Great and thoughtful post.
One insight you bring is that signal to noise is becoming a more significant problem. This is greatly exacerbated by the 400% acceleration in VP decay that HF19 brought, as decreasing our ability to curate to a 1/4 of what it was to a Steemit that sees an order of magnitude more posts than were made when HF19 was first discussed dramatically degrades our ability to curate content from this influx of new accounts.
Regarding account creation costs, I'll point out that, while account holders clearly benefit from gaining an account, Steemit also benefits. Estimates I have seen indicate the value of a new account to be at least $300, and range much higher.
Current political landscapes have created an expectation that new accounts not require financial investment. Various factors create this expectation, but they're not in Steemit's control. It simply isn't realistic to expect new accounts to be widely adopted in the current environment if doing so requires an investment in money, rather than merely time.
I'll point out that, regardless of whether I wanted to or not, I could not have opened an account on Steemit that required any financial investment at all. I hope few would think my presence on Steemit to be a bad thing for Steemit.
I strongly support a sponsorship model for new accounts, and this is essentially what the current model is, although the current model is broken by being to financially draining. I suspect that reputation, as it represents vetting by the community, may be useful as a means of doling out account creation tokens potentiating new account sponsorship.
Whatever is done needs to be financially without cost for new accounts, as people who aren't familiar with Steemit are incapable of understanding how it is different in practice from other social media, and therefore highly unlikely to be willing to invest in an account.
I'd love to see this tested. We can all argue all day long whether or not new users are willing to pay for an account, but it's probably just waste of time. Easiest solution would be to try it for a while and see what happens. My guess is that it's just a marketing problem: if we explain clearly why an account is valuable, potential users will understand it and pay without complaining.
God I love science! As you point out, it is something easily testable.
Let's test it. Well, let's hope 'they' test it =p