You'll be surprised how many scientific papers published in reputable journals are not "properly reviewed". In many cases there simply isn't enough time or knowledge to go through every detail of a paper. One example -- if a paper says go to a github repository to look at our source code, guess how many people actually review the code as part of the peer review process? Very few. There could be serious bugs in the code that would affect the results, or maybe the code is difficult to set up and run, but nobody would notice them for a long time.

It's definitely been my experience that reviews are hit-or-miss. Sometimes you get 1-2 reviews where there are valid points that need to be addressed, sometimes you wait 3 months for the reviews to come back, only to find that reviewers 2 and 3 dropped out and reviewer 1 basically said "looks fine".

Part of the challenge of peer review is that it comes _after_ the paper is written, when the cost of changing anything is quite high. I find much more value in peer review when planning experiments, and do so informally with my peers.