The paper investigates how often a polynomial function, when evaluated over a finite field, fails to hit certain output values. This is measured by the "value distribution" of the polynomial. The key insight connects this to a concept from topology and algebra called monodromy, which tracks how the roots of a polynomial equation get permuted as you move around in a parameter space. Specifically, output values that get missed correspond to permutations with no fixed points (called derangements). A random symmetric monodromy group misses roughly 1/e of all values, but a special algebraic structure called a sharply 2-transitive affine group can do much better, missing as few as 1/n values, which turns out to be the best possible.
The paper constructs an explicit family of polynomials that achieves this optimal bound. These polynomials are built from the arithmetic of finite fields using a parameter h that divides the field size minus one. The authors work out exactly how many times each output value is hit, across all relevant finite fields, including cases that fall outside the cleanest theoretical framework. In the extreme case, a specific simple-looking polynomial of the form U times (U minus 1) to a large power turns out to be the best possible non-permutation polynomial over every finite field containing a certain subfield, and the paper gives precise formulas for how close it comes to the theoretical optimum