> The platform splits pre-publish peer review from post-publish refereeing. Pre-publish review then becomes completely about helping authors polish their work and decide if their articles are ready to publish. Refereeing happens post-publish, and in a way which is easily understandable to the lay reader, helping the general public sort solid studies from shakey ones.

The author doesn't seem to understand the "referee" role of (associate) editors and reviewers. The core idea is about gatekeeping of a journal title's self-set publication limit in favour of articles that represent a substantive advancement of a field or concept.

However this itself is a problem for many fields of research, since it's a core factor underlying the failure to engage in replication studies and to check if results are valid. Moreover this would tend to encourage researchers with small but useful results to shelve them and NOT publish, since it will create a negative feedback mechanism for such contributions:

> Up votes increase the paper’s score and grant the authors reputation in the tagged fields. Down votes decrease the paper’s score and the author’s reputation in the tagged fields.

Part of this problem is because, like any site which uses an upvote or downvote function, the button will be used to express things other than the intended meaning. "Does not contribute" quickly becomes "I don't like X" (even if it is methodologically valid).