Because if that is what you're telling me, why should I believe that the current Witness collection who were supposed to be responsible for making sure the hard forts could maintain basic blockchain functionality are capable of doing that?

We do the best we can with what we have. which is a very limited test suite in the code base itself as well as a testnet which runs mostly independently of the main net and where it is a tremendous amount of work required to try to reproduce main net conditions (and even that can never be done perfectly).

In cases of small patches we can also do independent code reviews but that isn't feasible for major release on a large codebase, at least not given the current size of the economy (perhaps with more value to work with we could pay for comprehensive professional reviews, but currently we can not). Even with the new SPS, the resources available for tasks will be much smaller than the potential number of things those resources could be spent on. Again, we do the best we can with what we have.

If we wanted to wait until everything were perfect we would never get anywhere. Progress is being made and improvements will continue to be made for future releases as noted in the post. It is an ongoing incremental process.

Despite the number of times that you say so, upvote is not a signal of value and never has been. It's a signal of what some user or by believes other users and bots will vote on.

This is indeed a signal of consensus that the content is value-adding because the one thing that all stakeholders have in common is wanting their stake to increase in value. If you bet on content getting upvoted by other stakeholders when that content does not contribute significant value, you are much more likely to be wrong than when you bet on content that does. Sure, there are other factors at play and individual cases can certainly vary, but that is the core of it.

HF19/HF20 did not accomplish this very well. HF21 remains an experiment and we'll see if accomplishes it at least somewhat better. If not, well Steem itself is also an experiment in progress and we can iterate.

And since all of the above is scaled by SP, what you really mean is "votes should only deliver rewards if the content is deemed valuable by a whale."

No, that is not what it means and there are plenty of rewards paid out with NO whale votes. Yes, those who own a larger share of the system and with more value at stake have a larger say in how their resources are used and distributed, as with any business (although it isn't so common with most businesses that every user can earn or buy a larger stake in the platform). (Again, I'm referring here to how it is supposed to work and not how HF19/HF20 was working, which was indeed very poorly.)