"which arguably corresponds to free peer review" No, it doesn't. They are remarkably different. If I distribute free package X, which Amazon uses, then there's

"which arguably corresponds to free peer review" No, it doesn't. They are remarkably different. If I distribute free package X, which Amazon uses, then there's nothing to prevent anyone else from using X. Both can benefit the same way. If I spend 6 hours to do a peer review on paper X, which is then accepted, then Elsevier benefits financially, because they have a new paper to sell, but no other archive can benefit that way. There is also a gain in prestige to Elsevier from having published a higher quality paper. If the paper is not accepted, then the reviewer gets feedback from my review, but no one else does. (I am supposed to benefit in the future from a favor-in-kind by peer reviewers.) While it's true that much open source software is at no cost, I sell free software, exactly as described at https://www.gnu.org/philosophy/selling.html https://www.gnu.org/philosophy/selling.html . While AWS might purchase said software from me, Elsevier/ACS/etc. will never purchase my peer review.