We're using steemd to query the endpoints listed with condenser_api namespace. With hivemind, a sync process listens to the blockchain events and syncs related data into an RDBMS.
Which makes it a good alternative to use instead of steemd, directly. Also, since it's a relational database, there is a lot of potential for interesting SQL queries.
Currently, new endpoints are limited. They're listed on hive namespace on the root post. In the future, hive namespace may get bigger. (communities, extra calls, etc.)
RE: Introducing a public Hivemind node