Thanks for sharing this very interesting and informative piece of work! I very much enjoyed the read and several questions and concerns that I had about preprints (from an author perspective) were answered. I have some comments that I hope can be useful:
lines 31-69, on the "Landscape of preprint servers": Could there be more examples worth mentioning here than PeerJ preprints and BioRxiv? The journal Biogeosciences for example uses a model where submitted papers are published in "Biogeosciences Discussions" while they are in review with Biogeosciences, is this not also a form of preprint?
lines 58-59: "It is only possible to transfer a PeerJ Preprint for submission to PeerJ." This could be interpreted as if you can only publish in PeerJ after submitting a preprint to PeerJ Preprint, which is not what is meant I think.
lines 213-216: it was a bit unclear to me initially what this example served to illustrate, perhaps due to the following sentence. Were some of the image manipulated works retracted due to the preprint, before the final publication?
lines 236-237: relative novelty?
line 254: That preprints can be revised was news to me, this seems like it could be a major difference to peer-reviewed work where you are usually not even allowed to correct typos once the accepted article has gone online. Perhaps expand on the implications of this?
line 292-314: This touches on one of my major open questions about preprints currently. The examples online two different strategies on how to coordinate preprint submission with traditional journal submission: Either you submit preprint first, wait for comments that can be integrated into new version which is then submitted to journal (slow). Alternatively you submit preprint and journal version at the same time and use both preprint comments and reviewer comments to improve the MS for a future submission. I would tend to do the latter due to time constraints, but I am concerned that incorporating preprint comments parallell to reviewer comments could confuse (or worse: offend) journal reviewers and editors that are not familiar with the preprint model. Imagine you want to make changes to MS not called for by reviewers, will the editor feel compelled to send out the MS for another round of review thus slowing down the publication process? I guess this should not be an issue if changes are justified well in cover letter (perhaps linking to preprint comments?) and as long as editor is sensible and competent. Nevertheless the issue has crossed my mind when considering submitting preprints parallell to journal submissions.
line 329: "mSpehereDirect." what is up with the "." here?
Figure 1a: For a second I was confused by the grey bars until I realised they were indicating the border between years. Will perhaps be more visually clear when year 2017 is included? (very minor issue).
I'm looking forward to see future versions of this manuscript, preprint or published in journal!