A preprint counts 440 journals that left free-to-publish open access since 2009

Eighty-eight per cent of the 440 stayed with the same publisher through the switch. Matthias says that rules out the simplest explanation, which is that the titles were sold or acquired by someone with other plans ^[11].

The leavers were established journals. Matthias describes them as "disproportionately older, more established and larger titles, and not struggling young ones", and says that suggests the decisions were not necessarily driven by financial pressure ^[12]. Fifty-seven per cent of the journals in the data set switched after at least nine years of operation ^[13]. "It doesn't look like a survival story," Matthias said. "It seems like these journals and their reputation had become commercially worth converting to a different model" ^[14][15].

The commercial reading has a gap in it too. Commercial publishers accounted for 78% of the departures ^[16], but the journals in the data set did not markedly increase their output after changing model, "as might be expected from a pure profit motive", the authors write ^[17].

The 440 is a numerator. A 2021 analysis estimated as many as 29,000 active diamond OA journals worldwide at the time ^[8]; divide one figure by the other and you get roughly 1.5% ^[2]. That is not a rate. The analysis, a preprint published earlier this year ^[19], had to be assembled from several data sets because no single source records all diamond titles globally ^[9], and the authors say their list is not exhaustive, that they could not calculate the proportion of journals that stuck with the model, and that they did not examine why individual titles moved ^[10].

So the study measures the flips it can see, and it identifies the profile of a journal likely to flip. The health of the model is a separate question, and only the flips have a number attached here ^[10].

Diamond open access took off in the late 2010s partly out of frustration that high article-processing charges were pricing researchers out, especially in lower-income regions ^[18]. The charges these journals introduced span more than 600-fold from the lowest to the highest ^[3], and the professional publishers typically came in at the top end ^[4].