Via Brad Plumer: a traffic jam in in a bottle.

To me, it’s pretty remarkable how closely the real-world experiment above matches up with this java-based computer traffic simulator.

Reader support helps sustain our work. Donate today to keep our climate news free.

Warning: if you click the last link, and you’re at all geeky, prepare to lose your afternoon!

Grist thanks its sponsors. Become one.

A few years back I wasted hour after hour playing with the java settings, and watching "traffic" jams materialize and melt — just like in real life. My favorite quirk: for one lane-narrowing scenario, I could make traffic flow along beautifully at 40 miles per hour, but seize up like glue at either 20 mph or 60 mph. Another fave (and very relevant to congestion pricing debates) was letting traffic flow along smoothly at, say 1,400 "cars" per hour, and then increasing traffic volumes to 1,500 — and watching the traffic jam crystallize within moments.

Obviously, a java-based traffic simulator, or even a controlled experiment like the one in the video above, can’t perfectly replicate real driving. But it’s interesting to consider how sensitive traffic flow is to subtle changes. A seemingly negligible increase in traffic volumes, a tiny obstacle, a driver’s hesitation: any of those can precipitate a phase change, turning free-flowing traffic into viscous muck. But conversely, a seemingly intractable traffic mess can free itself — if you remove an obstacle, smooth out a choke point, or simply reduce the peak traffic volume.