A 2025 International Math Olympiad tiling problem begins with a deceptively simple constraint: leave exactly one uncovered square in every row and column of a 2025-by-2025 grid while covering everything else with as few rectangular tiles as possible. The difficulty becomes clearer once the goal shifts from merely constructing an efficient arrangement to proving that no better arrangement can exist. By noting that only six contestants earned full marks, the presentation establishes the problem’s formidable difficulty without using that difficulty as a substitute for actually explaining the mathematics.
The early decision to shrink the grid and experiment with different arrangements is particularly effective. Rather than dropping the optimal construction onto the screen, the explanation develops an intuition about efficiency by asking how many uncovered squares can border each tile. The seemingly unrelated cube-cutting puzzle then introduces the idea of proving a lower bound through correspondence. It is a lengthy detour, but it earns its place because it demonstrates the broader problem-solving lesson the presentation wants to emphasize: seemingly inspired insights often grow out of earlier mathematical experiences.
That groundwork leads naturally to the elegant windmill arrangement of square tiles, where the uncovered squares sit around tile corners and occupy distinct rows and columns. Generalizing the pattern to tiles of side length k produces a grid of side length k 2 , making 2025 being 45 2 suddenly meaningful rather than arbitrary. The resulting construction uses k 2 +2k−3 tiles. Even the observation that a similar pattern appeared on the Sunshine Coast Airport floor reinforces the larger theme that mathematical ideas can emerge from visual structures rather than purely symbolic manipulation.
The harder half is proving optimality, and this is where the presentation becomes unusually good at exposing the difference between following a proof and understanding why somebody might discover it. A first attempt at associating uncovered-square edges with tiles yields only the weaker lower bound k 2 −1. Instead of hiding that failure, the explanation uses its weakness to motivate a better construction: divide the grid into four directional regions using increasing and decreasing paths, then highlight edges according to the region in which each uncovered square lies. The crucial property—that no tile can touch more than one highlighted edge—is explained visually and conceptually rather than simply asserted.
Turning the locations of the uncovered squares into a permutation is the decisive abstraction. The two paths become a longest increasing subsequence and a longest decreasing subsequence, reducing the geometric problem to a statement about their lengths. From there, the arithmetic mean-geometric mean inequality and the Erdős–Szekeres theorem supply the necessary bound. Better still, the latter is not merely cited; its underlying argument is reconstructed by labeling permutation elements with the lengths of increasing and decreasing subsequences ending there, showing those pairs must be distinct, and fitting them inside a rectangular lattice. It is a dense sequence of ideas, but the progression gives each abstraction a reason to exist.
There are moments when the explanation strains under its own length. The repeated reminders that the problem is exceptionally difficult, along with several pauses to restate where the argument stands, occasionally slow an already substantial journey. The recruiting interlude also arrives immediately before the hardest portion of the proof and interrupts the concentration the presentation has carefully built. Yet the eventual summary is valuable, particularly because it restores a small nuance about intersecting paths that had previously been glossed over.
The closing discussion gives the mathematics a broader purpose without reducing the piece to a technology story. The claim that publicly available reasoning models could solve all six IMO problems by 2026, along with descriptions of earlier model performance, is presented as part of the narrator's account rather than independently substantiated here. More interesting is the distinction between possessing a proof and constructing a “motivated explanation” that makes its ingredients feel discoverable. The proposal that such explanations deserve greater academic recognition is explicitly an argument rather than an established standard, but it follows coherently from everything preceding it: the presentation has spent most of its runtime demonstrating precisely how much intellectual work can exist between knowing that a proof works and understanding why anyone would think of it.
Pros
- Builds the solution progressively from concrete tilings to increasingly powerful abstractions instead of simply presenting the finished proof.
- The cube-cutting analogy gives a memorable motivation for the correspondence argument used later.
- Visual intuition, symmetry, permutations and subsequences are connected into a remarkably coherent mathematical narrative.
- Explains the reasoning behind the Erdős–Szekeres result rather than relying solely on its name.
- The distinction between a valid proof and a motivated explanation provides a thoughtful framework for the closing discussion about mathematics and machine-generated proofs.
Cons
- Repeated reminders about the problem’s difficulty and several extended recaps make an already demanding explanation longer than necessary.
- The recruiting segment disrupts the flow at an especially concentration-intensive point.
- Claims about rapidly changing machine performance in mathematics are largely contextual assertions here rather than independently demonstrated evidence.
- One technical nuance involving whether the two paths intersect is postponed until the final recap instead of being incorporated when the lower bound is first developed.
The combination of geometric experimentation, combinatorial abstraction and careful motivation turns an exceptionally difficult Olympiad problem into a surprisingly approachable exploration of how mathematical ideas can be discovered. More importantly, the presentation succeeds at its larger goal of showing why understanding the path toward a proof can be intellectually richer than simply possessing the proof itself.
