I've learned to not believe that distributed software will work in practice in the way that its authors claim it will. The stronger the claims, the more important it is to have an independent test validating it before I even think of trusting it.
Totally understandable. We have been talking to Kyle. We have internal verification systems and will be publishing more later this year to that effect.
For what it's worth, we have built high-performance distributed systems before....so it's not just wishful thinking.