The Feasibility of Amazon Prime Air

Amazon prime air is in the media crosshairs.

About a week ago, Amazon announced plans to begin using drones to deliver packages within thirty minutes of ordering. They have released an ad that depicts a skateboard tool ordered via smartphone being packed into a box clipped onto the bottom of an eight-rotored flying robot, and flown over picturesque countryside to its recipient. All within 30 minutes.

It’s a neat idea, but most serious news outlets are understandably skeptical about the proposal. Reactions have ranged from mockery, to sober dismissal. This Slate article sums up the latter reaction pretty well:

Today’s drones are good at gathering information. Bigger drones are better at this than smaller ones. And only large, expensive drones flown by the U.S. government, are currently any good at delivering physical objects.

If thousands of drones are to fly around delivering packages across cities, they must become orders of magnitude more reliable than they are. Otherwise some will crash every day, and Bezos will have to hire an army of people to drive around, pick up the fallen drones, deliver the packages, and refurbish the drones. To satisfy the FAA, drones makers (and would-be operators) must prove that they are able to avoid airplanes, helicopters, and one another and to handle sudden changes in the weather.

On its surface, the Slate article makes a lot of good points. Current socio-technical regimes around both commercial flight and package delivery are not conducive to flying delivery drones. PrimeAir is incommensurable with current technology, policy and regulations. There are also infrastructural concerns, as shipping warehouses are not configured to launch aircraft, and very few people have miniature helicopter landing pads on their front lawns. Drone delivery in crowded cities where most people live in apartment buildings is a whole other concern. So a lot has to change before PrimeAir is feasible, meaning that we won’t be seeing it for at least a few years to come. 

But the Slate article and others like it make the elementary mistake of failing to consider how technology, regulations, and infrastructure can change. This is a classic mistake the media makes when it encounters new technology: At the turn of the century, cars were seen by many as expensive, unsafe toys that would never have any serious role in transportation. Yet cars had become indispensable in many places by the 1960s. There’s no reason to think that delivery drones could not have the same dramatic gain in feasibility and credibility. So let’s consider the various problems to look at just how insurmountable they are:

Firstly, technology. I’ll take Slate’s word for it when they say that small UAVs have neither the reliability nor the range to be used as Amazon proposes. But do they seriously suggest that drone technology will not improve? Making a reliable flying robot that can easily steer itself around obstacles (including other flying robots), deal with inclement weather, and not break down mid-flight to send a brand new coffee grinder crashing down onto the heads of hapless pedestrians below is a simple technical problem that has a technical solution. And there are quite a lot of people working on multi-rotor drones. If Google can make a reliable self-driving car that has to operate in a highly congested highway environment then surely Amazon can make one that can fly safely in the relatively open skies.

Secondly, policy. Slate is right to point out that the FAA will not allow Amazon’s scheme at present, but this criticism has a lot of problems. Firstly and most importantly, it only considers the US! Many other countries (including China and Canada) allow drones to be used commercially. If Amazon rolls out PrimeAir in these countries and it is successful, then it would be very hard for the FAA to retain its prohibition. And why should they? I can’t imagine that Amazon’s drones would fly at altitudes that would make them a threat to any other form of commercial air traffic. So once the technology is perfected to the point that nobody is at risk of being killed by falling broken drones, there is no reason to think that Amazon can’t use its undoubtedly considerable political muscle to push through some regulatory changes.

The last consideration is infrastructure. The infrastructural problems I mentioned above will probably make drones more expensive than trucks, at least in the short term. But Amazon seems willing to invest in these kinds of risky experiments, and there will be many consumers (myself included) who will be happy to pay a premium delivery charge for the added convenience and novelty of drone delivery. This will allow Amazon to establish a small market niche to hone the PrimeAir system. In the long-run, a worldwide rollout of PrimeAir will allow Amazon to massively downsize their human workforce*, meaning that PrimeAir will be profitable in the long run. So it’s worth their while to experiment.

I know there are a lot of reasons to criticize Amazon, but I’m rooting for PrimeAir because I see major environmental potential in it. Electric drones can eliminate a great deal of the carbon emissions created by delivery vehicles. If we can use drones and the internet to replace that system with small, electric vehicles, then that’s a good thing. If drone delivery allows internet shopping to capture more of the retail market, then that means we also reduce the environmental emissions from many stores, and we make it that much easier to live a car-free lifestyle. Those are all good things.

There’s no guarantee that PrimeAir will succeed, of course. There are a lot of contingencies involved, and it’s entirely possible that the whole thing is just an elaborate publicity stunt. But there’s a broader point here about sustainability: We’re not going to get it by staying in our current technological paradigm. If we’re going to implement lasting change, we need to imagine radical alterations to our current system. Critical thought is an important part of this, but if it lapses into cynicism then we’re not going to get anything done.

*I know that this is problematic, and I’m going to address it in a future post.