SOLVED (LEAN) on erdosproblems.com, the label as it stood on 2026-09-04, its Lean marker referring to the formalization reported in the forum and recorded below; with a formulation qualification. The site's statement above is preserved verbatim as it stood on 2026-09-04. Its annotation $f:X^2\to\{-1,1\}$ reuses the subset $X$ over which the maximum is taken, so it does not fix a single domain for $f$. If the domain is read as $[n]^2$, with arbitrary signs on ordered pairs, the resulting minimax is identically zero. The classical unordered-edge question has order $n^{3/2}$ for sufficiently large $n$. The site's label, solved, does not identify these two quantities or assert an exact finite-$n$ formula or leading constant for the latter.