without knowledge of the objects they contain. This will allow new
transaction or account formats to be added (possibly with larger data sizes)
without breaking the ability to parse the hash trees. It also simplifies
the operation-specific tree code (since it doesn't have to parse the raw
leaf data).
Google's protocol buffers cannot be used for the inner portions of
signed/hashed objects.
This SQL database layout will facilitate a cache-based transaction
processing architecture that will scale to billions of accounts and dozens
of transactions per second.
