This is a question of uncertainty-principle type stemming from Eigenvalue of a convolution and a restriction?
Let $f:\mathbb{R}\to \mathbb{R}$ be even, absolutely continuous and supported in $[-\frac{1}{2},\frac{1}{2}]$, with $f'$ of bounded total variation. Let $|f|_2=1$. What is the minimal possible value of $$I(f) = \int_{-\infty}^\infty |t| |\hat{f}(t)|^2 dt?$$ Which choice or choices of $f$ minimize $I(f)$?
Here "absolutely continuous" and "$f'$ of bounded total variation" are technical conditions that guarantee that $f(t)=O(1/t^2)$ as $t\to \pm \infty$.
PS. MathOverflow just noticed something I had forgotten, viz., I asked a question that turns out to be equivalent two years ago in Distribution $f$ such that (a) $\widehat{f}$ has compact support, (b) $\mathbb{E}(|X|)$ is minimal? . Clever robot! That question got 9 upvotes but no full answers, so it seems legitimate to ask a not-so-very-obviously equivalent question.
On another note: $f(x) = \cos(\pi x)|_{[-1/2,1/2]}$ is a natural choice - can one do better?