No, it's not ready to replace SQL, and I don't think it ever will. What are your requirements? If it's horizontal scalability (and you're actually hitting a performance wall) you should begin to think about it. Maybe also if you never do any joins.
Relational database systems (+ normalization) compromise everything to ensure the ACID properties, which for the majority of cases, is the most important part.
The unique location is the big plus, especially considering that the absence of geostationary satellites over the poles means this is likely the ONLY developer job on the planet that can't be done offsite, until they run fiber to Antarctica.
I'm just saying that all the author is observing is that the students with the most to lose, work the hardest. Foreign students obviously have a lot to lose since they're coming overseas for an education. On the other hand, the American students she interacts with are likely wealthy and don't necessarily need stellar grades to move on to a professional degree program e.g. medical school.
It doesn't explain any difference between American and other students, just that Babson's student body is different than most.
From personal experience i've found the hardest working students to be those needing to get into professional school (law or medicine) as well as low-income/disadvantaged students who couldn't afford the $50k tuition/year to attend a private college.
Backtype is a YC company that probably does some NLP stuff on the backend. http://www.backtype.com/
Also, Powerset and Wolfram Alpha.
The most obvious killer app is sentiment analysis applied to product reviews. Overall I don't feel like NLP has game-changing applications yet since good data is structured data. The research is in place, though.
Since you're in the bay area, you should look into regional meetups for people learning/using rails. There's one at Berkeley the third Tuesday of each month (Nov 17 this month)
The name of this article is really unfortunate and will probably mislead many readers.