tor and other group anonymity services aren't about hiding your data but about hiding your endpoint. the tor network sentinels publish their exit nodes and prioritize low latency high speed exit nodes, so it's not all that hard to tell, AS THE ENDPOINT, whether someone is using tor and in some cases the data they're sending (onion routing encrypts _routes_ not data). further, while tor exit nodes are published, tor _entry_ nodes are not, as anyone running an onion router can be an entry node. hence, the point of crowd anonymity is:
a) the endpoint doesn't know _who_ in the tor network is sending the data unless the host unmasks itself. it cannot analyze the route, because it only knows the route from the exit node.
b) what is far more important from the perspective of the tor inventors and maintainers is that a network monitor which only has visibility of the pre-exit-node overlay network and the host cannot tell where the host is sending packets, and ideally, whether a given host is part of the onion routing network (although this is a property that can easily collapse -- it is often protected by sheer numbers).
the tor maintainers prioritize the ability for MONITORED HOSTS to leave a MONITORED NETWORK without giving away their data's destination. while onion routing has other properties that can be maintained with varying implementation details, this is the priority that the tor maintainers focus on.
