It seems like every time I read one of these articles, it is written by someone with no knowledge of what is actually out there. I suppose that is normally because anyone with enough time to mouth off on the internet in article or blog form is not actually doing real work. The rest of us are too busy, you know, working?
Anyway, as someone that has been programming, particularly against just about every major database platform out there, I can tell you that there has always been a battle between relational vs. other types of databases, most notably graph databases. Full disclosure: I am partial to graph databases, but that is because I find a lot of utility in working in pure code and I also apply graph theory to a lot of my work.
There is no silver bullet and each kind of database is better at certain tasks. Long ago, hardware made much more of a difference than it does today and was one reason relational databases "won" out. Other reasons include marketing, the development of SQL and other standards, and the ease of applying relational mathematics which are easy to understand. I spent most of my life working with relational databases as they were simply in my comfort zone, until I realized I needed to get off my ass and learn more than just new languages and programming techniques (and realized relational dbs were such a huge fail at tasks like traversing graphs, storing dynamic columns ala a CRM, etc).
Particularly when you look at older languages like LISP, relational was a great fit vs. graph. Since then, many factors have caught up and many more languages, solutions, and designs are out there. In the meantime, graph databases never went away. It seems suddenly since the outbreak of "web 2.0" frameworks and ORMs, a lot of people who don't have a clue about SQL, and especially databases in general of all types find it a great idea to go out and make their own or put some huge hack on top of a relational db, or perhaps worse, try to come up with something entirely new that is not based on fact or need.
I've played around with some of the databases mentioned above, and just like MySQL, they are mainly reinventing the wheel badly. The google implementation is the only one I have seen that is not completely shoddy, but color me somewhat unimpressed as well. FYI, just because xyz company such as google or IBM uses some product does not make it good. On the contrary, it's often a warning sign as big companies are often what keeps sites like thedailywtf going.
Regarding graph databases, I'm currently using Gemstone with Smalltalk for instance, and I have used it with Java as well. I can tell you it is great, but it is no panacea. It's been around forever like many of its competitors which are also noteworthy, but a side-by-side comparison is best left outside the scope of this comment. Gemstone lets me avoid any ORM overhead and I can write and maintain queries, transactions, reports, etc in one place; my code where they belong for my projects. It's fast as hell and I have collections with millions of objects and there is no slow down. In fact, I migrated many of my databases from other systems including Oracle, DB2, MS SQL, and Postgres, and it smokes them all, but I am confident that if I can think of a huge laundry list of tasks where the opposite will be true.
For instance, I can do reporting using Gemstone as I mentioned, but it is left up to me writing my own code or using a library, where in something like MS SQL, I can just use reporting services to do more complicated cubes, or one of the 100 enterprise reporting tools of various size for Oracle. It also sucks in some ways if you need to share a lot of data between applications because you have to make decisions about what do you want to split out and how to manage the memory and load.
Even Gemstone itself is not so bold as to presume there is no use for relational databases. You can build a SQL layer on top of it and it has many tools to move data back and forth between any relational db, for instance.
The authors of the article would do well to take a look at real graph dbs instead of crap like Harry Potter DB and Couch surfer DB. Just reading over the philosophy behind Couch DB makes it clear to anyone who has spent time working on serious databases that the devs don't have a damn clue. Go back and take some math, learn about b-trees and calculus, and stop writing stupid articles.
I am going to tear my eyes out if I see "yet another tuple store or graph db." Welcome to the last century, please try again. In Smalltalk and Common Lisp alone, we have tons of graph DBs, ranging from Magma to Elephant. Please stop making more cheesy ones that suck little snowflake, and doubly so stop writing ill informed articles. /rant.