A&C and Elsevier


Older: v1

Back in 2012, Bob Mann and I pushed ahead with setting up Astronomy and Computing (A&C) as a journal for the astronomical computing community. We ended up getting Elsevier to publish the journal, with a hybrid access model. This overlapped with the ‘Elsevier boycott’. Below is my brief recollection of why we went ahead by this route. It's not intended to persuade anyone else that we were right or wrong (though I don't feel at all bad about it), or change anyone's mind, but instead to record that it wasn't a thoughtless decision.

I also don't plan to engage with the whole argument about (different colours of) open access or with various publishers' various wickednesses, nor really engage other than tangentially with the whole sorry serials crisis. I will dreamily recall, though, my confidently asserting that traditional journal publishers were clearly on the very point of collapse, and would definitely be gone before, say, 2000. Ho hum.


Bob and I organised a ‘birds of a feather’ session about ‘astroinformatics’ publication at the 2010 ADASS, described in the proceedings and on ArXiv (1103.1982). The problem, in a nutshell, was that people in this community were increasingly expected to publish in journals rather than semi-formal conference papers, but that the main astronomy journals were disinclined to publish primarily software articles, and the outlets we did use (eg PASP) didn't have the publication statistics we felt we needed. That conference paper describes our fairly informal survey of half-a-dozen journals, where we sent abstracts of a range of representative ADASS papers and asked ‘would you desk-reject this paper?’ (answer, most of the time: ‘yes’).

The mood was broadly in favour of new publication outlets and new publication types (eg ‘what worked’ articles), but stuck to the ‘let's publish more in PASP’ line mostly, I think, as a result of a strong conservative streak in astro computing. We also talked about open journals on that occasion, but the feeling in the room was strongly sceptical of the operational feasibility of relying on volunteer effort. We would very possibly have got a different conclusion from a similar BoF at a different conference, but ADASS is a pretty established conference in this circle, so probably a plausible indicator of community sentiment. It's probably relevant that a good chunk of the participants at ADASS are science operations folk, worrying about questions like ‘can we commit to this design/schema/format/SLA for the next 20 years?’

A year on, the hinted flood of PASP papers hadn't materialised. We took three things from this, and from other conversations at the time:

So this seemed to leave us needing a traditional-looking journal – any innovation in this line wasn't going to be in publication methods.


Our first instinct, of course, was to set up an open-access journal, using suitably open-source journal management software. Exercising heroic self-restraint, we decided against writing such software ourselves, and from a heart-to-heart with a colleague who'd gone down this route learned that:

I remember the figure of half an FTE of effort to keep the journal working. I can't remember if that was paid effort – and thus money that would have to be raised – or an estimate of the effort gifted to the project from the day-job, but in either case it was a soberingly large number.


In a way, it shouldn't have been a surprise. As everyone knows, who hasn't organised a conference, running a conference consists of nothing more than booking a room and putting up a web page. As everyone knows who has run a conference, there's a teensy bit more to it than that, involving a silly quantity of effort which it's almost impossible to account for afterwards. I think it's the same with publishing.

A couple of years after these events, I was talking to an editor at a (successful, cost-covering) academic press, who said that the cost price of scientific journal publishing was well north of £150 per paper. How much?! It seems that DOIs, feeds to Scopus and the like, ORCID integration, and all of the other tedious-but-fiddly back-office IT does add up alarmingly, that authors will not write sane LaTeX, that there's no APC revenue from rejected articles, and ten minutes later I'm whimpering slightly, trying to dig my shoulderblades out of the wall, and deciding that bitching about authors' bad LaTeX is a more fun direction for the conversation (yes, $\section{\gamma}$ will indeed get you bold maths, but no, I agree that's probably not great for long-term archivability please put down that bottle). But just as I'm unable to pin down, in retrospect, just what bit of being local conference chair will give you a nervous breakdown, I still can't put my finger on what specific bits of the journal publishing workflow are so irreducibly costly. They just... are.

There's an discussion of the economics of journal publishing in a Nature article from 2013 (10.1038/495426a). That's presumably at least a little out of date, but it can be handily summarised as: it's complicated.

Another data point comes from the Journal of Open Source Software, who have published a blog post describing both their actual costs (negligible), and an estimate of what their costs would be if they weren't relying on grants, volunteer effort, and effectively donated services. That estimates a cost per paper between $100 and $1300, depending almost entirely on whether or not, and how much, you pay editors. They keep costs down by requiring authors to produce camera-ready copy, by doing no copy-editing, and not paying editors; A&C, for completeness, provides post-acceptance copy editing and english-language editing, and pays editors (since you ask, above minimum wage iff you're time-efficient) but not reviewers. There are various positions one can take on the question of ‘editor compensation’.

Even without too much detail about the costs, the problem was that neither Bob nor I particularly wanted to run a journal, in the sense of operations rather than editing. We both had enough experience with the relevant sort of operations to know that the available quantum of giggles would be exhausted pretty promptly, and enough experience with academia to know that the associated glory (while of course undying) would set no heather alight, and the associated professional merit, though non-trivial, would be modest. Running a journal is an important job which we're all glad to praise someone else for doing.

But then, shortly after we published our thoughts about this, it happened that an Elsevier publisher approached us, having heard about our ruminations through the usual academic gossip networks, and said ‘I couldn't help noticing your paper...’. ‘But, the boycott...’. ‘And...?’ So on to the next phase.

Is this, then, a Good Thing?

Well, it's an adequate thing.

Without a publisher to take a commercial interest in this, I doubt the journal would have appeared. Perhaps a free (speech and beer) model could be found for this, but neither of us came up with anything that sounded plausible, and in any case we'd already formed the impression, as I mentioned above, that the community wanted something that looked respectable in a rather old-fashioned way. We hadn't decided ‘commercial or bust’, and I don't recall us shopping the proposal around, but if Elsevier hadn't turned up, I'm not optimistic that the project wouldn't have ended up in the bucket of life's ‘maybes’.


And Elsevier? Yes, they're a commercial publisher whose goal is to make money out of this sort of activity. Unlike many publishers through the ages, they currently seem rather good at the profit-making bit. I am reminded of PTerry's description of ‘Mr Goatberger', who, in publishing Nanny Ogg's dubious recipe book,

dreamed the dream of all those who publish books, which was to have so much gold in your pockets that you would have to employ two people just to hold your trousers up.

And you may recall that Gutenberg's bible project went over budget by a factor of several, whereupon he fell out with his venture capitalist/publisher, went bankrupt, and had to take his technology elsewhere (is this a cautionary tale about the vagaries of print, or just a reminder that nothing ever changes?).


Is profit-making wicked? For some social necessities – firefighters, healthcare, clean air, the drains and their associated ecosystems – yes, I think it probably is. For luxuries, not so much. Now, ‘knowledge’ in the abstract clearly is a social necessity (and I'm willing to be quite bang-on-the-table dogmatic about that); but for any particular instance of that knowledge – a journal on astronomical computing, to pick an example at random – I'm slightly less certain. There's a spectrum of ways of making a profit, some of which are more benign than others: if I found myself in Mr Goatberger's career (ideally sharing his problems with pockets), I'm confident I'd sleep happily at night.

The publication process

Is this the best way of organising all this? No, surely it really cannot be – there's got to be a less deranged way of getting text from academic writers to readers. The chain of payments from funders to institutions to authors, from multiple institutions and haggling consortia to publishers to printers and editors, and non-payments from publishers to referees/authors, is so far from the maximally efficient linear value chain that enthusiasts for capitalism insist must necessarily emerge if only the market can be free enough, that it's easy to feel that someone (possibly the enthusiasts) is being mocked.

Whether or not the current setup is justifiable or not, the problem is that it feels like someone's being rooked, so everyone involved ends up cross and shouty.

I should mention that Plan S is a Thing (it's a Thing which I currently have zero coherent opinions about).

For what it's worth, there is at least one possible alternative, that has the twin virtues of actually existing stably, and being describable without sounding like a gibbering idiot. That is SCOAP3. Here, a consortium of funders of HEP science have negotiated an agreement with a consortium of journal publishers – crucially including the three or four key HEP journals – so that the funders will give the publishers a large wodge of money, after which the communities can both publish in those journals for free, and read the contents for free. This cuts out the reluctant middlemen, means libraries don't have to haggle separately with journals and feel permanently but non-specifically diddled, means academics don't have to feel that they're taking part in some obscure not-quite-fraud they neither deeply care about nor want to understand, and means publishers get a secure contract to get money for the fiddly layout, editing, and archiving work they've made their own.

What I don't know is how astronomy would get there from here.

SCOAP3 works for HEP because (a) there's a small number of ‘key’ journals that have to be dragged into the consortium; and (b), crudely, HEP has CERN, which has a role to play in catalysing that community's haggling of multi-year funding and planning. It would be harder work in astronomy. But it gets the material out there, avoids authors having to think much about publication and, as a system, doesn't look permanently on the verge of rapid unscheduled disassembly. Which is good.

Norman, 2019 June 1

Updated 2019 June 9: Added a couple of further details and links