“Elon Musk Is Wrong about Artificial Intelligence and the Precautionary Principle” – Reason.com via @nuzzel
(disclaimer: I haven’t dug any deeper than reading the above linked article.)
Apparently Elon Musk is afraid of the potential downsides of artificial intelligence enough to declare it “a rare case where we should be proactive in regulation instead of reactive. By the time we are reactive in AI regulation, it is too late.”
Like literally everything else, AI does have downsides. And, like anything that touches so many areas of our lives, those downsides could be significant (even catastrophic). But the most likely outcome of regulating AI is that people already investing in that space (i.e. Elon Musk) would set the rules of competition in the biggest markets. (A more insidious possible outcome is that those who would use AI for bad would be left alone.) To me this looks like a classic Bootleggers and Baptists story.
Rick Weber has a good note lashing out against net neutrality regulation. The crux of his argument is that there are serious costs to consumers in terms of getting content slower to enforced net neutrality. But even if we ignore his argument, what if regulation isn’t even necessary to preserve the benefits of net neutrality (even though there really never was net neutrality as proponents imagine it to begin with, and it has nothing to do with fast lanes but with how content providers need to go through a few ISPS)? In fact, there is evidence that the “fast lane” model that net neutrality advocates imagine would
In fact, there is evidence that the “fast lane” model that net neutrality advocates imagine would happen in the absence of regulatory intervention is not actually profitable for ISPs to pursue, and has failed in the past. As Timothy Lee wrote for the Cato Institute back in 2008:
The fundamental difficulty with the “fast lane” strategy is that a network owner pursuing such a strategy would be effectively foregoing the enormous value of the unfiltered content and applications that comes “for free” with unfiltered Internet access. The unfiltered internet already offers breathtaking variety of innovative content and application, and there is every reason to expect things to get even better as the availabe bandwidth continues to increase. Those ISPs that continue to provide their users with faster, unfiltered access to the Internet will be able to offer all of this content to their customers, enhancing the value of their pipe at no additional cost to themselves.
In contrast, ISPs that chose not to upgrade their customers’ Internet access but instead devote more bandwidth to a proprietary “walled garden” of affiliated content and applications will have to actively recruit each application or content provider that participates in the “fast lane” program. In fact, this is precisely the strategy that AOL undertook in the 1990s. AOL was initially a propriety online service, charged by the hour, that allowed its users to access AOL-affiliated online content. Over time, AOL gradually made it easier for customers to access content on the Internet so that, by the end of the 1990s, it was viewed as an Internet Service Provider that happened to offer some propriety applications and content as well. The fundamental problem requiring AOL to change was that content available on the Internet grew so rapidly that AOL (and other proprietary services like Compuserve) couldn’t keep up. AOL finally threw in the towel in 2006, announcing that the proprietary services that had once formed the core of its online offerings would become just another ad-supported web-site. A “walled garden/slow lane” strategy has already proven unprofitable in the market place. Regulations prohibiting such a business model would be suprlusage.
It looks like it might be the case that Title II-style regulation is a solution in search of a problem. Add to it the potential for ISPs and large companies to lobby regulators to erect other barriers to entry to stop new competitors, like what happened with telecommunications companies under Title II and railroad companies under the Interstate Commerce Commission, and the drawbacks of pure net neutrality Rick pointed out, and it looks like a really bad policy indeed.
There’s a simple alternative to regulation: liability. We don’t need to tell companies how to be safe if we make them legally responsible for negligence.
It’s as though Mass’s government decided that back-to-school season calls for creating real-life rent seeking examples for my class. They’re going to start taxing ride-sharing customers $0.20 per ride with five cents of that going to the taxi industry.
“The law says the money will help taxi businesses to adopt ‘new technologies and advanced service, safety and operational capabilities’ and to support workforce development.”
New technologies like an app that gets more use out of otherwise idle cars? Or an app that makes it easy to hail a ride with little wait? Or an app that brings supply into harmony with demand when demand surges? Oh wait! We’ve already got that and it’s the thing that’s being taxed!
There are a few important economic lessons that Massachusetts’ electorate is evidently in need of. Let’s start with taxes.
Taxes don’t stick
“Riders and drivers will not see the fee because the law bars companies from charging them.” They won’t see the fee, but that doesn’t mean they won’t pay it. A business only exists by collecting money from customers and paying some portion of that to suppliers. The government cannot tax a business without taxing that business’s customers and suppliers.
Granted, part of the cost will be reflected in lower profits (although profits aren’t as big as people think) which means Uber’s shareholders will face part of the tax. But what does that mean? It means 1) a little less money in pensions, and 2) potential investment capital is moved from the people who gave us the best version of taxi travel to the people who gave us the worst version of it.
Money is fungible and I don’t know how to run a cab company
Safety, new technology, and workforce development all sound good, but taxi companies (at least those that deserve to stay in business) will already be doing these things. Safety is important because accidents are costly (especially if your fleet size is limited by regulation). New technology is being adopted by every other (competitive) industry without government support. Other companies invest in their employees.*
Supporting workforce development is part of a larger trend of people supporting specific fringe benefits without appreciating the tradeoff between monetary and non-monetary compensation. And all these ideas reflect a faulty logic: just because something is good, doesn’t mean we need to force people to do it.
Voters simply aren’t in the right position to know if some good thing is good enough relative to other options. If you go into the backrooms of any industry you aren’t already familiar with you will surely learn about techniques and tools you had no idea existed before. So why should we expect that cab companies need regulators to tell them what to do? Let them learn from their trade magazines.
But there’s good news. If we mandated that cab companies use this new revenue stream to pay for new tires, they wouldn’t simply waste the money by buying superfluous tires. They’d stop buying tires out of their own revenues and start buying them from Uber’s. Telling someone to pay from their left pocket simply leaves more money in their right pocket for everything else.**
Extra money in cab company coffers could allow them to invest in better service, happier employees, “and help so taxi owners could buy ‘flagship’ vehicles like a 1940s Checker or a Porsche.” But cab companies are already free to reinvest their profits if they think doing so would create value (i.e. greater future profits). The more likely outcome is that they will simply have more money than before.
Competition is not the problem, it’s protectionism
When we see problems in the world we need to look for their root causes if we want to actually make things better. More often we act like a doctor diagnosing cancer is the cause of the cancer. Don’t want cancer? Outlaw doctors!
Cab companies aren’t as successful as they previously expected and the apparent culprit is Uber. But they only exist because an inefficiency in the market created a profit opportunity. Cab companies are doing poorly because they don’t provide as much value per dollar. And that’s largely because of regulation that prevents competition. Much of it was put in place specifically to protect incumbents from competition.
A lot of these regulations sound nice enough, but they still created the market niche that Lyft and Uber filled. And they protected cab companies from competition right up until ride-sharing became feasible.
Regulation is not the answer
Let’s give cabbies the benefit of the doubt for a minute. Let’s assume that they aren’t really in it for the cash-grab and that they just want to help people get around safely and conveniently. Let’s even assume that NYC’s medallion system is about congestion rather than competition.
If that’s the case, then there are better ways to address the root causes of the problems cabbies tell us to worry about. We don’t need to address each of these problems individually if we can find a few key causes at the root of each of them.
Cabs have medallions but civilians don’t, so congestion will still be a problem in cities until congestion fees are implemented that balance the demand for road access with its limited supply. Safety is important, but mandating extra inspections for only some types of cars is a half-assed way of dealing with it.
There’s a simple alternative to regulation: liability. We don’t need to tell companies how to be safe if we make them legally responsible for negligence. This is an important lesson for how we think about regulation in all industries. The basic logic is also why economists vastly prefer pollution taxes to specific regulations; it’s usually better to name the outcome we want and create a cost for failure to meet it rather than mandate specific behaviors.
Perhaps this means we should modify the laws that require all drivers to be insured so that some drivers have higher minimum liability coverage. That would be far less invasive and do far more to alleviate the concerns Uber’s critics raise than mandating specific behaviors.
Concentrated benefits dispersed costs
Okay, so maybe this is too small an issue to be concerned with. If that’s not by intentional design, then it at least reflects an evolutionary logic. This policy is likely to survive because the people it taxes will face a cost so small it isn’t worth doing anything about. Yes, Uber and Lyft have incentive to lobby against it, but it’s so close to invisible that they’ll probably be able to pass it almost entirely on to drivers and passengers.
This is going to cost millions… with a tiny little m. At first I read it as a 5% tax and quickly realized that Uber rides are so cheap that I won’t even notice it. And 20 cents a ride is even less than 5%.
So why worry? Precedent. The problem with death by a thousand cuts isn’t any one cut.
*Of course we can argue about whether they do enough of that. There may be a tragedy of the commons if there’s asymmetric information between people looking to make human capital investments and businesses looking to gain access to specific human capital. Such a situation might create an opportunity for government to do some good by investing in public goods or subsidizing on-the-job training. But if that’s the case, it calls for very different programs (education reform, etc.) than taxing successful companies to subsidize their competition.
**Why is this good news? Because if cab companies did change their behavior it would imply they’re doing something where cost exceeds benefit. It would destroy value. Remember those stories of WWII rationing? Imagine that situation but with cab companies buying twice as many tires and just storing extras in the garage. It would clearly be a bad thing. Scarcity isn’t so urgent nowadays, but the basic logic remains the same.
I’m reading Complexity: The Emerging Science at the Edge of Order and Chaos which has the absolute best testimonial on the front cover: “If you liked Chaos, you’ll love Complexity.”
This book was written in 1993 so I’m pretty late to the show, but it’s worth raising the issue: complex systems require governance, but that need not mean government.
In the copy below the author is writing about how complex systems–systems with components that affect one another in simple ways resulting in emergent orders at the system-wide level–occupy an interesting space between chaos and order. Too much order and you end up with something fixed and unchanging. Too much chaos and you’ve got noise.
The second full paragraph misses an important point that should have been obvious to the author and the researchers who he’s paraphrasing. The government is an endogenous element in the wider economy. If we think of the economy as a network of people (individual nodes) who cluster into sub-networks (organizations), the government is just a collection of nodes and clusters that follow different rules than the rest. Granted, these clusters often serve important roles (e.g. courts). But the anarchist branches of economics have pretty clearly demonstrated that removing the state from these roles doesn’t always lead to chaos. Ripping the state out like a band-aid would be an awful idea, but gently scaling (scoping?) back the state need not be a disaster.
This band between chaos and order is wider than they’re giving it credit for. We can only examine this band from our own perspective… as human beings who are tiny components of this much larger network of networks. The range of configurations that could allow a peaceful, flourishing society is essentially infinite. Yes, governance is necessary, but strengthening any particular set of nodes cannot allow for governance of the system as a whole. It can only allow for governance of a sub-set of the wider network.
Emergent orders cannot be controlled from within.
Whoa! Yeah, I’m going to do this, but let me start with some caveats. First, this is an argument, not the argument. Every silver lining has a storm cloud, and acknowledging the silver lining doesn’t mean you’re in favor of tornadoes. Second, I’m being sloppy with the term libertarian; classical liberal is closer to the truth, but doesn’t make for as good a title. Most importantly, I think my argument is swamped by the traditional libertarian arguments against the FDA. All that said, this argument has some interesting implications for how we think about intervention generally. Here goes…
The human body is a complex system that we do not fundamentally understand. Although every complex system is unique, they have similarities. In the case of both the human body and society/markets, interventions lead to unintended consequences which can offset the (ostensible) gains from the intervention. At the end of the day, although the FDA intervenes in the complex system of human society, it also prevents intervention in the complex system of human physiology.
The Hippocratic Oath instructs its speaker to not play God and to avoid over-treatment, and the justification for that is made clear in a recent Econ Talk. The guest was on to promote his book which discusses the problem of medical reversal–the phenomenon of medical practices that are adopted and subsequently abandoned after evidence shows the practice to be ineffective or worse. From this position he argues that the FDA’s mandate to ensure not just safety, but efficacy, is especially important. His argument is that because of the cost of type II error the FDA ought to go further.
Let’s look at two extreme cases. In the “anything-goes” world, we might have a lot of people trying good and bad interventions with a lot of harm being done to the unlucky ones. You and I know that the real problem is one of information and that in a perfect world we would have “anything-goes-that-consumers-with-access-to-good-information-from-Consumer-Reports-®-or-a-competitor” but this world still leaves us with the problem (which we face in today’s FDA-evaluated world) that consumer trial-and-error is a poor substitute for randomized control trials.
At the other extreme we have the “first-do-no-harm-second-do-real-good” world of an ideal FDA. This world has very steep type I errors but instead of two steps forward, one step back, we would have one step forward, then another, and never any steps back…. but of course each step forward would cost a few billion dollars.
Neither extreme is ideal, but the second world is one where standards of evidence are taken very seriously. In that world I’d be a third grade teacher instead of a college professor. The standards of evidence are at the core of the problem of medical reversal, but also the problem of economic intervention (which is far less likely to be reversed, even in the face of good evidence indicating that it should be).
As far as medical intervention is concerned, my position is bullish on better efficacy evaluation of medical procedures but still bearish on the FDA itself. But looking at the FDA from this angle opens up an interesting thought experiment: what might be the effects of an Economic Intervention Standards Authority? In practice it would probably be awful (my guess is a federal bureau that attempts to quash Tiebout competition), but in a libertarian utopia it would be the bureaucracy that libertarian kids with administrative bents would dream of heading.