    Mar 10, 2016 at 15:00
  • 2
    @JeffE wouldn't the answer to that be the same way they have been for decades?
    – StrongBad
    Mar 10, 2016 at 15:51
  • 7
    Many modern libraries have developed a habit of discarding paper versions of documents because they have electronic versions that take much less storage space. Books go out of print and are damaged, destroyed, lost, or discarded, and publishers go out of business, sometimes before their content can be digitally preserved. My question stands.
    – JeffE
    Mar 11, 2016 at 2:53

3 Answers 3


Most publishers use a preservation network such as Portico or LOCKSS. The principle is that numerous electronic copies are saved, and released in case of certain events (e.g. the publisher stops permanently giving access to the works).


Open access journals and article repositories have a good option of ensuring very-long term access that is similar to how printed documents were handled in the past: Wide dissemination over several repositories.

In the life sciences, open access articles are often deposited in additional repositories by publishers themselves, for example at Pubmed Central or it's European equivalent Europe PMC. I think also additional repositories actively mirror some open access journals - at least I frequently seem to get results of such additional repositories through internet searches. That means even if the publisher goes bankrupt and shuts down its servers, the papers should still be easily accessible through these mirror repositories. See for example Biomed Central's statement on permanency of articles.

This strategy is only possible for open access journals, since only those typically allow redistribution of their articles and thus permit the mirror repositories to act as they do. I have no idea how commercial, non-open access publishers handle this problem.

  • This strategy is only possible for open access journals not really, subscription journals also post pre- or postprints on external repositories.
    – Cape Code
    Mar 10, 2016 at 14:51
  • @CapeCode The key word in your sentence is allow. If they wanted to ensure long-term availability, journals would require wide dissemination over several repositories.
    – JeffE
    Mar 10, 2016 at 15:01
  • Besides, how is dissemination over several repositories better than having multiple replicates of a database on several commercial servers?
    – Cape Code
    Mar 11, 2016 at 5:33
  • @CapeCode Because it's several institutions as well. If a publisher goes bankrupt, I'd expect all replicates to become inaccessible simultaneously, if no specific counter-measures are in place.
    – silvado
    Mar 11, 2016 at 10:19

Currently lib.gen/sci-hub build an archive of all papers and textbooks. Publishers fight the legal battle to stop the project but it's highly unlikely that they will stop lib.gen/sci-hub and it's various mirrors in every country of the world. Elsevier might win the suit in the US but they probably know why they don't pick the fight in Russia where the lib.gen/sci-hub servers reside.

There are applications like IBM's Watson that work better if they have access to more data. That means that companies like IBM, Google, Baidu and Facebook have an interest to have all the data of all scientific papers.

Isn't it expensive to store all that data? No, it isn't. In 2014 Facebook added 4 petabyte of data per day to it's database. That's likely more data than all scientific papers together. All the arxiv PDF's together are 270 GB.

  • 1
    This doesn't seem to answer the question, either at a literal level (sci-hub is certainly not how most journals intend to do long-term archiving) or in spirit (sci-hub faces exactly the same long-term archiving challenges as publishers do, plus the added difficulty of not having all the content in the first place). Mar 28, 2016 at 14:45
  • Note that the 270 GB figure is apparently as of February 2012, and likely is quite a bit larger by now. Mar 28, 2016 at 14:57

You must log in to answer this question.

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