Since this is simple to answer (remove all requirements to publish frequently, and hope that a lot of journals die naturally after that), the real question is: how do we distribute funding to scientists without forcing them to frequently show their work?

I could imagine a world where every scientist (starting from Ph.D student onwards) is evaluated only e.g. on the basis of a biyearly dissertation-style report, which includes all (positive and negative) results, all data/metadata/code/analysis. Rapid communication of interesting results can still happen at conferences and the remaining journals.

But then who reads, reviews and ranks all this work? Who gets positions and funding?