Well my platform is n number of $20 linodes to start. I'm clustering the python application across them using uwsgi+nginx (all I have to do is add an IP address in the config to scale), it's going to be a given that I shard the database across them as well. If you feel I should avoid Mongo would you recommend Cassandra instead?
Regardless, I think my initial question regarding when to denormalize data applies to any database including scaled MySQL, but perhaps was a better question for stackoverflow.
Cassandra has it's own hurdles, but I think if we're talking about getting your mind in the right place, it might be a better answer. Cassandra definitely has a much more mature scalability implementation that isn't caveat-ridden like MongoDB is. It's operating at scale at both Twitter and Facebook.
Cassandra has online compaction, but still requires up to 2x space for compaction. However, Cassandra does not have to do a full scan of the entire database to do compaction, and almost never actually uses the 2x space. It's also much easier to maintain a Cassandra cluster, because each instance shares equal responsibility, and replication topology is handled for you.
Despite what their fans will say, these are both beta-quality products.
Regardless, I think my initial question regarding when to denormalize data applies to any database including scaled MySQL, but perhaps was a better question for stackoverflow.