1. 程式人生 > >The Wave Of Attribution Is Coming To Car Culture

The Wave Of Attribution Is Coming To Car Culture

The Wave Of Attribution Is Coming To Car Culture

The advent of software has brought about an explosion of applicable knowledge across industries, upending every gut-driven business model in its path. Healthcare, finance, advertising, retail, entertainment… each of them now feeds on a steady diet of data in order to attribute their decisions: to find cause and effect in everything they do.

No sector has completely escaped the digital wrath of attribution, but automobiles have proven to be among the most elusive. Drivers still pay a pittance for land use. Employers still have zero accountability for the financial and environmental burden car ownership places on a region. Collisions still present such a mystery that

30% of them only exist on the honor system.

For all their claims of innovation, carmakers have kept vehicles and owners isolated from any semblance of attribution. Connected cars will shatter that blissful ignorance, and as V2X is crucial to the advancement of self-driving tech and shared access, connectivity may prove to be the last great automotive feature.

Take March’s fatal Uber crash. Tragic as roadway fatalities are, they make for extremely poor attribution models: a car of a different shape, or a pedestrian with a different reaction time, or even a city with different emergency response operations could have meant the difference between a fatality and an injury, none of which are factors that should alter the assessment of Uber’s technology or its driver.

Equal consequences, tied to behavior rather than outcome. Footage via Tempe Police Department

The wealth of data provided by advanced vehicles and traffic devices empowers us to attribute dangerous driving not to the fatal collision, but to the negligence, miscalculation, and environmental factors leading up to the collision. The upshot of such knowledge exposes millions of “phantom collisions” among drivers — a more relevant model for identifying and policing dangerous driving, even in situations where no crash occurs.

That is to say, human drivers are committing egregious, life-threatening driving errors every single day, and society will soon be able to hold them accountable despite the absence of a collision outcome. This revolution in attribution will not only give self-driving tech the more frequent comparison point it needs to prove safety superiority, but will also serve as a disincentive for human operation in a market where the vehicles they buy increasingly entrap them with a combination of greater autonomy and stricter surveillance.

RAND’s oft-cited report on the feasibility of quantifying comparative AV safety using conventional crash metrics: https://www.rand.org/pubs/research_reports/RR1478.html
RAND’s latest work, identifying “infractions” and “roadmanship” as additional measures of comparitive safety performance: https://www.rand.org/pubs/research_reports/RR2662.html

Beyond the act of driving itself, the broader proliferation of behavioral data will eventually extend deeper into legislation and law enforcement, as attributable inputs empower authorities to intervene without the need to wait for the confirmation of some unfavorable outcome. Private ownership’s veil of anonymity can no longer stem the tide of attribution.