Peter calls out attention to tens of thousands of scientific papers that are possibly (likely? who knows?) invalid. This is not a surprise to anyone who is paying attention. It comes from two very simple principles:
1. Success in a scientific career is determined by how frequently the scientist publishes papers in scientific journals.
2. Scientific journals are interested in novelty - new results that have not been published before. It's pretty rare to see a paper published confirming the results of a prior paper.
In short, there's absolutely no need for results confirming other papers - these will never get published anyway. Since there's no need for confirmation, a scientist can focus all his efforts on novelty. Because novel results sometimes must skirt the edge of what is seemly, we can expect spillover into the unseemly.
So how much spillover do we get? We don't know. And quite frankly, neither does anyone who is engaged in "Science". We've known for a long time that something is very wrong in the state of Science™, with the rate of major advances slowing noticeably. It could very well be that the reason is that there is so much bogus science being done, that distracts young scientists from other more significant areas of study. It may be that with no need to get reproducibility, it's just easier to put out novel garbage than it is to do more significant work. And when you think about which of those two patch will be better for a young scientist's career, it's no surprise that we swim is a sea of scientific crap.
None of this has to be deliberate, or fraudulent. It's just the way that the scientific game is played.
Science that is repeatable is called "Engineering" anyway. Strange how you don't seem to hear about lots of retractions from engineering journals. Science is like the stereotypical aging starlet still trying to get drinks in bars off of her old good looks and fame.