I keep seeing comments like this on the internet and I don't know where this idea comes from the "peer review" is some kind of double checking and reproducing of the scientific research done.

Peer review is where the journal makes sure the claimed work done is of high enough quality, and that the results are interesting and significant enough to be worthy of publication in that particular journal. If not, the work either needs to be improved or published in a journal that does not have as much "esteem". If anything peer review is the reason that so many fake and fraudulent results are published these days (though I'm not suggesting I know of any better alternative).

The alternative would be a journal/platform/wiki/repository of only reproducible results. Only double, triple, quadruple checked results would be accepted. In physics I believe this is called being above a certain value of sigma.

This should also require a way to prevent collusion.

All other content would be placed in a pending category.

I disagree. The alternative would be simply to make an arxiv like website the only place where papers are published and include a system where other scientists can leave public comment and ratings for each paper. Then everyone can follow scientists that they trust and see what they have to say.

This would never happen because the current system is about protecting academics' jobs, lest anyone discover that a majority of the work they do is so trivial and down right silly that anyone could do it with no training.

Not sure a centralised entity is desirable vs a distributed network.

And platforms for distributed networks do exist (1), (2).

The issue is incentives for switching and reaching critical mass.

(1) https://blog.science-octopus.org/

(2) https://osf.io/

I’m not sure the preprint matters here, what is being stated is a statistic generated by several fully integrated healthcare providers that have access to the data required to measure such factors as rare side effects of a new vaccine.