Balanced intersection size distributions in projective planes

Given a point set $S$ in a projective plane $\Pi_q$ of order $q$, each line $\ell$ determines a secant size $|S\cap \ell|$. We study how balanced the secant-size distribution can be for the line set $\mathcal{L}$ of the plane, in other words, how many lines must share the same secant size. We show that $\min_{ S\subseteq \Pi_q} \max_k |\{\ell\in \mathcal{L}: |\ell\cap S|=k\}|=\Theta(q^{3/2}).$ This shows a large contrast with the case of real projective (or affine) plane, where $\max_{k>1} |\{\ell\in~ \mathcal{L}: |\ell\cap S|=k\}|$ is always at least the third of $|\{\ell\in \mathcal{L}: |\ell\cap S|>1\}|$. We also discuss explicit constructions in addition to randomized point sets, that are asymptotically close to be optimal, and point out a link between the constructions and character-sum estimates. Finally, we explore the relation between balanced secant size distributions and legitimate colorings, studied by Alon and F\"uredi, and prove a result that might resemble the Erd\H{o}s-Faber-Lov\'asz conjecture.