What a 160-year-old theory about coal predicts about our self-driving future

The article argues that even if autonomous vehicles eventually become safer and cleaner than human-driven cars, that alone would not guarantee lower total emissions or fewer crash deaths. By making driving easier and more pleasant, AVs could encourage more trips, longer trips, and a shift away from transit, biking, or staying home.

Reading notes

#

• The article says self-driving cars are gaining momentum again after a difficult 2023, with Cruise rebooting under new management, Waymo expanding, and Tesla promising a robotaxi service.
• It frames the usual AV pitch as safer roads and lower emissions, especially in a country where transportation is a major source of greenhouse gases and crash deaths are high.
• It uses the Jevons paradox to argue that efficiency gains can increase total resource use instead of reducing it.
• It explains Jevons through coal and steam engines in 19th-century Britain, where better engines made coal use rise rather than fall.
• It applies the same logic to transportation, saying easier driving can lower the “cost” of using a car and therefore increase traffic.
• It says AV advocates claim self-driving cars will reduce safety risks and energy use by avoiding human errors and smoothing traffic.
• It adds that AV computers, hardware, and sensors also require substantial power and can introduce their own errors.
• It says AVs may still become safer and cleaner over time, but total crash deaths and emissions may not fall because people will use them more.
• It says AVs are designed to make travel easier and more pleasant, which can lead people to take longer trips or choose cars over other options.
• It cites a Transportation Research Board paper saying the likelihood of additional trips increases when autonomous vehicles are available, even when they are shared.
• It says second-order effects on land use could make matters worse if people move to larger homes on the urban fringe, where driving distances are longer.
• It concludes that the key question is not only how efficient AVs become, but how much extra driving they induce.