In Seoul and Ann Arbor, far from the limelight of Silicon Valley, a 31-year-old mathematician has patiently settled a question that has tormented experts for nearly six decades: what is the largest rigid shape you can push around a right-angled corridor without lifting it off the floor or bending it?
The sofa riddle that outlived generations of mathematicians
The story begins in 1966, when Austro‑Canadian mathematician Leo Moser formulated what sounds almost like a party trick. Picture an L‑shaped hallway, each arm exactly one metre wide. Now imagine a flat, rigid “sofa” that must be slid and rotated through the bend, always staying on the floor, never squeezed or folded.
The question Moser posed was simple to state and agonising to solve: what is the maximum area such a sofa can have?
The “moving sofa problem” asks for the largest possible area of a shape that can navigate a right‑angled one‑metre‑wide corridor without deformation.
Very quickly, the problem escaped academic journals and slipped into puzzle books and university courses. It became a kind of rite of passage for geometers and enthusiasts of applied mathematics, thanks to a mix of elegance and stubborn difficulty.
Early contenders: from rectangles to wildly curved shapes
You might first try obvious answers: a rectangle, a semicircle, a cross shape. All can be pushed around the corner, but they waste space at key points of the turn.
By 1968, British mathematician John Hammersley had produced a far better candidate. His quirky, partly rounded sofa shape reached an area of roughly 2.2074 square metres. For a while, this looked impressive.
Then in 1992, American mathematician Joseph Gerver pushed things further. He introduced a shape with many carefully calculated curves—so intricate that it is defined by multiple analytic arcs, not just a simple formula. Gerver’s sofa hit about 2.2195 square metres, becoming the new record holder.
Yet a nagging question remained. Was Gerver’s shape truly the best possible, or just a clever near miss?
- Hammersley (1968): about 2.2074 m²
- Gerver (1992): about 2.2195 m²
- Unknown: absolute theoretical maximum, until 2024
Over time, researchers leaned on computer simulations to tweak candidate shapes. Algorithms could nudge boundaries, adjust curves, and search through thousands of variants. The consensus grew that Gerver’s design was probably optimal—but no one could actually prove that nothing larger could possibly fit.
A young conscript meets a legendary problem
Enter Baek Jin‑eon, a South Korean mathematician who first met the moving sofa problem during his mandatory military service. Posted to the National Institute for Mathematical Sciences, he came across the puzzle almost by chance.
What caught his attention was not its fame, but its emptiness. Behind the colourful diagrams and guesses, there was barely any solid theory. The problem had been poked and prodded, but never really framed in a clean, systematic way.
Baek was struck less by the difficulty of the moving sofa problem than by the lack of a clear conceptual framework around it.
That absence turned into his motivation. He started to build the missing framework from scratch: definitions, constraints, and a way to encode every possible motion of a shape through the corridor.
Seven years, 119 pages, zero computer simulations
Baek’s work continued during his PhD at the University of Michigan and later at the June E. Huh Center for Mathematical Challenges in Seoul. While many modern proofs in geometry and optimisation lean on code, he made a radical choice: no numerical optimisation, no simulations, not even dynamic geometry software.
Instead, he set out to convert the sofa puzzle into a rigorous optimisation problem, where every possible path and orientation of the sofa could, in principle, be described and compared using pure reasoning.
After seven years, the result appeared at the end of 2024 in a preprint on the scientific repository arXiv: a 119‑page proof that locks the problem shut.
Baek proves that Gerver’s 1992 shape is not just good—it is mathematically optimal. No larger rigid shape can pass through the one‑metre L‑shaped corridor.
That statement settles the central mystery: the moving sofa constant—the maximum possible area—matches Gerver’s value. Decades of clever guesses, now backed by a fully detailed, pen‑and‑paper proof.
A win for pencil‑and‑paper thinking in the age of AI
Baek’s approach has drawn attention because it runs against the current. In an era dominated by machine‑assisted proofs and brute‑force search, here is a major result obtained by classical means.
The proof’s core is a new formalisation of the problem. Baek recasts the sofa’s motion as a path in a high‑dimensional configuration space: each point encodes both the sofa’s position and its orientation in the corridor. Within that space, the movement constraints become inequalities, and the task becomes a clean optimisation question.
From there, he uses geometric inequalities and careful case analysis to whittle down the possibilities. Piece by piece, he rules out any shape that would outgrow Gerver’s design while still satisfying all the movement constraints.
The work is currently under review at the prestigious Annals of Mathematics. If accepted, it will join a small set of landmark papers that closed long‑standing open problems with detailed, concept‑driven arguments rather than machine calculations.
Why such a “silly” question matters to serious science
At first glance, the moving sofa problem sounds whimsical. No one truly needs an optimally large couch to move through a corridor. Yet questions like this play a specific role in mathematics and engineering.
They force researchers to confront the limits of shape optimisation under rigid constraints, a theme with echoes in robotics, logistics and even medical imaging.
| Domain | Link with the sofa problem |
|---|---|
| Robotics | Planning paths for robot arms or drones through tight spaces |
| Manufacturing | Designing parts that must be transported through constrained factory layouts |
| Architecture | Understanding clearances for moving large objects in buildings |
| Computer graphics | Collision detection and motion of rigid bodies in virtual environments |
In all these areas, the same ingredients appear: rigid shapes, tight corridors, and the need to move without collision or deformation. The toy problem strips these issues to their bare mathematical bones.
Key ideas behind the puzzle, decoded
For non‑specialists, a few terms help make sense of what Baek has achieved.
- Rigid motion: The sofa can translate and rotate, but it cannot change shape. No stretching or squeezing.
- Configuration space: Instead of tracking a sofa in a corridor, mathematicians track a point in an abstract space that records all possible positions and orientations.
- Optimisation: Among all shapes and all valid motions through the corridor, the aim is to maximise the area.
These concepts show up in self‑driving car software, automated warehouses, and even video game engines, which must constantly decide if a moving object collides with its environment.
What this means for future mathematical puzzles
Baek has often described his research process as a cycle of hope and collapse: an idea seems promising, then breaks apart, forcing him to rebuild from the fragments. That experience mirrors the way long‑standing problems evolve across generations.
With the moving sofa problem now settled, attention is shifting to variations. One famous twist is the “piano movers problem”: how to move a more complicated shape, like a long object or a non‑convex piano, through a maze of obstacles. Another variant changes the corridor width or allows the shape to be flexible, raising new questions about trade‑offs between rigidity and adaptability.
There is also a broader lesson for young researchers. The moving sofa story shows how a simple, almost playful question can lead to new techniques, new frameworks and, eventually, a definitive answer that many assumed might never come.
For anyone fascinated by puzzles, an easy way to feel this difficulty firsthand is to sketch your own “sofa” on graph paper, mark a one‑metre‑wide L‑shaped corridor, and try to imagine every twist and slide needed to get it through without overlap. That mental juggling act—the need to picture shapes and motions at once—is exactly what Baek captured and tamed with careful, abstract reasoning.