Here is a key passage from Tyler Cowen’s 2011 piece on driverless cars that applies to a lot more than driverless cars:
The point is not that such cars could be on the road in large numbers tomorrow, but that we ought to give the cars — and other potential innovations — a fair shot so that a prototype can become a commercial product someday. Michael Mandel, an economist with the Progressive Policy Institute, compares government regulation of innovation to the accumulation of pebbles in a stream. At some point too many pebbles block off the water flow, yet no single pebble is to blame for the slowdown. Right now the pebbles are limiting investment in future innovation.
The lesson here is the one I emphasized in my post yesterday, “Clay Christensen, Jerome Grossman and Jason Hwang on the Agenda for the Transformation of Health Care”: allowing experimentation with innovations that at first seem like lower quality ways to do things (except for their cost and convenience) is crucial to many of the economic transformations that will do the most to improve overall standards of living. Needing a path through what seems at first like lower quality is exactly what Clay Christensen means when he says that an innovation is “disruptive.” Driverless cars provide a wonderful example. Ultimately, driverless cars will be much safer than human-driven cars, since it is unlikely that the overall skill and care of human drivers will dramatically improve from where it is now, while computers and sensors for cars can continue to get better and better and better. But we will get to those driverless cars that dominate human-driven cars in all respects (except for those who find driving recreational) if right now we allow driverless cars on the road that are better than human-driven cars in some respects and worse (within reason) in other respects.
I am saying that, because they are likely to ultimately be much safer that driverless cars should be allowed even if at first they are somewhat less safe, but in fact the relevant situation is more like this. At some point driverless cars will have a good safety performance in small-scale tests, but there will be some uncertainty about how they will do in substantial numbers in real world situations on the road. Even if at that point they would in fact have a better safety record if allowed on the road, opponents will argue that the uncertainty about how they will do means they should be banned. Such a ban–and its counterparts in other domains–are a very effective method to slow down technological progress.