Preprint servers face closure because of money troubles

Why on earth would anyone practice the same fee structure for repositories in developing countries as in developed countries? :confused: 1000 USD is about 5-6 times more than the monthly salary for many faculty members in Indonesia…

"INA-Rxiv, ArabiXiv, AfricArxiv and IndiaRxiv are run by volunteers around the world, but the servers are hosted online by the non-profit Center for Open Science (COS), based in Charlottesville, Virginia. The centre’s platform hosts 26 repositories, including more than a dozen that are discipline-specific.

In December 2018, the COS informed repository managers that from 2020, it would be introducing fees, charged to repository managers, to cover maintenance costs. The charges, which were finalized last December, start at about US$1,000 a year, and increase as repositories’ annual submissions grow."

https://www.nature.com/articles/d41586-020-00363-3

2 Likes

Statement from AfricArXiv on the matter:

I can see Nosek’s point in wanting to have a sustainable financial model for the service, but it does seem unfair for developing countries.

My initial suggestion was going to be to charge a heavily discounted fee (maybe scaling up from $100) to repositories serving developing regions (ArabiXiv, AfricArxiv, etc.) and raise the price for other repositories accordingly. But it sounds like other field-specific respositores (MarXiv, EarthArXiv), which presumably have some repository managers from developed countries, are also going to close because of the fee.

Clearly people value having pre-print repositories but aren’t generally willing or able to pay for them. Maybe a different funding model needs to be brought in, such that national funders from developed countries that encourage preprint submission are asked to pay the costs of all repositories. $230,000/year sounds like a lot but, realistically, it wouldn’t be much for a national funding agency in the EU or US to cover.

1 Like

Subsidizing these fees for preprint communities from developing countries (or just countries with very low resources for science) sems like a no-brainer to improve diversity in open science.

I’m thinking about different routes for this. Ultimately their independence and agency to run their communities is extremely important.

Anyway, for the time being it’s easy to solve in short term. Small amounts and people interested in diversity in open science can chip in a small amount.

Long term we need solutions.

2 Likes

From the Nature piece:

Dasapta Erwin Irawan, a hydrogeologist at the Bandung Institute of Technology who helped set up INA-Rxiv, says his repository received more than 6,000 submissions between July 2018 and June 2019, so the fees will come to about $25,000 per year, which he cannot afford.

$25,000 is a lot of money, but $4 per submission doesn’t seem too much to ask someone to contribute, even from a developing country. I’m a big fan of differential pricing, but I don’t like stuff being totally free in general because it will always get abused. If it cost someone from a Western university $20 to post a preprint I don’t think it would exclude many people — for four authors it’s the price of a beer each even if the lab/uni won’t pay. Of course, you’d need a payment system, but those aren’t too hard to add to web sites these days.

2 Likes

I would share this conversation with the indonesian open science team, who were quoted in the article, and hope they have time to respond. :slight_smile:

Will this be the beginning of the end of preprints?

I might be wrong, but the concept of preprint could (and should?) be a temporary one. Preprints have surged to give access to reports that are otherwise behind paywalls, or to disseminate preliminary findings before peer-reviews. However, the various archives are affected by some issues that could define their temporary nature:

  • They are not an alternative to peer-review processes
  • They could get filled with “junk” very easily
  • They are hosted by OSF (if it goes down, all the preprints go down with it)

Now, why should I pay OSF to host a preprint when I can put it on my university website, personal website or other clouds? Does the fact of being hosted on an official archive give my preprint more credibility?

Personally, the scientific community should aim to create repositories that also allow for collective peer-reviews. This would be an effective and better alternative to paid journals.

Meanwhile, publicly funded repositories for sharing peer-reviewed articles could be the best alternative.

I do not know how much it costs to run a server that hosts 6000 small files more per year, but that amount looks a lot of money to me too.

1 Like

As i said to my indonesian open science teammates, i found these publicly-funded examples to be very exciting. I encourage them to start the same example in indonesia.

http://research.un.org/ld.php?content_id=51248380

http://research.un.org/ld.php?content_id=51248392

I do not know how much it costs to run a server that hosts 6000 small files more per year, but that amount looks a lot of money to me too.

I don’t find the $25,000 to be excessive. It will depend on the extent to which COS spreads its overheads across its different operations. The size of the file isn’t the issue — it’s maintaining the code that allows access to them, plus keeping server operations up 99.9% of the time. That means having someone with a pager who can be woken at 3am local time when someone in Australia wants their files.

Over the last 20 years I’ve seen a lot of “open” projects founder, because people thought that free and open source software, combined with cheap hardware, also meant cheap operations. It turns out that the kind of people who are prepared to be on call at 3am like to be paid for that. :smile:

2 Likes

Do you mean for all disciplines or for the services hosted by COS? Arxiv.org has existed since 1991 and is hosted by Cornell Uni. We also have the raise of journals such as F1000 Research that employs new article and review formats. I’m convinced the positive development will continue, with and without COS.

Because personal websites and most(?) university websites are not linked to scholarly databases which means that even though the paper exists for free out there somewhere it can still not be easily found through traditional database searches. For example, I have made all my own papers publicly accessible on my personal website, but when searching for my name or the specific article in databases those versions won’t show up, only the paywalled ones. Which means that my articles are still not really open access in practice, only in theory. The preprint servers take care of that.

I thought so too initially but @dbernt corrected me: apparently the 230K USD that @briannosek referred to is a reasonable cost basically covering 2 full-time programmers to maintain the servers, if I understood correctly.

1 Like

Thanks for tagging me. If you are interested, here is a spreadsheet with info on how we calculated our costs for maintaining the service: https://docs.google.com/spreadsheets/d/1V0vKrf50K667CqM3e4S2itlFTdSMAqKAz17gJKxjVjE/edit#gid=0

And, here is a bigger picture blog post about the economic value for supporting preprints: https://cos.io/blog/conflict-between-open-access-and-open-science-apcs-are-key-part-problem-preprints-are-key-part-solution/

3 Likes

Also, the Nature article was a bit more doomsday than the reality. Here’s something I posted in a FB thread about it…

"I want to point out two inaccuracies in the article: (1) The title suggests closure is looming, and (2) the article attributes to me a statement that individual services will close if they don’t contribute by the year end. The reality is much more optimistic than both of those statements.

In brief:

  • Our present funders have been clear that they do not want to indefinitely sustain 100% of preprints, so the cost model is aimed to help tap the diversity of community resources that are available to cover the collective costs of maintaining the service.

  • It is not known where those resources will emerge, but it is unlikely that it would be equivalent across services, particularly for those in the developing world. A likely scenario as the sustainability model iterates is introduction of waivers and/or scaling by GDP as it becomes clearer where there is appetite from funding sources to provide support.

  • Services that are working collaboratively on identifying funding sources are not facing imminent closure. Some efforts will not yield resources, others may be abundant. The collective success will ensure sustainability, and the cost model will adapt to where the resources are available.

  • Some services do not wish to participate in the new cost model. That’s totally ok, and we can help them with a transition to a service that fits their desires.

  • The early signs of this more collaborative fundraising is very encouraging. Some services have already secured resources for their service, and a few have helped to tap engagement with library consortia that may cover a substantial portion of the total cost maintenance. Based on present fundraising numbers, I see it being very unlikely that any service would need to close in 2020 (though some may still wish to move to another service for their own reasons).

Also, I see lots of positive engagement from the community about identifying other solutions for communities that want to offer preprints. That is very good news from my perspective. We need some diversity of services to meet the varying needs across communities."

3 Likes

This post is very illuminating. I see the point of preprints shifting the emphasis of peer-reviewed reports on evaluation rather than publication. However, this is not entirely true in my opinion:

To this date, the publication/evaluation dyad stays in the hands of academic journals. This will be the case until other forms of published material will be cited in researchers’ works. Right now, at least in my field, I have never seen a preprint cited in a report. Hence, preprints at the moment do not solve the problem of publication but the one of accessibility .

Preprinting opens the door to decoupling these as discrete services that can emerge as competitive markets of their own. Journals can offer some or all of these services as a package or as individual services.

This is, in principle, an interesting solution. The problem lies, as usual, in the human nature. Imagine (private and paid) publication/editing services, peer-review services, marketing services. I can already see the competition and lobbying creating divisions between researchers who can afford this or that service, and research fundings spent to have the best peer-review, best marketing etc. because this is what will give your research good points and allow you to advance in your career.

Publicly funded archives, with ongoing and communitarian peer-review processes (with all reviewers converging in one archive to assure the readiness of the process, transparent reviews and the possibility for the community to judge the review process and the reports) are for me the only real solution to publication and evaluation.

Obviously, this is utopistic in a society where the distinction between “public” and “private” is more and more subtle (with good and bad outcomes). However, given that the costs of archives come mainly from server maintenance and hosting-related services, why not trying to turn archives into real alternatives to journals? (As a reviewer, I would much more in favour to provide reviews to not-for-profit archives rather than journals.)

1 Like

I’ve cited my own pre-prints in other papers and grants before, but I also don’t recall every seeing an article cite a pre-print that wasn’t by one of the authors. So I was surprised to see the following article that suggests pre-prints get too many citations!

1 Like

thanks for sharing our statement, Rebecca - here is our current crowdfunding effort, incl. for OSF: https://opencollective.com/africarxiv. Feedback and suggestions welcome - this is work in progress.

@briannosek so can you confirm that our expenses are already covered so we will be able to reassign the funds that have come and might still be coming in on the OSF tier and be allocated elsewhere in our operational costs? We urgently need income for overhead for team members, for necessary tasks such as communication, partnership building, infrastructure dev, and overall maintenance or your services to the African scholarly community.

Hi @jo.havemann. If you have already submitted your 2020 contribution, then you do not need to do anything else for supporting maintenance this year. All fundraising can be focused on your other needs and/or developing cash reserves for 2021 and beyond. We will focus on working with the other services that haven’t finished fundraising for 2020 yet. If you haven’t submitted that contribution yet, then yes, for the distributed model to function well, we need every group that is able to fundraise to contribute to the maintenance costs. If your forecast looks like you won’t reach the target, then the sooner you share that with our team with a confidence interval, the better. Then, they can help and problem solve. Brian

Has anyone managed to actually host the OSF platform somewhere else? I get reports that this is not actually possible, for all the usual reasons. Being “open source” and being portable are not necessarily the same thing. Just wondering if there’s a success story out there somewhere.

1 Like

Thanks for your response, @briannosek. We’ll be in touch to discuss.

I believe there was a launch of a new platform in early 2019 or 2018 and they were reusing OSF and even kept the whole look. Supposedly built on Ethereum (blockchain). But I can’t recall the name of the platform.

Thanks Rebecca. If you remember a URL that would be great.

1 Like