The style of tradeoff taken, in both cases, really exemplifies the difference in engineering priorities for MySQL and PostgreSQL. MySQL optimizes for reads, at the expense of a higher write load. PostgreSQL optimizes more for writes, at the expense of in-line bloat and reads.

When you go to scale out a RDBMS, traditionally you do it with single-master multi slave replication, to scale reads. Today, that's not an inordinately hard task. Scaling writes is a LOT harder, and I think PostgreSQL made the better choice, in the end.

Plus, PostgreSQL built in master-slave streaming replication in release 9, so scaling out reads is even less of an issue.