[tagh]

I'm one of the researchers whose had a paper removed. Confusingly, it was the 'accepted manuscript' version, which Elsevier explicitly states you may share [0]. It's very obviously watermarked too, by their own editorial management system. It's very frustrating that you can do everything right by the publisher and still lose like this.

[0] https://www.elsevier.com/about/policies/sharing

*Edit: typo

[leephillips]

The policy link lists, in some detail, how and where you may share your accepted manuscript—for example, on arxiv. I would not interpret their criteria to include Researchgate.

[tagh]

I think you're right. I've double checked their FAQ, which states that you can upload your preprint to Researchgate, but then they only explicitly mention arxiv for swapping the preprint with the accepted manuscript. I guess I'll just have to swap over to arxiv.

For the benefit of those who haven't submitted a manuscript before: the accepted version is just the preprint with improvements from the peer review process - and no "other publisher value-added contributions such as copy-editing, formatting, technical enhancements and (if relevant) pagination". Assuming the associate editor is a volunteer (true for the journal in question), Elsevier's only in-kind contribution at this point is their awful submission system, and the Chief editor's time in accepting the associate editor's decision... although the chief editor might also be a volunteer.

[leephillips]

And since the formatting is probably already done by the author with TeX, and (in my experience) the copy editing done by scientific journals usually makes the paper worse, there's not much “value-add” here.