"The system of peer review" in this context is the system where a scholarly journal editor sends your submitted paper out to other experts in the field to decid

"The system of peer review" in this context is the system where a scholarly journal editor sends your submitted paper out to other experts in the field to decide whether or not to publish it. This system came into use in the middle of the 20th century, and Einstein was famously outraged by it. It does not refer generically to every time someone reads or discusses a paper or replicates an experiment. I don't think survivorship bias is particularly relevant for three reasons. First, both papers from 50 years ago and papers from 150 years ago are already heavily filtered. Second, if you look at journal issues from 150 years ago, you will find forgotten papers in the same issue with the foundational ones, and the quality is still much better than today's forgotten papers. Third, what I'm really concerned about is not that bad papers about bad research are being published, but that good papers about good research that would have been done are not, echoing Higgs's remark about how he couldn't have done his work on the Higgs boson today because he wouldn't get tenure. Or, read Freeman Dyson's autobiography, and contrast the years when he was working on nuclear energy to the rather uninspired following 50 years, because, as he put it, it stopped being fun.