[munfred]

The author of this post raises valid points about some avenues for publishing scientific results. Journals have high barriers, rigid formats, paywall, and sometimes exclusionary reviewers. Twitter and blogs are fringe because they are not formal, but increasingly where the conversation is happening. He advocates for a model like Substack for scientific publishing. Which I don't see how it is different from blogs.

However to me the real problem is that he completely ignores the existence of preprint servers. This started with the arXiv in the 1990s for physics, math and computer science, but finally the model has begun to catch up with other fields. In biology the bioRxiv is well established and I'd say more than half of the papers I see do have a preprint deposited there. For medicine, which has a notoriously protective culture, the medRxiv was launched last year and got a gigantic boost due to the pandemic.

Other fields also have their own preferred preprint repositories. To me preprints solve the gatekeeping and cost barriers very well. I only wish they were a little more tied to comments from the readers (in depth comments of the kind you get during peer review, not the knee jerk ones we're used to seeing on the internet).

I say this because there is almost and "adversarial" relationship between the readers, who just want the bottom line in two sentences, and the authors would like their work to be perceived as a big important and novel contribution. Because of this often papers are written with more flourish language than actually necessary to make a point, and so it is really a hard to parse articles and tell good from bad. Often the "summaries" that peer reviewers wrote before making the comments are better than the official abstract of an article, but readers don't get to see and benefit from them. Each person that reads an article has to redo all that work of judging it for themselves, and with the deluge of articles we have it is simply impossible to keep up.

Initiatives such as openreviews.net are already a big step in the right direction. I think in the next 10-20 years this is where the bulk of scientific publishing is headed. A mix of preprint server coupled with curated community reviews that makes papers much easier to gauge and also incentivizes authors to write better and not bullshit, or risk getting wrecked publicly with the public reviews.