It sounds like the problem is more or less summarized as "No one knows how to evaluate work, so loud people get promoted." Not biasing towards loud people is d

It sounds like the problem is more or less summarized as "No one knows how to evaluate work, so loud people get promoted." Not biasing towards loud people is difficult, but at least at Google, the process is pretty harshly designed to avoid it. The criteria for promotion is clear. The people deciding on it are considering many different promotions at once, so they have a decent sample to look at. The people with incentive to promote don't get to make the decision or they play only a small role in it. There is a strong demand to see evidence that the person's work fits the job ladder description that they are trying to promote to, etc. It's a lot of overhead and everyone hates it, but it seems to be pretty robust at avoiding arbitrary promotion.