Intelligence confined to further analysis in a dead end isn't nearly as useful as that sort which makes connections and goes places others haven't. Consider, for instance, what perhaps was Richard Feynman's greatest contribution to physics. Julian Schwinger, too, described quantum electrodynamics (QED), using virtuosic, massively complicated mathematics. Feynman did it using instantly understandable diagrams. Freeman Dyson proved the approaches equivalent. Feynman's quickly became universally used, even amongst those who speak math the way Shakespeare wrote sonnets.
Stephen Jay Gould's 'The Mismeasure of Man'--still intellectually invaluable, and a great read--centered, in its most rigorous parts, on Spearman's 'G', a statistical umbrella reducing multiple measurable entities into a single number, and its misuse in describing intelligence as both a single quantity and one that could be inherited. Just about any person of genius demonstrates that there's more than one way to be intelligent. Some might even call that a defining characteristic of greatness transcending competence.
One of Gould's observations is the statistical validity of phrenology, as opposed to its accuracy or utility. Reminiscent, seems to me, of Feynman's observation about parapsychologists' self-assessment of the statistical validity of their assertions. I had the great good fortune to have encountered Prof. Lloyd Motz of Columbia's astronomy department, who was, amongst other things, investigating the intersection of gravity and particle physics when doing so wasn't cool. When the Venus probes found its surface far hotter than anybody had predicted, Prof. Motz credited priority to a long-term intellectual nemesis, Immanuel Velikovsky, who predicted it out of his potted cosmology. An act of intellectual honesty, and a rare one.