Questions about how much education “matters” become unproductive insofar as they are drawn into the nature/nurture frame. The nature/nurture polemic comes down to a question about explaining measured outcomes: i.e., the quantitative magnitude of traits.
What fraction of any given man’s intelligence/capability, or his stupidity/incompetence, can be explained by each source? Decomposing the two contributions allows you to ask: how much did his upbringing and education improve him? And extrapolating from these quantitative results, it seems that we can just as well ask: how much would upbringing and education have improved this (poorly educated) fellow? Then, when we abstract away from concrete counterfactuals about individual morons, we get the question: how much could upbringing and education improve the human mind?
This is a very misleading series of questions. The first question is sound, the last question is teetering on absurdity. The errors of the modern liberal theory of education (upheld by most postwar Western conservatives, no less than by bolsheviks) are only due in part to empirically wrong answers to the first question. Yes, intellectual virtues are mainly a function of heredity, and anyone who doesn’t realize this will have a distorted view of everything else about human society. But education does matter, and attempting to convey the irrelevance of upbringing to the absolute magnitude of mental power without accounting for the dimension within which education is relevant only makes confusion inevitable.
The inevitable confusion is the one that makes people think of educating a child like paving a road or building a tower. The more you educate the child, and the more efficiently, the greater his learning: as though an education could get “longer” or “taller”! People will perceive that instruction, discipline, and intellectual cultivation make the difference between a jungle savage and a gentleman, and if you do not give them useful metaphors for this difference they will go off and find a terrible metaphor on their own.
The analogy that I have used in the past is language learning. As an analogy for education as a whole it has a number of minute advantages (go read Darwinian Reactionary if you want to understand why the analogy works in such fine detail), but it also has one overwhelming advantage for escaping the “nature/nurture” dichotomy: language learning aims at fluency.
Fluency is the final fulfillment of the educational project. You are fluent when you speak the language just the way natives speak it. You will then be on the same page, which allows you to communicate with them. You cannot go beyond fluency: if you attempt to learn to speak the language in a way that is “better” than the rest of the community, it must in some way be different from the language they speak, and therefore will be worse, to the extent that it does not match how they speak and is thus unintelligible to them.
So general education in any particular domain aims at fulfillment, at conformity. One can be educated to the same standard as other members of the community, but not beyond it. He who overshoots the standard misses his target. “Beyond”, in fact, is meaningless; whether a poor upbringing is the product of unusually few years of instruction or unusually many years of unusually bad instruction, what makes it poor is that it fails to fit in with the upbringings of the others.
We can educate our children to achieve coherent organization, to teach them fluent conformity, to synchronize their interactions so that their thoughts and deeds resonate with the thoughts and deeds of their neighbors and their civilization. But the goal is, and can only be, 100%.
Once 100% is achieved, you cannot build on your success by bringing the next cohort of students to 110%, and the subsequent cohort to 121%. Two families can raise two sons to speak their common language with equal fluency, but the smarter boy will say smart things fluently and the other will say stupid things fluently. To attempt to change how the boys speak to change what the stupid boy says is, at best, wasted effort. To whatever extent such attempts do change how boys speak, it only impedes their fluency and damages the organization of the linguistic community. Worse still: to whatever extent changing how the boys speak equalizes the intelligence of what they have to say, this can only come about by making it difficult for the smarter boy to express himself.
(By the way, on my claim that fluency is the final goal of language acquisition: I have sharp commenters, who are quick to nitpick. Yes, style and elocution can improve even after one has achieved fluency; yes, one can become more literate; a polyglot; an expert in the jargon and formalism of various fields; yes, one can intentionally speak in novel ways to achieve particular effects. I could elaborate on these topics if there was interest, perhaps, but after convolutions and minutiae you would understand the validity of the analogy more vividly without being able to learn more about the nature of education from it.)