Now that I've spent some time toiling on different types of research (qualitative, quantitative, confirmatory, etc.), I think that I can finally sum up the requirements for good research:
1. Rigorous. Researchers have to sweat. The process can' t be easy. It must involve a good deal of work doing... well, something: collecting data, interviewing people, digging in archives, translating from an obscure and dead language, etc. The only black box in research is the researcher him- or herself. The best apparatus, data set, or algorithm is only as good as the investigator. And becoming a good investigator requires works.
2. Commensurate. To borrow a hackneyed expression from Newton, the best researchers stand on the shoulders of giants. Hence, they must create research that both builds from and extends the work of those giants. Breaking new ground in unexplored wilds is good for prophets and psychotic charismatic leaders; researchers should stick to development in the suburbs. Without some basis of comparison, research can only be considered fringe.
3. Rhetorical. I don't mean obvious or not requiring an answer. Instead, I mean capable of supporting an ongoing discussion and capable of defending against hostile attacks. Rigor and commensurability obviously support the rhetorical aspect of good research. But so does appropriate positioning and establishing bridges to those stable extant pillars of research. The rhetorical nature of good research is really one of performance. While the first two concerns are inherent elements of research, rhetoric (written or verbal) depends on the execution of the researcher. Again, rigor in research approach and intellectual randori are crucial.