I agree with the author that hstore is very interesting, but the data structures are not the key selling point in the NoSQL space in my opinion. The most overlooked advantage to things like Cassandra and Riak are the fact that you have no single point of failure in the system. If an individual node fails, there is no operational impact.
Postgres does have (finally!) a nice replication story, so you have data protection on a second server, but the failover mechanics are much more complicated and you still have a single point of failure at the master/writer node. The story gets even more operationally complex when you talk about sharding and the fact that you now have to have a replica per master -- and it really needs to be the same size as the master if you want to be able to trust that you have enough capacity to fail over to it. Suddenly you need to have half of your database capacity sitting idle.
Now, don't get me wrong, I think Postgres is a wonderful database server. For the vast majority of applications it is the correct default choice. Very few applications ever get to the point where they need to scale further than they can get vertically on a single node and most can tolerate a few minutes of downtime to fail over to a backup server. Hand-wavy "sharding is easy" comments, however, ignore a lot of operational reality and that's dangerous.
Understand your use case. Understand the failure modes. Decide how available your datastore needs to be and how much data you need to support. Know the consequences of the choices you make.