> [T]he latter is the one you'd want to read ... if you wanted answers to questions.
I agree with this assessment, but seeking the answers to questions is only part of why I go looking for papers to read: a lot of the time, I am looking for tools / techniques to solve my particular problem, and a paper that offers a novel technique (even if the author(s) have only applied to to a single specific problem) is very enticing in that case. Papers built around a novel technique get a lot of citations because they engender new attempts to problems that were difficult to solve before the technique existed.
And when everybody is suddenly exploring a new technique, testing the boundaries of what the technique can do is very interesting, making negative results significantly more publishable than normal. Lots of people know that $OLD_TECHNIQUE can't solve $PROBLEM_FOO, but the fact that $NEW_TECHNIQUE can't solve it says something quite interesting about the capabilities of the technique.
But surely, it's clear that using the technique is the "terminal value," and inventing the technique is the "instrumental value." The inventor of the technique will be greatly rewarded, but the only reason what they did was valuable is that others will come along to use their technique. It must be kept in mind that the article is suggesting eliminating your role (that of the person trawling the literature for existing techniques to solve their problem), to focus on the invention of new techniques singularly.
I agree with this assessment, but seeking the answers to questions is only part of why I go looking for papers to read: a lot of the time, I am looking for tools / techniques to solve my particular problem, and a paper that offers a novel technique (even if the author(s) have only applied to to a single specific problem) is very enticing in that case. Papers built around a novel technique get a lot of citations because they engender new attempts to problems that were difficult to solve before the technique existed.
And when everybody is suddenly exploring a new technique, testing the boundaries of what the technique can do is very interesting, making negative results significantly more publishable than normal. Lots of people know that $OLD_TECHNIQUE can't solve $PROBLEM_FOO, but the fact that $NEW_TECHNIQUE can't solve it says something quite interesting about the capabilities of the technique.