2. Whereas with Cryptonote (and Zerocash) what needs to be unencrypted is neatly compressed with complete organization on the block chain, off chain routing can create mazes of extreme complexity. In the asymptotic case, the authorities would need to cross correlate every encrypted packet ever sent on the internet.
Only if things are implemented "properly" at every level. That is a huge assumption I'm not prepared to make with any of the current altcoin efforts. I'd rather rely on well-designed and well-vetted encryption. Implementations of large systems that require sending information around the internet to mixing nodes and such has a
huge attack surface. Cryptography implementations are potentially orders of magnitude smaller (and Cryptonote really isn't that complex) so far, far easier to vet at a systemic level. Granted the issue of quantum computing vulnerability is valid, but it likely to be a acceptable risk to a great many users, but not to you. Fair enough. Build something better.
3. Cryptonote has no IP obfuscation built in (yet), thus unless you are using Tor with it, the on chain anonymity is already cracked. Which means even if you use Tor, if the others in your anonymity set ring didn't use Tor, then you are de-anonymized. And even when Cryptonote adds I2P or Tor support by default, it isn't planned to be supported for mining, and those low-latency mixnets are shown in research to be vulnerable to timing analysis. There are mathematically characterized better designs for IP obfuscation for crypto-currency than I2P and Tor.
Tor is already supported for mining. There is no reason why you can't connect to a pool using Tor. Most pools require no registration, so any coins that go back to you via pool payments can't be traced to you by any mechanism other than attacking Tor.
5. You won't get decentralized mining without off chain anonymity.
Not proven, nor proven that you can get decentralized mining with off chain anonymity.
So again I reiterate, why risk it with on chain anonymity when there can be designs that are exponentially more secure with your anonymity into the future?
Why risk off chain routing attacks when (some future non-existant) on-chain system can be made exponentially more secure, with greater scrutiny of the components parts (encryption) then some large and nearly impossible to analyze set of interconnected elements.
See, FUD works in either direction. Stop doing it, and start building. When we see how much better your system is, we will all be convinced! (The salesmen representing other coins, including other off-chain anonymity coins won't, but you can't ever convince them no matter what you say or do.)
6. The claim that Cryptonote has a larger anonymity set because it can mix from the entire history of the block chain, whereas CoinJoin has a simultaneity constraint, is not true because to be prunable the rings must be restricted to small groups, and as I showed in my bounty algorithm upthread, if you allow widely overlapping mixing then the rings can in theory be de-anonymized.
You didn't show anything at all about the scope and degree of unmixing, so we have nothing to say here, just more FUD.
In short, prove on-chain wrong by constructing something better. Until then you are behaving similarly to the shills you hate.