> For all its faults, peer-review is still the best mechanism to keep science in right track.

Peer review is often effective, but it can't reliably block fraudulent publications like those described in the posted article. Most bad papers are rejected, but the authors can always try again at another journal. Any paper will probably get published somewhere, eventually, even if only in a Hindawi or MDPI journal. The journals aren't accountable to anyone, and as long as they have enough good articles to serve as cover, academics will need to pay for access because citing relevant prior work is obligatory. The publishing system is very weak against fraud.

> people with huge twitter followings [...] would dominate science

Isn't that at its core the same as with scientific journals? People trust these journals to curate science in the same way you suggest twitter would come to curate science if it made the move online.

1. It's already possible to call attention to a paper through twitter, regardless of whether it's published in a journal or not. Paywalls gate-keep the content somewhat and makes sharing easier, but that's a minor side effect of a very broken system.

2. Papers (and involved data) being available on public platforms like github that already have mechanisms for reporting and tracking issues as well as built-in review tools, in githubs case even a separate discussion feature now, would allow for much quicker discussion critizising bad methodology.

3. Working with a VCS like git would automatically make it clear who wrote, edited or removed what.