On your 7 points:
the thief has agreed to arbitration because it's in the Constitution. Unless you take it out ... :-)
defence in depth. This works well if there are 3 day delayed stakeholded amounts or other defences. Use defence in depth. Make it your friend. Actually, many friends.
is wrong, total bunk. Yeah I understand that people might believe that, but people used to believe the earth was flat and the moon was made of cheese. Let's move on.
yes - but that's humanity and removing humans because we're corruptible means removing everyone.
Ask yourself this: is it easier/faster to corrupt the DPOS system or the Arbitration system? if your answer is the Arbitrators you are empirically wrong.
- compared to what? What's your definition of 'centralised' v. 'decentralised' ? Let's cut to the chase. ECAF is decentralised across N arbitrators. Each arbitrator takes on a case, essentially at random. So in order to reliably breach the ECAF system you have to (a) put a dodgy arb in place (b) make sure your dodgy arb takes on the case, and (c) deal with all the checks and balances that will stop a dodgy ruling.
In contrast, to breach BPs, you just have to trick one BP once to get a bad tx through. Guess which one happened already - twice.
Bunk. It scales much better than BPs, when funding arrives :-) What's scaling now slowly is volunteer ECAF with only 6 arbitrators. do the math... with funding, we'll move past 21 within a few months.
bunk. if you call someone 'police', they will act like police. It helps the argumentation to use the right words.
The BPs are (like) the police. The Arbitrators are (like) the courts. Literally, practically, and intentionally, the original judgement capability within DPOS has been taken away from the BPs, leaving them the policing or enforcement function, only.
RE: Proposal for an incremental Constitution and Dapp layer governance on EOS