Braess' paradox isn't really a paradox, it's just a counterintuitive graph. Notice the edge connecting the top and bottom paths — it has a cost of 0 to travel on. Should reduce overall traffic time, right? Well, watch what happens as the agents reach equilibrium.
(And yes, this DOES reflect real life: some studies have supported lane closures decreasing traffic congestion!)
Stats Box
Pigou's example is a simple graph with one constant-cost edge and one variable-cost edge. Here, the variable-cost edge has a linear cost, but some variants have polynomial and even exponential costs to serve as thought experiments.
Stats Box
Here's a custom graph adapted from here. Can you find the Nash flow?
Stats Box