Programming language stuff happens over LOTS of time. There is something of a proverb saying it takes 10 years for a language to mature to the point of being stable, useful, and popular enough to actually take off. Trying to predict which language will be the next big thing is like trying to predict what where you will be 10 years from now.
The languages that seem to catch on are not the ones that do academically interesting things; in fact it seems to be the more down to earth languages have become the most popular...the McDonald's languages I call them *cough* PHP *cough*. What happens in the wide use of programing languages seems to have little to do with technical merit and more to do with ease of setup to do a job, the libraries available, the historical use, and advertisement than anything. I've kinda lost hope of seeing truly interesting languages in any wide spread usage.
Perhaps a universal language-bridge will evolve. I shall call it Bridge . Kinda like the CLR or JVM. We can plug it into our development environment and choose to develop in ANY language we fancy. My pal can develop in his language of choice and we use the Bridge to pass messages.
I'm going to say that LISP based languages are going to gain popularity exponentially within the next 25 years.
Now as to why: Most modern languages add new features, ones that were invented in LISP ages ago. The features that are deemed new are already there and far expanded upon in LISP, and people searching for the next greatest thing after a language gives them a couple pieces of meat is the full steak. The more languages evolve, the more they approach LISP.
Until a more superior language is released, LISP will be the inevitability.
TL;DR: Lemur's First Law - All languages will approach LISP as they evolve, until they become a LISP.
The closer a language is to LISP, the more likely someone is to find it and make a jump. Ruby programmers are starting to flock around Clojure and Erlang/Elixer en masse because Ruby itself isn't too many deviations from LISP in some aspects. The metaprogramming and power that we love in Ruby is so much more in a LISP that it makes sense to follow in that path.
TL;DR: Lemur's Second Law - Programmers will seek more power, and in doing so make a jump to more LISP like languages.
Now on the subject of abstraction, computers are becoming so much more powerful that if we spend forever optimizing in low level nothing will ever get done. As soon as Quantum comes, you'd have to be a fool to try and justify writing thread safe code in something like Assembly or another low language. It just would not make sense at all to spend months on something that gives a trivial time difference, such as Binary today.
TL;DR: Lemur's Third Law - As computers evolve, lower levels of abstraction will become irrelevant if not completely useless due to the raw power of processing.