Alexandra Freeman, in an August Scholarly Kitchen interview about her new publishing platform Octopus:

Of course people can publish elsewhere as well. I suspect in the short term Octopus will act a bit like a preprint server in that respect. Authors will hopefully want to publish quickly in Octopus to ensure they get the credit for their work as soon as possible, protecting their “priority” over it, but then write it up for a traditional journal at some point. In the longer term I think that second stage will modify, so that journals will not be taking primary research papers and organizing peer review, but instead be commissioning reviews, syntheses, editorials, and short “news” articles announcing findings — so authors will instead be pitching those once they have published in Octopus.

There's lots to digest about Octopus, including its modular model. What's most interesting about Freeman's interview, though, is her vision for the future of journals. Plainly influenced by her background in media (and science journalism in particular), she sees journals moving toward public- or professional-facing summary pieces, with the soup-to-nuts details for fellow specialists hosted on Octopus or its future peers:

It struck me that journals are trying to fulfill two roles at once in the current system. On the one hand they are disseminating important findings to their readers, particularly to professionals who may be able to implement those findings in practice. On the other hand they have the somewhat thankless task of being the primary research record, where researchers publish all their work in all detail to a minority readership. These two roles pull in opposite directions: an article that gives full details, all the twists and turns and blind alleys of the real research process, does not make a nice, easily-digested narrative for those who mainly want to know the findings and implications; an article that supplies the ‘edited highlights’ does not give the full details that other researchers need to learn from. Journal articles are best suited to the dissemination role — the “news and views” and editorialized, narrative-driven pieces.

It's an interesting vision for the division of publishing labor—readable digests, separated out from the weedier research record. Freeman just got £650,000 from Research England to build out her end: Octopus is slated to launch in spring 2022.