I'm working on a problem where I need to prove that a posterior variance of binomial data with a uniform prior is less than 1/12 (the variance of the prior). Currently I have $(\frac{y+1}{n+2})$ $(\frac{n-y+1}{n+2})$ $(\frac{1}{n+3})$ For the variance. Since the first two fractions sum to 1, I know conceptually that the maximum product is $\frac14$. However, how can we prove that the maximum product of two fractions who sum to 1 is $\frac14$?
I know from there how to prove the inequality, but I just don't know how to prove the maximum product.
We have two real numbers $0 < a, b$ such that $a + b = 1$, so $b = 1-a$. We want to maximize $a\cdot b$.
Substitution gives $a \cdot b = a(1-a) = a - a^2$. What is the maximum possible value of that expression?