An efficient journal (2012) | Hacker News Reader

An efficient journal (2012)

blogs.harvard.edu

55 pointsjampekka18 comments

@dougmccune makes a great point that "with the right conditions a quality journal can be run incredibly cheaply." Discrete Analysis is prime example of this (and there are others), the editors have been able to run the journal on a small grant using Scholastica for peer review and publishing via the arXiv model. Full disclosure - I am a co-founder of Scholastica. I think the key here is that in order to make journal production cheaper it has to be _despecialized_. Journals have come to reply on corporate publishers because printing and early online publishing were too cumbersome for them to manage on their own. In turn publishers have built up levels of specialization for their role. But today, with existing technology and services, journals can assume a pared down publishing model that addresses their needs in a much more affordable way and allows them to retain control of content and its distribution.

As someone involved in publishing (but on the literary side) I've always made the point that if we reduce the value that we provide to something mechanical, we will eventually become dispensable.

The key to that kind of publishing (and I believe this story illustrates that it matters in academic publishing also) is providing a framework for the best to come to the top. That means supporting, encouraging and publishing good authors.

The JMLR has the great unspoken advantage that all of the people involved in its production have a job that allows (or encourages) them to pursue this project also. That won't always be the case. Often the key people involved will have to have salaries if the enterprise is to thrive. But the point remains the same: if you measure your value by the bureaucratic production aspect then eventually people will simply say: what do we need these guys for?

> The JMLR has the great unspoken advantage that all of the people involved in its production have a job that allows (or encourages) them to pursue this project also. That won't always be the case.

I'm not sure what you're saying here. The vast majority of hours have always been donated by referees. Then there are the editors, who are either volunteers or paid only a small amount. Aside from that, it's a matter of posting the papers on a website.

Well I'm saying that all these people have a livelihood. Within that they are able to afford to volunteer time. That is a somewhat privileged position to be in. In a really good and positive way.

Nonetheless a lot of people that want to do important work in one field or another, need to be able to make that their livelihood. Usually because what they do has not got the automatic financial value attached of something like Machine Learning.

This is definitely a good example of how with the right conditions a quality journal can be run incredibly cheaply (on the order of < $10 per published article). And it looks like JMLR has a better impact factor (in the 2.4 range) than the Springer journal (IF ~1.8) that the editorial board resigned from [1]. So they're obviously doing it incredibly cheaper, with good results. I don't think you can extrapolate from this that the same model can be applied to every field (as the author acknowledges), but it's certainly a good example to try to emulate. Another would be Discrete Analysis [2] for an example in Mathematics (also a field well-suited for efficient publishing).

I know this submission comes on the heels of the discussion of scholarly publishing on HN a day or two ago [3]. It's certainly a good counterpoint to my previous argument that you can't run a journal super cheaply, although I'd argue that one or two cases in one or two fields don't prove you can scale such a system to work for all of academia. But it certainly shows that it's possible to do at a small scale, and maybe there's someone clever enough to figure out how to scale it up.

[1] http://www.springer.com/computer/ai/journal/10994 [2] http://discreteanalysisjournal.com/ [3] https://news.ycombinator.com/item?id=15265507

There's also the journal Quantum for research in quantum science: https://quantum-journal.org

Discussion about APC pricing would be a lot more fruitful if there was some honest and open discussion about how laborious the publication and it's different work phases actually are. As I said, and as you verified, in the previous thread, no publisher wants to give any breakdown of the costs. This post on JMLR is one of the few where an outsider can actually get some kind of grasp of what it actually takes to publish a journal, and as seen, it isn't that much.

If the actual cost structure was better known, it would be a lot easier to evaluate if the price level is reasonable. Also, it would give a push towards "unbundling" most of the APC. Eg, the author could do the typesetting themselves (as in JMLR) or buy it from somewhere else for cheaper, and thus save grant money that could be used to pay at least something to a grad student or a postdoc currently working for free instead of, as I cynically assume, line up pockets of rich investors and overpaid execs.

I find it a very odd claim that somehow doing publishing at large scale should be MORE expensive than doing it on small scale. My hypothesis is that it's not the production costs, but that at the moment academic publication is so ridden with network effects that publishers can ask for greatly inflated prices to make huge profits and/or be done very inefficiently. It's also such a good business that journals/smaller publishers tend to get bought out or at least invested into by larger companies (most of this was actually done in past couple of decades), who of course have an interest to keep the profit margins as high as possible. And this can happen quite naturally even without full monopoly or cartel, although I wouldn't be surprised if there's some kind of "common understanding" among the big publishers that price wars aren't good for any of them. Also, it's one of those businesses where almost nobody actually pays anything from their own pocket directly, which generally isn't very good for price competition.

But in the end, I find most of this discussion sort of moot. I quite strongly believe, and can't seem to get any evidence to the contrary, that the actual labor going into the publishing is so low that the industry in it current form shouldn't be, and hopefully won't be, sustainable given the current, let alone future, technology. But OTOH, even if I'm correct, they have huge lobbying budgets, democracy is in such a sorry state and people are so indoctrinated to be selfish, short sighted greedy creatures, I wouldn't be too surprised that the industry can keep the racket going for decades.

Here's a good article on why it's hard to calculate per-article costs: http://cameronneylon.net/blog/pushing-costs-upstream-and-ris...

But yes, prices are not a reflection of cost, but of how much publishers can ask. Unfortunately, given that publishers sell subscriptions in bulk ("Big Deals"), there's no downwards price pressure: if they add another journal to the deal, they can ask more money, without that leading to libraries cutting subscriptions elsewhere. There's no real pressure for APCs as well: authors want to publish in "reputable" journals, and don't pay the APCs themselves, so publishers are free to raise them quite considerably.

Perhaps this is my bias showing, but if you read through this whole interaction [1,2,3], Kent Anderson comes across as a real dick, while both Yann LeCun and George Monbiot come across as pretty reasonable albeit ticked off about academic publishing (imo quite rightfully).

[1] https://www.theguardian.com/commentisfree/2011/aug/29/academ...

[2] https://scholarlykitchen.sspnet.org/2011/09/01/uninformed-un...

[3] https://blogs.harvard.edu/pamphlet/2012/03/06/an-efficient-j...

I often feel the same, although like you, I'm afraid it might have to do with my bias. The Scholarly Kitchen is an interesting place to catch up with the views of the publishers, but interactions like this and articles like most of Joe Esposito's don't really help with the image of "evil Elsevier et al"...

I feel like the "let authors pay for copy-editing themselves, if needed" could be extended to typesetting as well. I can't imagine most manuscripts being _that_ hard to read in the author's format, especially not if they know that it might affect how often they get read/cited.

The field of author services (as actual paid-for services, not bundled with journal publication) is an interesting field. Peerwith [1] is one new entrant into the field, which is part of the Springer-Nature incubator Digital Science. They're sort of a marketplace for academics to connect with (and pay) other academics to help with manuscript services, everything from peer review type work to copy editing.

[1] https://www.peerwith.com/

I agree, and if it became commonplace, I suspect people would offer templates that would help homogenize things.

> In fact, JMLR, Inc. didn’t even have a bank account until recently; there was no need.

Heh. Those Elsevier and Springer profits look gross but from the other direction. What are they doing with all that revenue?

Make profit margins of 30%-40%... [1]

(And also sponsoring academic conferences and societies, etc. But mostly profit.)

[1] https://www.theguardian.com/science/2017/jun/27/profitable-b...

These academic conferences are insanely expensive (typically ~$1000 tickets) and quite poorly run, especially compared to industry conferences. Industry conferences are not only cheaper and better run, but they even have niceties like coffee (!!) and the occasional snack. The last academic conference I was invited to speak at charged me $1000 as an invited speaker – which I paid, but never will again. There was no coffee or organization at the breaks. In short, I'm quite certain academic publishers are also making a tidy profit off of their conferences.

My point was given how little they do, even a 40% profit margin is hard to justify. It should be in the 70-90% range!

Well, OK, there's more. For example, like the author relies on submitters delivering a properly typeset article (using LaTeX), traditional publishers often discard the original formatting, re-typeset it using an XML-based format that can then produce HTML and PDF outputs again (in the journal's branding). In this process they often introduce errors, and one can question whether it's actually necessary. Then again, the publishers are realising that themselves as well, and are more and more often relying on author-typesetting as well - which in that of course boosts their margins some more.

Reply on news.ycombinator.com