Reviewing process

Like APS does it:
https://journals.aps.org/OutstandingReferees

Discussion forums Would be interesting for readers, but could restrain authors from being perfectly clear, since the forum would give them a second chance to explain properly during the review process. It could also increase the reviewing time.

Blindness "Blind authors, non-blind reviewers" would seem the best to me (blind authors implies less biais, non-blind reviewers encourages high-quality reviews). Unfortunately this is not compatible with the concept of epijournal (see Loïc Salles' comment). The remaining choices are "Blind reviewer" and "Non-blind reviewer". For quality reasons, I am in favor of the second choice but the reviewers should indicate, before accepting to review, that no relationships (professional, or friendships) prevent them from rejecting the article.
If the article is eventually published, reviews should be available as well as the reviewers names, to promote high-quality reviews.

Non-significant results Only significant results should be published. However, failed replications or unexpected outcomes may very well constitute a significant result.

Deux options possibles:
- profiter de ce "vent frais" pour modifier les pratiques en profondeur
- se contenter d'un changement de portage des pratiques actuelles : des éditeurs classiques aux epi-revues

Je suis favorable à un changement en douceur pour entrainer le maximum de scientifiques possibles ; le single blind étant le standard actuel en mécanique, je propose de le maintenir.

The double blind system seems to me the easiest to put in practice, that guarantees the liberty of honest reviewers to express negative opinions without being affraid to suffer in the future from possible retaliations. Of course that the anonymity of dishonest reviewers may sacrifice a good study, but the role of the editors is precisely to avoid unfair evaluations to prevail... In a double blind system it is always difficult to hide the identity of the authors but there is always an element of doubt about the complete composition of the authorship, which seems to be good.

Above all, the quality and agility of the reviewing process should be of major concern. Only true experts on the subject of a submitted paper should be asked to give an opinion. Upon the reception of a manuscript the decision to whom to send the manuscript for revision should depend on an enlarged group of people an not only on one or two (in some cases obsolete) editors.