Great post as always. One thing I'd like to just clarify is that one would be forgiven for getting confused over whether the social data is being "moved" from the blockchain to hivemind, versus it being mirrored in hivemind. The social information is not being removed from the blockchain and placed into hivemind. The social information remains on the blockchain, what's changing is where the APIs route applications that are asking for that information. It used to be that no matter what an application is looking for the APIs would route the query directly to the blockchain; a database that is optimized for establishing consensus, not powering real time applications. This lead to the blockchain nodes getting huge. But most of the information people wanted to pull from the blockchain had nothing to do with tokens, it had to do with information that had no monetary aspect to it. This is essentially information that was not required to be perfectly accurate every 3 seconds (i.e. consensus). In other words: social information.

Hivemind just copies this information from the blockchain and makes it accessible to developers in a more appropriate database that uses a more developer-friendly programming language. The sacrifice essentially comes down to the information taking more like 6 seconds to be perfectly reliable (still way faster than the majority of other blockchains), but the upside is clear: dramatically reduced node sizes and the ability to add the types of features to Steem-powered applications that user have come to expect from web applications. Esteem has already demonstrated how search can be improved by leveraging hivemind and that was before this big release, so I'm personally really excited to see the possibilities that are unlocked by the release of hivemind.