I might've misunderstood the "flag" somehow, I thought it was intended to mark content that people consider inappropriate, and not provide tangible feedback to the author.

In that sense the flag provides both functionalities: it warns other readers and gives feedback to the author so that they have a chance to fix the post.

What I had in mind was a system where negative votes offset positive ones. That way you can properly review stuff, a bit like in real life.

That's indeed a good way to look at it.

What occurred to me is that having a strong feeling about something that cannot be expressed is not the equivalent of censorship, but a technical limitation.

You have a good point here. This "technical limitation" does not mute you (as you can still write a comment saying: "I think this post is overpaid") but it makes you powerless even though theoretically you have the power. I don't really know how to call this state.

After doing some introspection I think this is what disturbs me the most: we remove part of our freedom assuming it will have bad consequences before this fear was proven to materialize in real life. For me, limiting our freedom is a last resort, not a preemptive action aimed to yield to our fears.

But the biggest paradox here is that currently you are able to do what is feared by those who oppose the downvote option. You are just invited to do collateral damage to reach your goal.

This logic is completely beyond my understanding: even if we assume that lowering somebody's payout is damaging, they prefer you to do more damage (by using the flag), than give you an option to do less damage (by giving you a downvote tool).