We have presented a design for a content-neutral decentralized anonymous storage service and investigated the kinds of anonymity required for such a service. The design protects anonymity at least as much, and in some cases more, than related work. Unfortunately, we have also shown that this does not say that much, because large problems remain with evaluating the level of protection provided by any of these systems, much less providing ``good'' protection. Until the risks involved in using such systems can be better evaluated, they cannot be used in good conscience for situations where failure is not an option. Much more work remains.