They only guarantee consistency as long as you don't use them over a network, i.e. communications with the database are always reliable. But once you do use them over a network - CAP theorem comes in and forces you to either use something like two-phase commit or no promises of consistency. Which is the opposite of what his post implied, like there is CAP consistency with those databases. But there never was!

Although I kind of got used to RDBMS crowd not understanding consistency, it's just another technology cult.

Proper RDMS databases have two phase commit with transactions.....

Which ones? Traditional mainstream RDBMSs, like Mysql and Postgres don't use two-phase commit protocol. Obviously new distributed ones do it properly, but we are not talking about them.

Wait, CAP theorem just says that you have to sacrifice availability of you want Consistency and Partition tolerance.