Scientific publishing - where are we going?

I think it's safe to say that anyone involved in scientific publishing will tell you that it's a mess and the trends are worrisome.  This week, this news release/article came out about this preprint which shows a number of the issues.  In brief (not all of this is in the preprint; some is me editorializing):

Figure 1 from this preprint

• The number of scientific papers being published is growing at a rate that looks completely unsustainable.  In my opinion, it's problematic on multiple levels.  There aren't enough reviewers (though that doesn't bother all publishers) and the average paper gets smaller and smaller readership (raising the question of why bother to publish papers that no one reads).  Does it make sense that the number of papers is skyrocketing while the number of PhDs granted is falling?
• Some publishers (especially Frontiers, Hindawi, MDPI) have boosted this by drastically cranking up the number of papers that they publish, through launching specialized journals with "special issues" designed to have super-short review times (assuming that review is even truly part of the process).  Lest you think this is only the provenance of publishers previously accused of being predatory, this week alone I have received five different "special issue" announcements from AIP journals.
• Why do people do this?  To try to game the impact factor calculations.  I've aired my grievances before about why journal impact factor is a lousy metric.  
• Why do people want to inflate impact factors?  Because that's how journals keep score, and some countries put in place big-time incentives tied to impact factor.  A publisher worries that if its journal's impact factor falls below some threshold, then the government of China, for example, will no longer view that journal as important, and then thousands of authors will stop submitting....
• Open access is a complicating factor, with some publishers charging absolutely sky-high charges, while at the same time having very high profit margins.  In the US, at least, those charges can be much larger than what grants will support.
• Over all of this is the concern that massively inflating the amount of scientific literature lowers its quality and undermines the credibility of science in general.  

Coincidentally, this week we hosted Steinn Sigurðsson for a colloquium.  He is now the scientific director of the arxiv, the preprint archive that went from a quick and dirty preprint sharing site in 1991 to an enormously important part of the global scientific enterprise.  In his talk he hit on some wild numbers.  The arxiv is up to around 20,000 papers per month now (!) (in part because new disciplines like quantitative biology are using the arxiv).  Thankfully the arxiv has recently landed some good support.  Their annual operating budget is around $3.5M, and this is an enormous bargain by any measure.  The arxiv is partnering with volunteer developers who are adding some neat functionality.  Unsurprisingly, generative AI is a serious concern, even more so than for the publishing houses.  

It's a transformative time, for sure.  Maybe what we are seeing is analogous to the fluctuations that happen when approaching a 2nd order phase transition, and we are headed for a real change in the way publishing works.  It's hard to see how the current trends can continue unabated.