"...you can't build communities when you don't know...whether two accounts are of the same person or of two different persons..."

Why not? If someone manages two different accounts differently, you can relate to either identity per their behaviour. If they simply have multiple accounts they manage with the same behaviour, why would you treat either account differently?

What difference does it make who is the actual meatbag behind the behaviour?

"I agree the game is rigged (a guy from Nairobi or Aceh has little chance to be able to buy several hundred $ worth of bot voting) but the solution is not "opininon wars" that are ultimately destructive ("I think your content is not valuable so I harm you") but rather correcting the problem from its root."

Well, the root being discussed is botvotes for profit. Thus decreasing the profitability of the botvotes by flagging is directly getting to the root of the problem.

"To get back to solving Steemit abuse, I am convinced that the solution is to have true, real, provable identities."

How would that impact buying botvotes, selfvoting, or circlejerks?

I can easily see how it could prevent people from speaking forthrightly. I do not see that as a benefit.