Braess' Paradox is a counterintuitive result in traffic and network science: sometimes adding a new road or connection to a network actually makes overall travel times worse, not better. This happens because individual travelers each choose the route that minimizes their own travel time, and their collective selfish decisions can lead to a worse outcome for everyone. The paper studies this paradox in a specific, clean setting: rectangular grid networks where roads follow a simple linear relationship between congestion and travel time, and a single new "shortcut" edge (called a chord) is added somewhere in the grid.
The authors use a clever analogy between traffic flow in these networks and the behavior of electrical circuits, which lets them calculate how much travel time changes when a chord is added. They work out exactly which chords can potentially trigger the paradox (depending on where in the grid they are placed) and find that such chords are actually quite rare. The fraction of paradox-inducing chords is highest in grids with roughly a 2-to-1 aspect ratio, and the results are computed across all grid sizes up to 100 by 100.
A key quantity researchers care about is the "Braess Ratio," which measures the worst-case factor by which total travel time can worsen after adding a bad edge. A famous prior result showed this ratio cannot exceed 4/3 in general affine networks. This paper tightens that bound for grids specifically, showing the ratio approaches about 1.207 in square grids and only reaches 4/3 in very elongated, thin grids. The authors also prove that any chord achieving the absolute worst-case ratio must itself have zero travel time cost, meaning free shortcuts are paradoxically the most dangerous additions to a network.