No, merge works recursively. Merge is just a computational operation whereby two syntactic units are combined together to form another syntactic object (for example, "the cat was killed" and "when it was hit by a car" produce, when operated upon by merge, the sentence "the cat was killed when it was hit by a car"). Merge is recursive because it is an operation that can be applied to its own output (in the last example, the inputs were "the cat was killed" and "when it was hit by a car", the computational operation was merge and the output was "the cat was killed when it was hit by a car"; via the application of the same merge operation, this output can then be combined with yet another syntactic unit, like "that was traveling at sixty kilometers an hour" for example, to form another linguistic unit: "the cat was killed when it was hit by a car that was traveling at sixty kilometers an hour"). So merge works just like any recursive function does in the theory of computation, which is why merge has the property of recursion. There's no "downgrade" from recursion to merge, you just didn't understand the theory.
Moreover, Everett's argument is fallacious in the most vulgar sort of way. Chomsky's claim is roughly the following: (1) if humans acquire recursive languages, then humans must have a genetic capacity that allows them to mentally process recursive languages. Anyone who knows their logic is aware of the fact that this conditional does not imply the following proposition: (2) if humans have a genetic capacity that allows them to mentally process recursive languages, then humans must acquire recursive languages. That is, in general, "if A then B" does not imply "if B then A". In finding an example of a human language that does not feature recursion, all Everett has done is to disprove proposition (2), but this proposition is not implied by Chomsky's actual claim, and so Everett's (very much disputed, if not refuted) discovery is irrelevant as far as the viability of Chomsky's program is concerned.
As for a younger generation of linguists abandoning ideas of innate grammar, well that might be partially true. At the same time, there are academics trying to argue that there are no such things as genetic programs, and that all talk of information being encoded in DNA is meaningless. In both cases, this doesn't represent the majority view, which is fortunate because in both cases it simply demonstrates a shocking lack of understanding of theory of computation. As far as I can tell, dissent from this basic view of Chomsky's amounts to an infantile and misdirected fear of genetic determinism (of course, many people, like Pinker for example, dissent from Chomsky in the finer details, and that's where the media coverage should be concentrated; at the moment, what the media is doing is analogous to claiming that the general theory of relativity has been disproved instead of focusing coverage on the more subtle disputes--the ones of actual importance to physicists--that exist between scientists who accept general relativity).
Re: “Language: Innate or Invented?”
No, merge works recursively. Merge is just a computational operation whereby two syntactic units are combined together to form another syntactic object (for example, "the cat was killed" and "when it was hit by a car" produce, when operated upon by merge, the sentence "the cat was killed when it was hit by a car"). Merge is recursive because it is an operation that can be applied to its own output (in the last example, the inputs were "the cat was killed" and "when it was hit by a car", the computational operation was merge and the output was "the cat was killed when it was hit by a car"; via the application of the same merge operation, this output can then be combined with yet another syntactic unit, like "that was traveling at sixty kilometers an hour" for example, to form another linguistic unit: "the cat was killed when it was hit by a car that was traveling at sixty kilometers an hour"). So merge works just like any recursive function does in the theory of computation, which is why merge has the property of recursion. There's no "downgrade" from recursion to merge, you just didn't understand the theory.
Moreover, Everett's argument is fallacious in the most vulgar sort of way. Chomsky's claim is roughly the following: (1) if humans acquire recursive languages, then humans must have a genetic capacity that allows them to mentally process recursive languages. Anyone who knows their logic is aware of the fact that this conditional does not imply the following proposition: (2) if humans have a genetic capacity that allows them to mentally process recursive languages, then humans must acquire recursive languages. That is, in general, "if A then B" does not imply "if B then A". In finding an example of a human language that does not feature recursion, all Everett has done is to disprove proposition (2), but this proposition is not implied by Chomsky's actual claim, and so Everett's (very much disputed, if not refuted) discovery is irrelevant as far as the viability of Chomsky's program is concerned.
As for a younger generation of linguists abandoning ideas of innate grammar, well that might be partially true. At the same time, there are academics trying to argue that there are no such things as genetic programs, and that all talk of information being encoded in DNA is meaningless. In both cases, this doesn't represent the majority view, which is fortunate because in both cases it simply demonstrates a shocking lack of understanding of theory of computation. As far as I can tell, dissent from this basic view of Chomsky's amounts to an infantile and misdirected fear of genetic determinism (of course, many people, like Pinker for example, dissent from Chomsky in the finer details, and that's where the media coverage should be concentrated; at the moment, what the media is doing is analogous to claiming that the general theory of relativity has been disproved instead of focusing coverage on the more subtle disputes--the ones of actual importance to physicists--that exist between scientists who accept general relativity).