Some Feedback:
1) Why bother with Phase 1? I think what people want is fully decentralized, trustless anonymous coin. The first coin to do it will be very popular. Implementing phase 1 seems to me like it may simply divert a lot of time and energy from achieving phase 2. Not just in terms of initial development but also building out the infrastructure, bug fixes, documentation and support, etc, etc. So if phase 2 is achievable, why not just announce that is the roadmap/goal/plan and go for it? Requiring a trusted third party (phase 1) just does not interest me and probably many others.
2) The document does not seem detailed enough for any third party security researcher to fully grasp the workings and audit the logic. I don't believe any coin will reach full value/potential until that hurdle is cleared.
3) It is unclear if mixing will be a) the only possible way to send, b) the default way to send, c) opt-in way to send. My recommendation for an anonymous coin would be (a). If it is known that everyone is mixing, then mixed transactions do not stand out or look suspicious. There is much more noise, and correspondingly less signal to follow than in an opt-in scenario. (a) would also mean the highest level of fungibility, which is generally considered a useful property of a currency.
4) I did not see defined how a node (phase 2) would know which of its peers can be used for mixing.