(“Better is the enemy of good”, Voltaire)

Google and the nirvana fallacy

Google’s self-driving vehicle project appears to be in serious trouble, apparently as a result of something closely related to the nirvana fallacy, the idea that one must reach the perfect solution, which in this case is the total elimination of human involvement in driving.

From the outset, Google’s project was about developing a vehicle in which humans were seen as the source of all faults and problems, the weak link to be eliminated, to the point of making it impossible for there to be any involvement by eliminating the controls needed to drive. Meanwhile, its competitors have been focusing on the development of progressive developments to reduce driver involvement under certain circumstances, all related to specific business models.

The result is that Google, which was originally perceived as a pioneer in the field of autonomous driving, now seems to be losing ground to a series of rivals who every so often announce a new development: the first completely autonomous taxis developed in Singapore does not bear the stamp of Google, but of NuTonomy, a small startup incubated at MIT.

But in addition to this project, which it could be argued is an isolated case, other companies seem to be making the headlines: Volvo keeps announcing new developments, such as its association with Uber in Pittsburgh, a city that has been able to establish itself as a technological showcase.

If the announcements are to be believed, Ford sees its future as a service provider. Tesla has recovered from the fatal accident a few months ago involving one of its vehicles by presenting a new version of its Autopilot which makes much more intensive use of the radar. Uber is exploring all possible alternatives and besides trying out self-driving taxis in Pittsburgh with Volvo, is launching a fully electric vehicles initiative in London, while buying Otto to develop a self-driving truck to transport goods, which will probably be the first truly autonomous vehicles we’ll see on our roads.

The approaches taken by other competitors are markedly pragmatic: anything goes if it means advancing the development of the elements needed in a future business model. Tesla prefers to talk about self-driving cars as an advance from assisted driving, in the same way that Uber’s pilot project with Volvo has a driver “just in case” and another person taking notes, transporting passengers free of charge. In both cases, the point is to obtain data, maximize learning, to educate their algorithms. To do that, as with the nirvana fallacy, they do need not to have the best: good enough will do.

The difference between these initiatives and Google’s seems to be, aside from their pragmatism, is the business model behind them. In the case of Google and its radical approach, everything is unknown: it seems, as with Android, to be create a platform that we can put in the hands of manufacturers willing to incorporate it into their vehicles. Google has a fleet deployed in northern California circulating under normal conditions, albeit subject to certain restrictions such as the need for a human driver and a speed limit, the potential of which is still unknown: could end up being part of a fleet of robotaxis that is considering for San Francisco using Waze, or part of a project developed with Fiat Chrysler vans, or anything else, but there doesn’t seem to be any strategic direction clearly marked.

Other competitors certainly have it clearer. Tesla wants to sell cars that work for us when not in use. Uber wants to monopolize public transport city by city, based on business models that claim a fleet of taxibots would eliminate the need for 90% of private cars and drastically alter the appearance of cities.

For Google, once a pioneer, to now have problems due to a lack of competitive focus and an excess of ambition is hardly surprising for a company that sets out to solve issues it considers important, true engineering challenges, but treating the business model as a secondary issue, along the lines of “if we solve it, money will come”. The real buzz however, comes from solving the problem. In many cases, the engineers in these projects end up being frustrated and leaving the company to look for greener pastures where they can become true entrepreneurs committed to a genuine business model and a set of real life goals. Reaching nirvana is probably great, but not everybody has the required patience to do so…

This is something we discussed a few days ago, and it isn’t alone: it appears that Apple, a company known for its obsession with secrecy and opacity, is also rethinking its self-driving vehicles strategy, and may even be thinking about pulling out of the race completely.

As the industry begins to establish self-driving vehicle testing procedures, it becomes clear that rather than simply solve problems, some companies have business models for making money from self-driving vehicles.

We must now prepare to share the streets and roads with vehicles with nobody behind the wheel, seeing the autonomous vehicle as an increasingly common part of our daily lives. What is less clear is which make this a reality.


(En español, aquí)