It's hard because of the consistency guarantees that PostgreSQL makes. If you need true multi-master replication (you probably don't) the best solution is to use some form of queue to manage writes across the cluster. The DRBD solutions sound awesome but trade one SPOF for another, and have drawbacks in terms of speed (replicating block devices makes them slow) and risk (network latency, write contention) and are limited to at most two masters.

If you find yourself needing a replication solution beyond hot standby; it's possible that PostgreSQL isn't the right tool for the job.