Re: That old chestnut
The problem the article (and, presumably, the original paper) highlights is that because of this one-way communication, new ideas for new protocols and new services are just not happening, except within the private networks of the largest players (Facebook, Google, ...).
If you have an idea for a great new service which relies on communication being initiated from your server to your client, you can't do it - unless you are Google or Facebook (which almost everyone keeps a live connection to all the time).
We see that with email polling. It is an abomination that every phone needs to keep an open TCP connection all the time to its email host in order to be told about new emails.
Yes, it is true that the security and privacy implications of discarding NAT need to be better addressed (particularly for the cheapest of devices - like typical domestic IoT). But they haven't been because NAT is hiding the real underlying problems.