This is simply not true. The world has changed a lot the last 10 years. In many scientific fields, you are actually expected to publish your "preprint" in the open and also your source code.

In what fields are you expected to publish your source code? In the fields I’ve interacted with it was always an “above and beyond” thing that nobody would raise an eye at you not doing.

i was thinking about image processing or machine learning, but i'm sure this is the case in other fields as well. Reproducible research is a "hot topic" in many places, and for good reasons.

ecology, more and more biology, chemistry...

That's not at all the same as the proposal I was replying to.

They seemed to be saying that people should only publish in Sci-Hub, making it a prestigious publication (that would appear in press releases, one assumes) instead of just a mirror.