Hence what, for lack of a better name, I'll call the Python paradox: if a company chooses to write its software in a comparatively esoteric language, they'll be able to hire better programmers, because they'll attract only those who cared enough to learn it. And for programmers the paradox is even more pronounced: the language to learn, if you want to get a good job, is a language that people don't learn merely to get a job.
Paul wrote that in 2004, and since then, many companies have embraced these languages. For proof, check out the percentage of jobs on indeed.com that mention Python or Ruby. According to indeed, 0.6% of all job postings for all positions reference Python. We have reached the point where these technologies are no longer 'comparatively esoteric', where plenty of developers learn these languages just to get a job.
It's tempting to say that the logic behind the Python Paradox still holds true, and that we simply need to find other languages. To attract curious, great programmers, maybe we should be building our software in the Clojure or Erlang. I don't think this is right. Even back in 2004, it was easy to build great web apps with Python, while I wouldn't wish building a non-trivial Haskell web application on my worst enemy. The current wave of new languages are interesting, but I think they work best for a subset of problems and that many of these problems won't be encountered by most web startups.
I still do think that startups can find great developers based on technology choices, just not languages choices.
Ladies and gentlemen, I present the NoSQL Paradox: if a company chooses to store its data in a comparatively esoteric datastore, they'll be able to hire better programmers, because they'll attract only those who cared enough to learn it. And for programmers the paradox is even more pronounced: the datastores to learn, if you want to get a good job, are the datastores that people don't learn merely to get a job. See what I did there?
NoSQL stores like MongoDB and Cassandra are where Python and Ruby were, comparatively speaking, 7 years ago. They're featureful, they can be applied to many different problems, they're driving a lot of innovation, they have passionate followings in the development world, and best of all, the vast majority of companies aren't using them. To prove that last point, I point you to the indeed figures for job postings mentioning MongoDB and Cassandra, each of which are about 0.02% of all postings or 1/30 the Python figure listed above. While the vast majority of web companies may not need Erlang, they can certainly use one of these databases.
Anecdotally, I can vouch for this. At Famigo, we build everything in Python on top of a variety of different datastores, including Mongo, Redis, Solr, and CouchDB. As we're looking to hire more people (if interested, email codypo at gmail!), I talk about our stack with a ton of programmers. The good ones want to talk about Python, while the great ones want to talk about Redis and Mongo.
Just like other companies eventually grokked the Python Paradox and hired accordingly, they'll do the same here and mitigate this advantage. It'll take a while though, and until then, I'm going to build and hire accordingly.