Hard fork 21 actually failed this test for quite a while

You mean 20?

Again, it's not personal. It's purely professional

Taken as such.

By all means, vote against all of the current witnesses (indeed especially the witnesses at the time of HF20) but it is my sincere belief that other witnesses could do no better under the circumstances, and could certainly do worse. It is not even clear to me that you know what witnesses do. But that may not matter to you and wanting to seeing someone fall on a sword is a defensible response to bad outcomes.

You literally have three jobs. Make sure the content can continue coming in. Make sure the content can be voted on. And make sure there are new people coming in to help drive the platform. Everything else is secondary at best and way down the line at worst.

I'm afraid I must disagree to a point. Those tasks are important for sure, but as you say the user interaction is still shrinking month after month. What good is that? Looking at the broader picture I do believe that we need to balance making progress in improving the platform with the uninterrupted operation of the platform. Rejecting all forks or rejecting all forks with are not rigorously tested and audited (nearly equivalent in practice at this juncture) would certainly optimize uptime and minimize risks but it would not necessarily optimize for improvement which can ultimately turn around the (lack of) growth trajectory. My view is that this balance takes precedence, even though it isn't always an easy task.