Self-driving cars – the sweet, sad sound of bursting bubbles

Balloons burst singly, with a satisfying bang, and leave shreds of their existence. Bursting bubbles, like foam on the water, are unnoticed, scarcely audible, leaving just a drop of water, a puff of air, and a small stain. Amid the reports and press releases and shedloads of money, it is starting to be noticeable that there is a rising tide of doubt about connected autonomous vehicles (CAVs). The doubts are raised particularly, but not only, by researchers based in behavioural studies, policy analysis, and transport futures. 

This is triggered by a thoughtful recent paper from the University of the West of England, by Graham Parkhurst and Glenn Lyons [1], who discuss the way “Political enthusiasm, advocacy from technologists, and media hype can portray a sense of inevitability for citizens themselves: ‘a CAV future is coming’”. They carefully unpack the assumptions, appraisals and projections underpinning this ‘inevitability’, and find the substance wanting. They cite an interesting observation from Graham Currie of Monash University who criticises “the unscrupulous use of the word ‘sharing’ by technologists to imply that new mobility modes (including CAVs) are good”, a concern also reflected in the powerful report All Change? by Greg Marsden and his colleagues on the Commission on Travel Demand. And the late John Urry, who suggested that among other behavioural obstacles to driverless cars are the “pleasures… especially for many male drivers, of driving too fast and aggressively”. 

This is not out of the blue. A 2016 report [2] by Zia Wadud and Jillian Anable at the University of Leeds, published with the Institution of Mechanical Engineers and the Low Carbon Vehicle Partnership, emphasised that carbon, mobility and other benefits were not guaranteed, behavioural responses were uncertain, and that there were therefore critical policy conditions on demand management, system specification, technologies and regulation. These were almost self-evidently true, much repeated, and in practice continuously ignored. And it’s not only the academics. The motor insurance industry is increasingly nervous about even using the label ‘autonomous’ because of its ambiguity about responsibility, and Thatcham Research, one of its advisors, has suggested not going ahead with partial automation, but ‘wait until they are ready’ for full automation, a timescale where there are significant divisions in the industry. Greg Marsden points out that we might well have fully automated forklift trucks in Amazon sheds before we manage lane keeping only on motorways. And in political discussion, Christian Wolmar, writing in the staunchly conservative Spectator, suggests a time scale of ‘never’. Search the BBC News app for ‘driverless cars’ to see the changing tone of coverage. 

Why is this shift of expectations happening? I’d suggest it was in part due to a fundamental flaw in the brilliant, universally used, typology for degrees of autonomy, devised in 2014 as part of an exercise in vehicle standardisation. It defined the now famous six-part classification ranging from 0 to 5, with levels of vehicle autonomy ranging from none to complete, summarised as none, hands-on, hands-off, eyes-off, mind-off, and steering wheel optional. (Steering wheel illegal was not included, but may have been in an early draft, replaced by the more benign ‘optional’).

This typology is a logical classification because it spans zero to everything, so captures every important possible different degree of autonomy that could be achieved by vehicles. As such it is nearly faultless. In the same way one might imagine a biological classification system in which every living thing was described successfully by the degree of height it could reach, from the bottom of the sea to high in the sky, by fins, legs, wings, etc. 

But while it is (broadly) true that life first evolved in the oceans, reached the land and then the sky, the reality of sequence was that flying reptiles preceded land mammals, and – crucially – the outcome is that fish, reptiles, insects, mammals and birds all coexist simultaneously, and for staggeringly long periods, in a complex web of interacting ecological niches. Similarly, hand saws and power saws can still be bought in every tool shop. The mistake was to see the six levels as six sequential stages, through which vehicles would evolve, from the inferior to the superior. Technically that makes some sense – calculators were necessarily developed before programmable computers. But in terms of human society it makes no sense at all. The classification does not define a trajectory.

Armed with that view we start to see a different set of issues, which are not about the way in which society steadily and monotonically adapts to vehicles of increasing autonomy, each displacing the previous less advanced form. Instead, they suggest a long lasting – we might as well say permanent – interaction of vehicles of different degrees of autonomy and people with different degrees of response. Then the trajectory, barriers and policy requirements are quite different.

Here’s one example, that I owe to Tom Worsley, formerly a lifelong DfT official and now a Visiting Fellow at the Institute for Transport Studies, University of Leeds. As a keen cyclist and pedestrian, he asks what would pedestrians do if they knew, with certainty, that vehicles would stop for them. They would just walk across wherever and whenever it was most convenient. So truly safe autonomous vehicles would become the self-defeating device that restores the balance of power of pedestrians over vehicles, makes zebra crossings redundant, improves the quality of street life, and makes sitting in the stop-start conditions of autonomous vehicles themselves intolerable. 

The industry doesn’t talk about this much, but it knows it, sort of.  One hears grim whispers of algorithms for autonomous aggressive driving, to make such impolite pedestrians feel as unsafe as possible while not actually being harmed. Or painting autonomous cars in nature’s warning signs, such as black and yellow stripes. But then think of John Urry’s aggressive young men, who as pedestrians devise a new version of James Dean’s game of chicken. This fantasy of course generates its own solution. Initially, there would be demands for the reintroduction of safety barriers, protecting safe vehicles from unruly pedestrians. From there, as Jillian Anable suggested in 2016, there would be guard rails everywhere, as well as disruptive streetscapes and interference with cycling and walking due to having to have so many more drop off/pick up spaces everywhere. Is that the end of the ‘place’ advantages of streets? The end of nearly half a century of rediscovering the joys of traffic-free town centres? The logic leads to fenced or tunnelled roads, with no pedestrian access, between towns, accompanied by strictly regulated and corralled arrangements for interchange, especially between vehicles and walking, in the areas of dense activities in town centres and residential areas. 

But we already have modes of transport that do exactly this. They are called trains. What’s the point of reinventing them, at huge expense in investment, money, resources and energy, and the loss of civic amenity and social welfare? 

Is this all fantasy? Well, yes and no.  Yes, it’s an extreme case that could not actually develop to the limit of its own logic. But no, the concept is valid about the dynamic interaction of people and vehicles of different degrees of autonomy, and of the vehicles with each other. This is not a temporary blip in the smooth penetration of a new technology, but inherent to the social complexity of traffic in streets. It is long lasting, and there is no mechanism that somehow guarantees that a ‘superior technology’ is necessarily a successful technology, even if it is backed by governments, industries, public opinion, vast budgets and all the hype in the world. Think of Concorde. 

As I write, an old song is wafting through the open windows. 

I’m forever blowing bubbles

Pretty bubbles in the air

They fly so high

Nearly reach the sky

Then, like my dreams, they fade, and die. 

References

1. http://tinyurl.com/yatkv898 

2. http://tinyurl.com/y8rb52ux

---

Phil Goodwin is emeritus professor of transport policy at both the Centre for Transport and Society, University of the West of England, Bristol, and University College London. Email: philinelh@yahoo.com