Maybe if the academic world behaved like that, I wouldn't have burned out and fucked off to the IT sector a decade ago.

Kind of adjacent to this, years ago around said burnout I kept floating this idea around my head where I was thinking if there was some way out of this tradition of creating these giant monolithic papers every time -even if your effective research result could be distilled down to a paragraph with some numbers and preliminary handwaving- where you need to pad the whole thing out with a big-ass literature review to keep the citation circlejerk going.

So why not just have "papers" consist of effectively a few paragraphs? The citation tree of how you got there is still relevant, but you can put all of that stuff in what's effectively metadata and not clutter up the whole thing with it.

Have an idea for a lab experiment? Publish the methodology as-is, and link it to whatever other tidbits of knowledge make it relevant. Did the experiment and got results? Publish the data, and link to the experiment. Got some new theory out of the results? Publish the theory and link to the experiment results in the metadata. And so on.

Maybe this would have weird side effects of its own, but I can't help but imagine this would make the whole process so much less painful, and allow also for better organizing of the knowledge produced.