Last week marked the annual celebration/marketing event that is Open Access Week, and this year it seemed something of a mixed bag. Open access (OA) is growing into maturity, and has rapidly become integrated into the scholarly publishing landscape over the last fifteen or so years. We have now reached a point where experiments have been in place for a while and results can be analyzed. Early assumptions can now be measured and the move to OA seems to have reached something of a crossroads.

Make no mistake, OA is here to stay, and there is no crisis of confidence, at least as far as the continuing growth in access to the research literature. But the repercussions of the business models and methodologies chosen for OA are beginning to be recognized.

The first red flag during OA Week came on Monday, when Jeffrey Beall announced that he was adding Frontiers Media to his infamous list of predatory publishers. I don’t know quite what to make of Frontiers. They do publish some very reputable journals, yet at the same time, their practices have been widely questioned. Researchers have publicly called out Frontiers’ worrisome peer review practices. Earlier this year, the publisher fired some 31 medical editors who complained that Frontiers’ policies “are designed to maximize the company’s profits, not the quality of papers, and that this could harm patients.” Holtzbrinck, the parent company behind the Nature Publishing Group (now merged to form Springer Nature) invested in Frontiers in 2013, and many public efforts were made to tie the Frontiers brand to that of Nature. In 2014 the two companies officially severed ties and decided, “never to mention again that [Nature Publishing Group] has some kind of involvement in Frontiers.”

This does not paint a pretty picture, and perhaps it exemplifies what many fear about the economic pressures that gold OA brings to bear, the structural incentive to increase the number of papers published (and hence profits).

But have all those aggressive moves from Frontiers paid off?

Frontiers recently published a marketing piece touting their “financial commitment to open access publishing.” While the post is clearly self-serving (I’m not sure what the difference is between their “financial commitment” and what any company faces as far as the costs of doing business), it offers us a rare glimpse into the expenses ledger of a commercial, for-profit publisher.

Looking at the Frontiers blog post, they state expenses of some $20M in 2014, and breakdowns of where that money was spent are offered. Frontiers publishes 54 journals, (decentralized across 55,000 editors, or an astonishing 1,018 editors per journal) and from my calculations, published 11,210 articles in 2014. Frontiers has 4 different author charges, based on the type of article, with levels at $1,900, $875, $250, and free.

Doing the math (and since the expenses include $1.9M for waivers and discounts, I’ll assume these cancel out), it becomes increasingly obvious that if the expense figures are correct, Frontiers ran at a deficit for the year. If as little as a few percent of articles where charged the lower rates (I’ve tried to determine actual percentages of article types but Frontiers search functionality has been “currently not available” for the last week), then it’s hard to get to $20M in APC revenue.

Take this with a grain of salt — financial reports are slippery beasts and there’s much wiggle room in how companies treat things like overheads. But still, this is really worrying. If one of the most driven publishers, one that has taken measures to lower costs and increase revenues that are so extreme that some have declared them unethical, still cannot reach profitability, then where does that leave the rest of us?

Shortly thereafter, PeerJ, which has been widely praised for its innovative membership business model, announced a move back to the pack, adding a standard gold OA APC option for authors. PeerJ does not seem to have reached the levels of scale needed to reap the profits promised by their original model. The announcement suggests that some of the problem lies with authors being unable to spend grant funds on memberships.

I suspect that they also overestimated the importance of cost savings to researchers. In this annual survey, the responses from authors remain consistent — quality and reputation are always among the most important factors in choosing a journal. UK authors strongly prefer to publish in established hybrid journals, particularly those with higher citation impact measurements, which correlates to those with higher APCs.

The PeerJ membership model could offer some savings to authors, but those savings aren’t recognized until a second paper is published. Authors likely think more short term, and want to get the maximum career impact out of this paper, rather than focusing on saving a few hundred dollars on the next paper. Being locked in to publishing in one journal at one level may also have proved unattractive to ambitious authors or those looking to target specific research reports to specific audiences.

The value that researchers place on cost savings will continue to be tested by the new (low) APC. The big question is whether this is a sign that PeerJ‘s venture capital backers are reaching the end of their patience with this experiment.

The APC model in general is increasingly under fire. Cameron Neylon, one of the most thoughtful of OA advocates, recently stated that the APC model may be “completely wrong” and is exploring the question of whether it is sustainable through a series of blog posts. There seems a growing recognition that the goals sought have more to do with the deeply ingrained culture of academia than with publisher practices, and that real change is difficult and slow, and unlikely to be achieved through short term moves meant to disrupt the publishing industry.

One thing that is becoming increasingly obvious is that many have grossly underestimated the costs of high quality scholarly publishing. While some continue to base policy on these unrealistic analyses, real world companies are seeing the actual costs. As stated above, it seems as if Frontiers, despite their extreme measures, is still failing to make ends meet at their current APC price levels. PLOS recently raised their APCs to cover rising costs (though oddly, they raised the price on their one profitable journal, asking PLOS ONE authors to spend even more to subsidize their high end journals which run at a deficit despite a $2,900 APC). eLife managed to gain some benefits of scale in 2014 by more than doubling the number of articles they published in 2013. Even so, they still spent nearly $8,300 per article.

The list of business model problems and unintended economic consequences goes on. The largest commercial publishers seem to be reaping significant profits from OA, likely due to the benefits of much greater scale, further concentrating power (and profits) in the hands of this sector of the industry. Productive institutions are doing their own calculations and realizing that OA would cause them to shoulder a greater financial burden then they already carry. Libraries are struggling with the administrative expense caused by inadequately funded policies. Costly and effort-consuming repositories are increasingly looking incomplete and potentially irrelevant due to networks like ResearchGate and Academia.edu, for-profit, venture capital-driven businesses built to spy on researchers and sell their data to the highest bidder.

And so, this OA Week of 2015 leaves us at a crossroads, or at least a recognition that this is an evolutionary process, not a revolutionary one. Evolution takes a very long time, and occurs through variation, but not every variant confers a selective advantage. The ideological hurdles have, in many ways been cleared, and acceptance of OA continues to grow. But as things have moved from the theoretical to the practical, questions of sustainability and cost-effectiveness are making themselves clear. The dominant methods of this early period are being tested in that crucible, and history may or may not look back on them as dead ends.