Science in the Open

This post was prompted by Donna Lanclos tweeting a link to a talk by Eamon Tewell:

https://twitter.com/DonnaLanclos/status/992843570238377984

His talk, on the problems of deficit models chimed with me on issues of tacit knowledge. I’m still noodling around an underpinning theory of knowledge for my work (blog post currently has spent nearly 12 months in the draft folder). The core to the model is that (general? non-local?) knowledge is made when local knowledge is translated across group boundaries.

This draws heavily on people like Ludwig Fleck and Jerome Ravetz and indeed most social models of knowledge. Ravetz makes two critical observations. First that locally bound contextual knowledge is rooted in what he calls “craft skills” and what Harry Collins would refer to collective tacit knowledge. In our models this is language bound, context dependent, and often tacit. Second that the (attempted) transfer of this knowledge across group boundaries is lossy.

This family of versions of the facts[…]will show diversity in every respect: in their form, in the arguments whereby they can be related to experience, and in their logical relation to other facts. They will however, have certain basic features in common: the assertions and their objects will be simpler than those of their original, and frequently vulgarized out of recognition.

Ravetz (1971) Scientific Knowledge and its Social Problems, p234, emphasis added (1996 Edition, Transaction Publishers)

There is abstraction and generalisation, but also loss of specificity, and loss of detail. Generalisation (reproducibility amongst other things, for instance in Collins’ example of building a specific form of laser) requires that we make the tacit explicit, that we convert collective tacit knowledge, first to relational tacit knowledge and then to explicit statement.

This model works for me because I’m interested in the question of what it is in the culture the academy, or of scholars (not the same thing!) that makes it successful. Where I’m getting to is that we share a culture that provides some common practices and scaffolding which forces us to attempt to make our local knowledge more general (publication for a start) while at the same time pushing us into groups and communities that don’t share language and scaffolding so that we are forced to work at this translation process across groups. My hypothesis is that our long standing culture and practices actually manage this tension successfully and do so at all scales.

So what does this have to do with deficit models? We never talk about the deficit model within the scholarly community. We, by definition don’t have a deficit. Having the thing is how we define the “we”. Within the scholarly community we are (more or less) happy to do the work of translation of trying to explain our context for those who do not understand it because it is part of what we do. Complaints about referee #2 and “those damn scientists” aside we subscribe to a rhetoric of how much we benefit from explaining our thing, that “if you can’t teach it you don’t know it”. Making the tacit explicit – including the tools we apply to force ourselves to do it; teaching, reproducibility, data sharing, peer review – is a core part of our culture when dealing with each other.

But when we talk to outsiders we invoke deficit models. These fail (as various responders to our tweet conversation helpfully provided links to the literature on) precisely because there is not a deficit on their side, of knowledge, there is a deficit on our side of effective translation. Translation is not merely dumbing down here but the work of making concepts relevant, embedding them in local scaffolding, seeking to identify or build boundary objects that can be shared across the border between communities.

If this works then there are a couple of useful things that arise. One is that deficit model rhetoric is a signal that the boundaries of “our” scholarly community have been reached. That’s more useful than you might imagine. Defining where we start and end is not a trivial exercise. Second that working to identify and understand the cultural practices that seek to make the tacit explicit within our communities takes me to the core of my problems of interest. I’ve already speculated a little on the role of teaching here but there is more to this.

What is perhaps most interesting is that we might be able to make something positive out of the failures of deficit model thinking by using failure modes to help us understand the translation deficit. Something that I really haven’t worked out remains the challenge of resource allocation between translation focussed on a known group and that for potential users whose characteristics are as yet unknown. This again, is not a new question but one that I’ve been grappling with for a long time. This might provide a new way in.

It’s not my joke but it still works. And given my pretty much complete failure to achieve even written resolutions it’s probably better to joke up front. But…here are a set of things I really need to write this year. Maybe more for my benefit than anyone else but it’s good to have a record.

Blog Posts: I have a few things that either need finishing or need writing, these are relatively immediate

1. Against the 2.5% Commitment – argument that fixed top-slicing of scholarly communications budgets is not the right way to think about funding infrastructure and support services. Rather we need some market-like social incentives that are internal to organisations. My proposal is a shift to thinking about budgets as investments, including the extent to which they recirculate value or money to the scholarly community.
2. The local context of peer review – I owe Hilda Bastian a more detailed response as to why, while I agree with her analysis of where the problems lie and what the optimal state is, that the studies we have of peer review don’t show what we might think. Part of the reason for trying this is that I need to make this argument in the broader context for a book (below)
3. Some hard truths on APCs, exclusion and charity – There is a received wisdom circulating that APCs are the source of all problems with exclusion. I think, while APCs have their problems that this is looking in the wrong place, in fact that APCs may surface issues of exclusion in a way that could be ultimately positive. The real risk of exclusion lies in work-flows and no-one seems to be focussed on the interaction of structural power in publishers and submission systems and the risks they pose.
4. Sketching a theory of knowledge – first step in working out the article (4 below) to get some ideas out and down

Articles: Things that really need writing, mostly have been sitting for a while

1. Institutions for productive conflict in knowledge-making – working on this at the moment. It’s not really feasible to describe the argument in a few sentences, which means I haven’t got it fully refined yet! It’s a much evolved version of parts of the argument from this paper from ElPub last year.
2. The structure of scientific evolutions – we have a theory of everything (in knowledge communications and production). It’s quite exciting, and quite simple in outline, complex in its implications. A draft of this exists but it may be a small book rather than an article. I desperately need to do the background reading to properly support the arguments.
3. A clash of peer review cultures – a kind of auto-ethnography of the experience of the peer review process for the Excellence paper. There were many weird moments when an attitude amongst the researchers involved (including editors and referees) in favour of open review clashed with the journal’s assumptions of double-blind review. Mainly blocked because I don’t know diddly squat about doing ethnography properly and need to talk to someone who does…
4. Repertoires, learning and social theories of knowledge – I’ve been using and developing an implicit social theory of knowledge that is really just a mix of notions of community from a wide range of different places. It needs to be properly fleshed out, ideally with someone who knows what they’re talking about. This is likely more review/synthesis than new ideas per se, but as far as I can tell that synthesis doesn’t exist across the silos I’ve been looking at.

Books

1. The core piece of work I need to do on the book I should be focused on. It is code-named Telling Stories with the subtitle A personal journey from the sciences to the humanities (and back). The first (very rough!) drafts of the introduction and first chapter I posted as Telling a Story and Portrait of the Scientist as a Young Man.
2. Possibly pull together a series of texts into a thing on Networked Knowledge. I put together a “book” based on old posts and reflections a few years back. That doesn’t in and of itself merit formal publication (that’s the view of the referees incidentally) but, particularly with the resurgence of interest in “federate all the things”, there are a series of texts that could be pulled together in a kind of manifesto of the technological possibilities.
3. Political Economics of Scholarly Publishing – I’ve been occasionally adding to this list of posts, and this new paper is relevant, but that work needs drawing together in some form. Might be worth seeking a grant to cover the time to finish it off properly perhaps…

That looks like a long list…

This is a piece I wrote for Jisc, as part of a project looking at underpinning theories of citation. There are a few more to come, and you can read the main report for the project at the Jisc repository. This post cross-posted from the Open Metrics blog.

Citations, we are told, are the gold standard in assessing the outputs of research. When any new measure or proxy is proposed the first question asked (although it is rarely answered with any rigour) is how this new measure correlates with the “gold standard of citations”. This is actually quite peculiar, not just because it raises the question of why citations came to gain such prominence, but also because the term “gold standard” is not without its own ambiguities.

The original meaning of “gold standard” referred to economic systems where the value of currency was pegged to that of the metal; either directly through the circulation of gold coins, or indirectly where a government would guarantee notes could be converted to gold at a fixed rate. Such systems failed repeatedly during the late 19th and early 20th centuries. Because they coupled money supply – the total available amount of government credit – to a fixed quantity of bullion in a bank, they were incapable of dealing with large-scale and rapid changes. The Gold Standard was largely dropped in the wake of World War II and totally abandoned by the 1970s.

But in common parlance “gold standard” means something quite different to this fixed point of reference, it refers to the best available. In medical sciences the term is used to refer to treatments or tests that currently are regarded as the best available. The term itself has been criticised over the years, but it is perhaps more ironic that this notion of “best available” is actually in direct contradiction to intent of the currency gold standard – that value is fixed to a single reference point for all time.

So are citations the best available measure, or the one that we should use as the basis for all comparisons? Or neither? For some time they were the only available quantitative measure of the performance of research outputs. The only other quantitative research indicator being naive measures of output productivity. Although records have long been made of journal circulation in libraries – and one time UK Science Minister David Willetts has often told the story of choosing to read the “most thumbed” issue of journals as a student – these forms of usage data were not collated and published in the same ways as the Science Citation Index. Other measure such as research income, reach, or even efforts to quantify influence or prestige in the community have only been available for analysis relatively recently.

If the primacy of citations is largely a question of history, is there nonetheless a case to be made that citations are in some sense the best basis for evaluation? Is there something special about them? The short answer is no. A large body of theoretical and empirical work has looked at how citation-based measures correlate with other, more subjective, measures of performance. In many cases at the aggregate level those correlations or associations are quite good. As a proxy at the level of populations citation based indicators can be useful. But while much effort has been expended on seeking theories that connect individual practice to citation-based metrics there is no basis for the claim that citations are in any way better (or to be fair, any worse) than a range of other measures we might choose.

Actually there are good reasons for thinking that no such theory can exist. Paul Wouters, developing ideas also worked on by Henry Small and Blaise Cronin, has carefully investigated the meaning that gets transmitted as authors add references, publishers format them into bibliographies, and indexes collect them to make databases of citations. He makes two important points. First that we should separate the idea of the in text reference and bibliographic list – the things that authors create – from the citation database entry – the line in a database created by an index provider. His second point is that, once we understand the distinction between these objects we see clearly how the meaning behind the act of the authors is systematically – and necessarily – stripped out by the process. While we theorists may argue about the extent to which authors are seeking to assign credit in the act of referencing, all of that meaning has to be stripped out if we want citation database entries to be objects that we can count. As an aside the question of whether we should count them, let alone how, does not have an obvious answer.

It can seem like the research enterprise is changing at a bewildering rate. And the attraction of a gold standard, of whatever type, is stability. A constant point of reference, even one that may be a historical accident, has a definite appeal. But that stability is limited and it comes at a price. The Gold Standard helped keep economies stable when the world was a simple and predictable place. But such standards fail catastrophically in two specific cases.

The first failure is when the underlying basis of trade changes, when the places work is done expands or shifts, when new countries come into markets, or when the kinds of value being created changes. Under these circumstances the basis of exchange changes and a gold standard can’t keep up. Similar to the globalisation of markets and value chains, the global expansion of research and the changing nature of its application and outputs with the advent of the web puts any fixed standard of value under pressure.

A second form of crisis is a gold rush. Under normal circumstances a gold standard is supposed to constrain inflation. But when new reserves are discovered and mined hyperinflation can follow. The continued exponential expansion of scholarly publishing has lead to year on year inflation of citation database derived indicators. Actual work and value becomes devalued if we continue to cling to the idea of a citation as a constant gold standard against which to compare ourselves.

The idea of a gold standard is ambiguous to start with. In practice citation data-based indicators are just one measure amongst many, neither the best available – whatever that might mean – nor an incontrovertible standard against which to compare every other possible measure. What emerges more than anything else from the work of the past few years on responsible metrics and indicators is the need to evaluate research work in its context.

There is no, and never has been, a “gold standard”. And even if there were, the economics suggests that it would be well past time to abandon it.

In this final post about the IDRC data sharing pilot project I want to close the story that started with an epic rant a few months ago. To recap, I had data from the project that I wanted to deposit in Zenodo. Ideally I would have found an example of doing this well, organised my data files in a similar way, zipped up a set of directories with a structured manifest or catalogue in a recognised format and job done. It turned out to not be so easy.

In particular there were two key gaps. The first was my inability to find specific guidance on how to organise the kinds of data that I had. These were, amongst other things, interview recordings and transcripts. You’d think that it would be common enough, but I could not find any guidance on standard metadata schema or the best way to organise the data. That is not the say that it isn’t out there. There is likely some very good guidance, and certainly expertise, but I struggled to find it. It was not very discoverable, and as I said at the time, if I can’t find it then its highly unlikely that the ‘average’ non-computational researcher generating small scale data sets will be able to. In the end the best advice I got on organizing the package was ‘organise it in the way that makes most sense for the kinds of re-use you’d expect’, so I did.

The second problem was how to create a manifest or catalogue of the data in a recognised format. First there was still the question of format, but more importantly there is a real lack of tools to generate well formed and validated metadata. You’d think there would be a web-based system to help you generate some basic dublin core or other standard metadata but the only things I could find were windows only and/or commercial. Both of which are currently non-starters for me. This really illustrates a big gap in RDM thinking. Big projects manage data which gets indexed, collected, and catalogued into specialist systems. Single data objects get reasonable support with generic data repositories with a form based system that lets you fill in key metadata. But for any real project, you’re going to end up with a package of data with complex internal relationships. You need a packager.

At this point, after I’d posted my rant, Petie Sefton from the University of Technology Sydney got in touch. Petie has worked for many years on practical systems that help the average scholar manage their stuff. His work mixes the best of pragmatic approaches and high standards of consistency. And he’d been working on a tool that did exactly what I needed. Called Calcyte, it is an approach to data packaging that is designed to work for the average researcher and produce an output that meets high standards. Petie has also written about how we used the tool and presented this at eResearch Australasia.

What does Calcyte do?

Calcyte is ‘is experimental early- stage open source software’ that generates ‘static-repository[s] for human-scale data collections’. We worked with Petie running the software and me filling in the info, although I could potentially have run it myself. What this looked like from a user perspective was very simple. First Petie and I shared a folder (we used Google Drive, sue me) where we could work together. After the first run of the tool each of the subdirectories in my folder had a CATALOG file, which was an Excel file (yeah, sue us again…).

Opening up the Excel file showed a partially filled in spreadsheet with information about the files in that directory. Which I then filled out.

Running the tool again generated a top level CATALOG.html file, which is a human-readable guide to all the files in the package. The information and relationships is all hoovered up from the entries in the various spreadsheets.

…but more than that. The human-readable file has relationships both about the files, like for instance copyright assignments, people and external identifiers where available. Here for instance, we’ve done something which is generally close to impossible, noting that I’m the creator of a file, with links to my ORCID, but also that the Copyright belongs to someone else (in this case the IDRC) while also providing file-level licensing information.

But it’s not just information about the files but also between files, like for instance that one is a translation of another…

All of this is also provided in machine-readable form as JSON-LD and it is all basically schema.org metadata. We rinsed and repeated several times to refine what we were doing, how to represent things, until we were happy. Things that were not obvious included what the canonical ID for the project was (we settled on the DOI for the formally published grant proposal), how to represent the role of interviewee as opposed to interviewer, and a whole series of other niggly things, that mostly related to questions of granularity, or particular choices of how to represent relationships (versions of objects that were not updates or mere format shifts).

Finally all of this was packaged up and sent of to Zenodo. It is a shame that Zenodo can’t (at the moment) surface the CATALOG.html in the root directly as a preview. If it did and this ‘DataCrate’ approach caught on then it is easy to imagine the generation of a set of tools that could help. I can imagine (in fact I’ve already started doing this) writing out simple CSV files in computational projects that record provenance that could then be processed to support the production of the CATALOG files. Working in a shared folder with Excel files actually worked extremely well, the only real issue being Excel isn’t great for text data entry, and its habit of screwing up character encodings. It is easy to imagine an alternative web-based or browser-based approach to that walks you through a cycle of data entry until you’re happy. Adding on a little conventional configuration to guide a user towards standard ways of organising file sets could also be helpful. And actually doing this as you went with a project could help not just to organise your data for sharing, but also provide much better metadata records as you were going.

The happy medium

When I was getting fed up, my complaint was really that online support seemed to be either highly technical (and often inconsistent) advice on specific schema or so general as to be useless. What was missing was what PT calls ‘human-scale’ systems for when you’ve got too many objects to be uploading them all one by one, but you have too few, or your objects are too specialist for there to be systems to support you to do that. For all that we talk about building systems and requiring data sharing, this remains a gap which is very poorly supported, and as a result we are getting less data sharing than would be ideal, and what data sharing there is is less useful than it could be. The challenge of the long tail of research data production remains a big one.

In the report we note how crucial it is that there be support where researchers need it, if the goal of changing culture is to be achieved. Researchers struggle both with good data management within projects and with packaging data up in a sensible fashion when subject to data sharing mandates at formal publication. Because we struggle, the data that is shared is not so useful, so the benefits to the researcher are less likely. Add this to a certain level of antagonism caused by the requirement for DMP provision at grant submission, and a virtually complete lack of follow-up on the part of funders, and the end result is not ideal. Of course, doing the practical work of filling these gaps with good (enough) tools is neither sexy nor easily fundable. Implementation is boring, as is maintenance, but its what’s required if we want to move towards a culture of data sharing and re-use. The kind of work that Petie and his group do needs more support.

Images in this post are (c) Peter (Petie) Sefton, taken from http://ptsefton.com/2017/10/19/datacrate.htm

Open Access week is a fitting time to be finalising a project on Open Data. About two years ago I started working with the Canadian development funder, the International Development Research Center, to look at the implementation of Open Data policy. This week the final report for that project is being published.

Everyone, it seems agrees that opening up research data is a good thing, at least in the abstract. While there are lots of good reasons for not making data open in specific cases, its hard to make a case against the idea that, in general, it would be good for more of the outputs of research to be more available. The traditional way to achieve this, at least for funded research projects is through funder policies requiring data management and data sharing.

But while there’s a steadily increasing number of funders requiring data management plans and data sharing there is less work asking whether they work. There’s a lot less work asking what might be an even harder question, what does it actually mean for them to work? For the past two years I’ve been working with the IDRC, a Canadian funder of development research to probe that question. In particular I’ve been looking at how policy, and support of policy can support the kind of cultural change that leads to good data management and data sharing being the default for researchers.

The reason the project focused this way was due to a striking early finding. My first task was to interview a set of international experts on research data management and sharing for their views on what worked and what did not. What emerged was a sharp divide between those who thought that Data Management Plans were the best way to get researchers thinking about RDM, and those who were worried that the requirement for DMPs was leading to a tick box “compliance culture”. There was a concern that this might actually be damaging the process of culture change. I wrote about this concern in the review written for the project, published in RIO Journal (where all the formal documentary outputs for the project can be found).

The main part of the project was focussed on seven case studies of IDRC-funded projects. We worked with the projects to examine the data that they were generating, and then to develop a Data Management Plan. Then we followed the challenges and issues that arose as the projects implemented those plans. IDRC funds development research done in developing countries. We had projects based in Vietnam, Niger, South Africa, Columbia Egypt and Brazil working on projects in areas from copyright to climate change, health issues to sexual harassment, ecology to economics.

When we looked at the projects we found any of the challenges you might expect for data sharing in places away traditional centers of research prestige. Lack of good network connectivity, limited capacity and resources. We saw serious issues arising out of the English-centric nature of not just research communication, but systems and infrastructures. And we saw that the funding of infrastructures was inconsistent, periodic and unstable.

But what we also saw were challenges that arose at the funder level. These were not restricted to IDRC, and not limited to development funding or even data sharing, but general issues of policy design and implementation for research funders. Funders work with many projects, and those projects have different contexts and issues. Often the relationship between funder and project is mediated through a single person, the Program Officer. This single person then has to be the conduit for all policy communication, implementation, and monitoring.

Its hardly news that the hard working people within funders have a lot on their plate. What I certainly hadn’t seen before was how this limited progress towards culture change amongst researchers. Within our pilot project we had a group of highly engaged program officers that were learning about data management as they went. New policies required them to become instant experts able to advise, and where necessary push, funded projects towards new practice. There are a lot of policy initiatives within funders at the moment, and the load means that not all of them get the attention they need.

This turns out to be a problem because of the message it can send. Culture change really involves change within both researchers, and the funder. Each on their own can change practice, but embedding that practice so it becomes a culture, of “that’s just how things are done” requires mutual reinforcement. It is not enough to say something is important, you have to show you care about it. Many policies say nice things, but without the backup, and the infrastructure to support it, the message that can come across is “we talk about this, but we don’t really care”.

I had a very engaged and sympathetic group of program officers and projects, but even for this group, data wasn’t always the highest priority. We had to plan to regularly reinforce, and provide support for, the message that data was important over the course of the pilot project. Across a whole funder the risk of balls being dropped is very high. This raises issues for policy design and implementation. In the final report, just published at RIO the following recommendations are made.

1. Policies serve at least two distinct functions. They serve as a) signals that a specific issue, in this case improvements in data management and sharing, are an issue that the policy maker takes seriously and b) as a means of creating interventions in the behaviour of those subject to the policy. These two aspects of policy intent can be in tension with each other. Policies in general have been designed with the intent of requiring and driving behaviour change. The aspect of signalling, and its role in supporting culture change has been taken less seriously. In the initial review it was noted that the aspirational signalling aspects of some policies were at odds with the content of their interventions.
2. The signalling function of a policy is important for culture change. Researchers and funder staff take note of the direction of travel. Articulation of aspirations, particularly when they align with existing sympathies and narratives is important. The details of policy design seem less important and can be antagonistic to cultural change where the aspirations are not matched by capacities provided by the policy maker.
3. Internal audiences are at least as important as external. Policy needs to be designed with a view as to how it can effect the desired cultural change within a funder as much as among researcher communities. How does the policy articulate the importance of its aspirational goals, does it give permission for individuals to act, how does it help create a strong culture that aligns with the policy goals? Perhaps most importantly does it provide the necessary internal levers to ensure that those charged with implementing policy can access the necessary resources, infrastructure and expertise?
4. The primary failure mode for policy is overreach. The most common issue with policy implementation is where demands are made that cannot be met or evaluated. Generally the concern is whether researchers have the capacities, resources or expertise to deliver on policy requirements. However the risks can be substantially higher where the shortfall is at the funder. The core of culture change is a sense that the aspirations underpinning the change are both important and shared amongst stakeholders. Where a grantee falls short, allowances can be made in a way that reinforces the importance of the direction of travel to the grantee. However when a funder fails to provide the support and infrastructures – including Program Officer time, skills and expertise – necessary for policy implementation, and most particularly when implementation is not evaluated, the message is sent that the issue is not in fact important. This in turn leads to instrumental behaviour and in turn a lack of, or even negative, cultural change.
5. The interventions required by policy must be properly resourced, continuous, and self-reinforcing. The second aspect of policy remains important. Interventions at the level of requirements and practice have a role to play, both as opportunities to engage – and therefore transmit culture – and as a means of driving best practice and therefore achieving successes for grantees from their behaviour change. For this goal to be achieved the interventions need to be embedded in an ongoing and reinforced narrative that engages both researchers and funder staff.

This being Open Access week it seems worth thinking about how we design policy, not just within funders but more generally so as to effect culture change, not just change behavior. What might we do to better design both policies and implementation to build a greater sense of a collective direction of travel, and how do we ensure that travel is monitored in a way that is sustainable and builds momentum rather than becoming a tick-box exercise?

I’m not quite sure exactly what was the reason but there was a recent flare-up of the good old “how much does it cost to publish a scholarly article” discussion recently. Partly driven by the Guardian article from last month on the history of 20th century scholarly publishing. But really this conversation just rumbles along with a regular flare up when someone does a new calculation that comes to a new number, or some assertion is made about “value” by some stakeholder.

Probably the most sensible thing said in the most recent flare up in my view was this from Euan Adie:

There isn't a "true cost" of publishing an article independent of context. More things with no true costs: bananas, newspapers, tweets

— Euan Adie (@Stew) July 7, 2017

There is a real problem with assigning a “true” cost. There are many reasons for this. One is the obvious, but usually most contentious, question of what to include. The second issue is the question of whether a single cost “per-article” even makes sense. I think it doesn’t because the costs of managing articles vary wildly even within a single journal so the mean is not a very useful guide. On top of this is the question of whether the marginal costs per article are actually the major component of rising costs. One upon a time that was probably true. I think its becoming less true with new platforms and systems. The costs of production, type-setting and format shifting for the web, should continue to fall with automation. The costs of running a platform tend to scale more with availability requirements and user-base than with the scale of content. And if you’re smart as a publisher you can (and should be) reducing the costs of managing peer review per article by building an internal infrastructure that automates as much as possible and manages the data resources as a whole. This shifts costs to the platform and away from each article, meaning that the benefits of scale are more fully realised. Dig beyond the surface outrage at Elsevier’s patent on waterfall peer review processes and you’ll see this kind of thinking at work.

Put those two things together and you reach an important realization. Driving per-article costs to platform costs makes good business sense, and that as you do this the long tail of articles that are expensive to handle will dominate the per-article costs. This leads to a further important insight. It will be an increasingly important part of a profitable journal’s strategy will be to prevent submission of high cost articles. Add back in the additional point that there is no agreement on a set of core services that publishers provide and you see another important part of the strategy. A good way to minimise both per-article and platform costs is to minimise the services provided, and seek to ensure those services handle the “right kind of articles” as efficiently as possible.

Pushing costs upstream

Author Selectivity

Nature, the journal has one of the best business models on the planet and, although figures are not publicly available, is wildly profitable. The real secret to its success is two-fold. Firstly, that a highly trained set of professional editors are efficient at spotting things that “don’t fit”. There is plenty of criticism as to whether they’re good at spotting articles that are “interesting” or “important” but there is no question that they are effective at selecting a subset of articles that are worth the publishing company investing in. The second contributor is author selectivity. Author’s do the work for the journal by selecting only their “best” stuff to send. Some people of course, think that all their work is better than everyone elses’s but they in turn are usually that arrogant because they have the kind of name that gets them past the triage process anyway. The Matthew effect works for citations as well.

It may seem counter intuitive but the early success of PLOS ONE was built on a similar kind of selectivity. PLOS ONE launched with a splash that meant many people were aware of it, but the idea of “refereeing for soundness” was untested, and perhaps more importantly not widely understood. And of course it was not indexed and had no Journal Impact Factor. The subset of authors submitting were therefore those who a) had funds to pay an APC b) could afford to take a punt on a new, unproven journal and c) had a paper that they were confident was “sound”. Systematically this group, choosing articles there were sure were “sound” would select good articles. When ONE was indexed it debuted with a quite high JIF (yes, I’m using a JIF, but for the purpose it is intended for, describing a journal) which immediately began to slide as the selectivity of submission decreased. By definition, the selectivity of the peer review process didn’t (or at least shouldn’t) change.

Anecdotally my own submissions follow exactly this path. I’ve never had a research paper sent to be refereed at Nature, every one was rejected without review. Each one of those articles went on to get a lot of citations – several are considered important in their field (although no-one ever really got the thermal equivalence thing, shame really). Similarly the first article my group sent to PLOS ONE (after rejection at several other journals) hasn’t done too badly either (as an aside, you can get to the articles via Scholar, I don’t want to artificially inflate altmetrics by linking directly). As an aside there’s an irony here. One of the ways to maximise the “quality” of articles a journal gets is to avoid being indexed, but have a profile amongs traditional academics in North America and Europe, and publish fast. That way you will systematically get submissions people think are good, but have been rejected by other journals. These are systematically likely to include important articles unrecognised by more conventional journals.

If you follow this logic through then having systems to rapidly reject papers that are likely to be problematic makes sense. Some journals have internal (but obviously secret) systems that reject everything from specific countries (one large publisher had an automatic rejection for everything from Iran at one point) or that fit specific patterns (GWAS studies that link single, or small numbers, of gene up or down regulation to a specific stimulus). Many editorial processes apply various forms of conscious or unconscious bias to “positive” effect here, rejecting articles from unknown institutions.

Technical measures

It isn’t just editorial processes that push costs upstream. It’s also possible to get authors to do extra work by technical means. The most obvious of these is journal submission systems. Once upon a time all the metadata for an article would have been extracted by the publisher or an indexer from the manuscript itself. Submissions involved some number of hard copies and a cover letter being posted. Today the many hours of filling in forms that are typical of submission systems places that work onto the authors’ shoulders.

Similarly the many, many efforts to get authors to use publisher-supported authoring tools seek the same goal. If successful, which they are generally not, they would push work out of the publisher workflow upstream to authors. One of the problems that has arisen with this is that authoring technologies keep moving faster than publishing ones. In some cases there are journals that successfully use the outputs of LaTex directly but these are the exceptions.

The best example remains that of the IUCr journals that essentially provide web based forms to upload standard data formats and small chunks of text into predefined sections. By obtaining structured data from the authors in a standardised format the publishing costs are reduced massively. This is a near ideal case where many reports that are structurally very similar can be supported by a platform. Doing this for articles in general is clearly much harder and there are no good examples of this that I’m aware of for articles in other fields (some fields have side-stepped this for particular data types but that’s different in subtle but important ways).

Obviously there are limits here. It’s not feasible to simply impose new technical requirements and expect all authors to simply acquiesce. And as noted previously in this quasi-series there are challenges relating to the control that submission system vendors hold over publisher workflows. Nonetheless this is an area where publisher costs can be substantially reduced at least in theory.

Pushing costs (and risks) downstream

If one solution to reducing costs is pushing the upstream, then another is clearly pushing them downstream. This can mean a range of things – costs to authors, funders, institutions, or “the system”. As these are often not obviously direct cash costs they can be diffuse. These are the costs of the publisher (or really of any actor in the publishing workflow) not doing something.

An extreme example of these is so-called “predatory” publishers that do not do peer review. To simplify the example we can put aside the question of whether publishers “add value” through peer review and treat this simply as a question of validation and certification. In a world where we assume that being published in “a journal” means an article has been peer reviewed, what (and where) are the costs of that not happening?

Eve and Priego (2017) have dissected the question of who is harmed in a recent article. In particular they note that the harm, largely in the form of risk, falls largely on those relying on “publication” in “a journal” as a quality indicator or a certification of peer review. There is an accumulating risk of being exploited by unscrupulous authors for systems and institutions that rely on such markers. The cost may materialise either as wastage when a poor decision is made (to hire someone or to fund a project) or more directly as the cost of replicating the validation process that was supposed to occur. Whatever you may have thought of Beall’s list, and I really didn’t think much of it, it was a product of donated labour. When it folded, many of the evaluation systems that had come to depend on it discovered those costs and had to find a way to cover them. What is worth noting is that, the more institutions rely on such markers (including eponymous lists) as reliable proxies, the greater the risk and the greater the cost when problems arise.

This is a toy example, albeit one that really did play out in practice. The risks also run the other way as Eve and Priego note, the loss of perfectly good work that is published in an otherwise unreliable venue. Similar kinds of cost and loss do play out across the publishing system where venue is relied upon as a proxy; at the extreme end the costs (financial and in terms of mortality and morbidity) of cases such as the Wakefield paper, assumed to be reliable because it was published in the lancet, and on the other of highly promising young researchers neglected (and the investment in their careers wasted) because they happen not to have the required paper in the right journals, or book with the right publisher.

There are many examples where publishers have to make a choice to internalise costs or push the risk downstream. Checking for image manipulation is a good example. This remains costly, and in most cases prohibitively so, to apply to every submitted image so spot checking is common, or restricting checks to accepted articles. But that passes the cost downstream, either to the next journal in the submission chain, or to the community of readers. Readers may even be unable to check images because of format shifting and resolution changes in the production process to reduce publisher costs. Unfortunately most of these decisions are done quietly without any publicity. It remains almost impossible to determine across journals exactly what checks are carried out, with what frequency, and at which stage.

What does this mean?

First and foremost, don’t take anyone who says “the cost of an article is [or should be] X” seriously unless they are very, very, clear on exactly what parts of the process of publishing they are talking about and the context in which those parts are taking place. Second, we need to get much more explicit about what services are being provided. There have been a range of efforts to try and improve on this, but they’ve not really taken off, at least in part because they’re not really in the interests of the big publishers, who benefit the most from playing the games I’ve described here. As I’ve argued elsewhere we need a much better understanding of how shifting the costs from the publisher balance sheet, to the academic worker, or institutional infrastructure, play out. How do cash costs of managing a strong quality assurance process within a publisher scale when they become the non-cash costs of an academic editor, or referee’s time? When they become the time wasted by researchers on following up on fatally flawed research?

All of the above relates to costs, and none of the above relates to price. We also need a better understanding of the political economics of pricing. Particularly we need to understand where we are creating a luxury goods market where price (either in work to get through the selection process or in money) drives the perception of brand, which reinforces the perceived value of the investment, and therefore leads to an increase in price. Be very wary of claims that publishing is a “value driven market”. This is both a way of avoiding a discussion of internal costs, which in some cases are outrageous, and of avoiding the discussion about precisely which services are being provided. It’s also a captive market. Another value-based market is plane flights out of Florida this week…

Above all, we need a much more sophisticated understanding and far better models of how costs are being distributed across the system in cash and non-cash costs, in labour and in capital. Overall we can point at the direction of travel, and the problems with existing pricing and cost structures. But its hard to make usable predictions or good design decisions without a lot more data. Getting that data out of the system as it exists at the moment will be hard, but the direction of travel is positive. There’s nothing wrong with publishers making a surplus, but any conversation about what kind of surplus is “appropriate” needs to based on a much deeper understanding of all the costs in the system.