"Fact" in science needs to be based as much as possible on reproducibility by anyone and as little as possible on the support of those with institutional authority.
Alternatives to peer review are seen everywhere these days: open review and discussion and even the open source software model.
A scientist does some work, writes it up in rough form, puts data files online, and tells some friends/colleagues about it. After some feedback and corrections, the scientist puts the writeup online and notifies discussion forums. Then, anyone interested can look it over and provide feedback of all sorts: good, bad, and ugly.
The scientist then incrementally upgrades the paper, including edits and rewording and donated graphics, etc., along the lines of open source software and pull requests. I'm not claiming that many eyes make all bugs shallow, but if someone points out that your algorithm omits the first line of each data table, or your evidence of warming only measured temperature during the day while comparing to temperatures measured day and night, it's true or not regardless of the credentials of the person making the claim.
The paper doesn't have to be fossilized but can be versioned, because "paper" doesn't mean paper anymore, and your institutional worth (as opposed to scientific worth) can be based on citations of all versions rather than the prestige of the journal.
You might fear that authors will be overwhelmed by mostly low-quality feedback, but most papers are read by so few people that this isn't much of a risk, in my opinion. The biggest risk is deliberate lies, but we already have a reproducibility crisis, and only reproducibility--not the filters of authority--can ultimately solve it.
Alternatives to peer review are seen everywhere these days: open review and discussion and even the open source software model.
A scientist does some work, writes it up in rough form, puts data files online, and tells some friends/colleagues about it. After some feedback and corrections, the scientist puts the writeup online and notifies discussion forums. Then, anyone interested can look it over and provide feedback of all sorts: good, bad, and ugly.
The scientist then incrementally upgrades the paper, including edits and rewording and donated graphics, etc., along the lines of open source software and pull requests. I'm not claiming that many eyes make all bugs shallow, but if someone points out that your algorithm omits the first line of each data table, or your evidence of warming only measured temperature during the day while comparing to temperatures measured day and night, it's true or not regardless of the credentials of the person making the claim.
The paper doesn't have to be fossilized but can be versioned, because "paper" doesn't mean paper anymore, and your institutional worth (as opposed to scientific worth) can be based on citations of all versions rather than the prestige of the journal.
You might fear that authors will be overwhelmed by mostly low-quality feedback, but most papers are read by so few people that this isn't much of a risk, in my opinion. The biggest risk is deliberate lies, but we already have a reproducibility crisis, and only reproducibility--not the filters of authority--can ultimately solve it.