If I peel back the onion further, I find the root of the problem is how to measure someone who is good at science. How should we measure scientific output to determine compensation (not just monetary or tenure, reputation itself can be considered a form of compensation)? That the way we reward scientist means that few are pushing to reform academic publishing is barely a footnote compared to some of the problems produced by our current model. Look at the lack of replication that occurs because replication is far less rewarded than new research even when the new research fails replication (once someone does eventually get around to it). The way to fix this is to find a method to measure good science that works better than the current method, but given that us tech people can't even figure out a decent way to measure proficiency in our own field, the outlook isn't hopeful for our ability to disrupt science.
Yes, at the core of the problem is how we measure good science. If you want to try to crack that nut I absolutely salute you. Improvements here could pay massive positive dividends for society as a whole. Meta[1], which was bought by Chan Zuckerberg[2], is sort of trying to do this (in addition to trying to tackle better discovery of the literature) by using an AI model to try to identify important research trends earlier in the process.
