The only types of areas where this is even remotely the case are things where the rest of the field consistently undervalued a certain concept/category of problems. That's why there is really no language that can do concurrency as easily as Erlang.

The only way to figure out whether a language is good for you or not is to use it for whatever you can think of. Trying to be convinced by an Internet forum that a language is good because someone else made something in it is pointless. People have made whatever you can think of in almost every language older than a few years. Their successes say nothing about whether or not the language is good or not or whether or not it was a painful process to make it.

HN the site, looks deceptively simple. It solves a number of problems simultaneously [0] yet seems to work at scale [1] without bugs. [2]

The success of Arc, a Lisp derivative is in the everyday usage of the site and the hard problems it solves. [3] The original comment stands.

[0] http://paulgraham.com/hackernews.html

[1] "1.6 million page views and 200,000 unique visitors on a given weekday" 2013 ~ http://techcrunch.com/2013/05/18/the-evolution-of-hacker-new...

[2] In fact almost 10 years ago I reported a bug, pg actually responded it was a server reset. https://flickr.com/photos/bootload/406590838/

[3] Submitting high quality links, with high quality discussion while increasing users. That is a hard problem.