Earlier quoted context omitted.
I wonder if national averages actually matter very much beyond being a vanity statistic? While it's good to have a generally more mathematically capable population, you would expect most actual "progress" (papers, discoveries, things built) to be made by the outliers. I guess there's another layer tho in 2025 where, assuming they are rational, you wouldn't necessarily expect their efforts to benefit Romania.
In a global economy I think a good national average is more important than the peaks. The genius invention will spread, and there is significant value in having a population which can utilise it. As another example, if cancer is cured tomorrow it won't necessarily be the country which found the cure which will be cancer free first, but the one with the most competent doctors, able to use and apply the cure.
Second, I would caution against the trope that invention is some exclusive province of the "genius".
That out of the way, we can look to network effects. Consider that a populace with high literacy will permit more of the kind of collaboration that has literacy as a prerequisite, while one with low literacy and a few outliers won't. The outliers are constrained in how much they can collaborate, and collaboration is essential to scientific and economic development.