The Braess Paradox: When a New Road Makes Traffic Worse
Braess's Paradox shows that when each driver independently picks the fastest-looking route, the network settles into an equilibrium that is worse for everyone than the original network — and removing a road can paradoxically help.
A complete interactive classroom, not just a preview.
Start when you are ready to enter this Stage's 9 scenes and explore, respond, and learn as you go.
Why can adding a new road to a road network make every driver's travel time longer?
City planners add a new road to relieve congestion, yet every driver's commute gets longer. This actually happens — and math proves it.
Adding capacity should help, not hurt. Something counterintuitive is hiding in how individual drivers choose their own fastest routes.
An interactive network simulation where the learner adds and removes a shortcut road and watches travel times rise and fall for every driver simultaneously.
A new road can make everyone worse off because selfish route choices push traffic into a Nash equilibrium that is not the social optimum — so adding a selfish shortcut widens the gap.
More roads mean more options, and more options should reduce congestion, so adding a road should never make traffic worse.
- Real-world city case studies beyond the classic toy network
- Detailed game-theoretic proofs beyond the intuition of Nash vs. optimum
- Tolls, pricing, or policy remedies (other than as a one-line resolution nod)
- 01The City That Built a Road and Got SlowerslideQuestion
Open with a concrete story: a city adds a shortcut, drivers celebrate, and then commute times rise for everyone. Frame the driving question before any answer is revealed.
- Cities routinely add roads to fight congestion
- In Braess's Paradox, the new road backfires for every driver
- The puzzle: why should adding capacity ever hurt?
- 02Try It Yourself: Will the Shortcut Help?interactivePrediction
Learner manipulates a small two-route network: choose travel demand, then predict whether adding a shortcut will shorten or lengthen average travel time before running the simulation.
- Adjust the number of drivers on the network
- Predict the outcome before the network is solved
- Commit to a guess: faster, slower, or unchanged
- 03Watch the Paradox in ActioninteractiveEvidence
Run the network in real time: cars enter at S, choose routes through A or B, and converge on a self-selected equilibrium. Toggle the shortcut on and off and read off average travel time.
- See drivers redistribute after the shortcut opens
- Compare equilibrium travel time with shortcut vs. without
- Notice which links become bottlenecks when the shortcut is added
- 04The Numbers Don't LieslideEvidence
Display concrete numbers from the classic Braess network: 4000 drivers, each link with a linear latency function, and the resulting equilibrium travel times with and without the shortcut.
- Without shortcut: every driver takes about 65 minutes
- With shortcut: every driver takes about 80 minutes
- The shortcut makes the equilibrium strictly worse for all
- 05Why Selfish Choices BackfireinteractiveExplanation
Interactive diagram: drag drivers between routes and watch link latencies update. Learners feel how each individual switch raises the latency of the shared bottleneck until no driver wants to move.
- Each link has a latency that grows with how many cars use it
- Switching to the shortcut loads the two shared links it connects
- The new equilibrium is stable — no driver can improve alone — but worse for everyone
- 06Nash Equilibrium Is Not the Social OptimumslideExplanation
Explain the gap between individual rationality and collective good using simple latency curves, without heavy game theory notation.
- Social optimum: minimize total travel time across all drivers
- Nash equilibrium: no single driver can switch and do better
- The shortcut moves the equilibrium further from the optimum
- 07Transfer: Close the Shortcut on PurposeinteractiveTransfer
Learner tests the reverse move: starting from the paradoxical equilibrium, they close the shortcut and watch travel time fall. Reinforces that removing a road can help.
- Apply the insight in reverse: sometimes less road is better
- Observe the system relaxing to a better equilibrium
- Generalize: any selfish-shortcut link can be a candidate for closure
- 08When Does the Paradox Show Up in Reality?slideBoundary
Be honest about limits: the paradox needs specific latency shapes and selfish routing. Real networks are messier, but documented cases exist in Seoul, Stuttgart, and other cities.
- Classic result assumes every driver picks the fastest current route
- Real traffic mixes habits, information limits, and signal timing
- Still, several cities have closed roads and measured improvements
- 09Answer: Why a New Road Made Things WorseslideResolution
Directly answer the driving question and resolve the opening tension with a one-line takeaway and a nod to pricing as a remedy.
- Adding a selfish shortcut pushes drivers into a worse Nash equilibrium
- Individual incentives and collective welfare can pull in opposite directions
- Tolls or coordination can align the two — sometimes closing the road is the fix
Discussion threads for a Stage aren't available yet.
