Ask HN: Detecting popularity stress

I've read today's post about ooc language ("Ooc - modern object oriented c"), I've started thinking about things that should be popular, but aren't... yet.. (think "This must already exist"). Like ooc or maybe someone's blog, which is obscure, but insightful.

There probably should exist an algorithm for detecting some "popularity stress"(tension), i.e. a discrepancy between actual popularity of something and the "should be" popularity. Think of Google Trends or Flu prediction, which is somewhat obvious.. but also think of the stock market - there probably exists an algorithm that can pick some signals about when some stock is going to become much-much more pricey (popular).

Well, Google Trends is obviously detecting post-factum popularity - if something is popular - it will find out that X queries today is much more than Y queries yesterday...

...but I'd like to talk about predictive algorithms.

How can we detect something (a website, a blog, a technology, programming language, scientific paper, direction in which humanity is going to be developing) that has the potential to be popular, but yet lacks the momentum , therefore is under "popularity stress" (maybe there's a better phrase for what I mean).

One signal that I can think of - is whether this "something" is going to make our life much easier, without requiring any/much effort/resources on our part. I.e. C is making our lives tough, but ooc makes it easier. But can we make computers find out what's going to make our life easier? Or maybe some crowd-sourcing effort about what makes people's life miserable and tough?

What differs something that is born dead from something that is going to be huge? (Aside from billions of $$$ in marketing?)

Do you have any ideas for detecting this kind of "discrepancy"? Algorithms? Signals?