Our previous post raised the interesting idea of a benevolent dictator, possibly by the name of Watson, to guide us all to greater prosperity. Despite almost instinctive dislike of the notion of such an arrangement, it certainly does have its merits, and it’s a fascinating discussion to be had.
As a (somewhat) educated, self-described liberal, I like to take a first crack at most issues economic and political by asking myself: what would this do for the (admittedly vague) notion of personal liberties? The interesting thing is that, even starting with a “traditional”, human dictator, the answer to that question in this case is far from clear. It is telling, for example, that two of the staunchest defenders of liberalism, Friedrich August Hayek and Milton Friedman, were openly sympathetic to some dictatorial regimes such as that of Pinochet that came to power in Chile in 1973. Sure, their support in this case came mainly as a response to the policies implemented by Allende, who ran the Chilean economy into the ground and trampled economic freedoms considered the central safeguard needed to uphold all other freedoms. Still, let’s see what they had to say in their defense.
Hayek argued that democracy lived out to its fullest cannot work because it is riddled with deep internal contradictions, creating forces that end up destroying itself. It is certainly nothing totally unreasonable and definitely worth thinking about. It is important to note, however, that he did not support dictatorships as a long-term institution, but rather as a transitional system to reverse the damage done by the inconsistencies of democracy and collectivist policies. A sort of “reset” button once “the voice of the people” came back to haunt them. In his words,
“a dictatorship may be a necessary system for a transitional period. At times it is necessary for a country to have, for a time, some form or other of dictatorial power. As you will understand, it is possible for a dictator to govern in a liberal way. And it is also possible for a democracy to govern with a total lack of liberalism. Personally, I prefer a liberal dictator to democratic government lacking in liberalism.”
While no amount of progress on whatever front can justify the human rights violations, political persecution and other atrocities committed by Pinochet, he did peacefully hand over power to a democratically elected president in 1990. And while the Chilean economy faces massive problems today, it is the second richest country of South America in terms of GDP per capita. The cleanse seems to have worked.
But let’s move on to the hopefully less ideologically charged issue of an impersonal benevolent dictator. For this, I have to go back to how I view the role of government. In a distilled form it’s basically that government should to do whatever the market cannot do. In any case of market failure, it is the government’s role to step in. Healthcare, for instance, is a perfect example of an area where pretty much every known form of market failure occurs. It makes sense for the government to run it. In terms of externalities, it is the government’s role to get actual prices in line with socially optimal prices. It is the government’s role to get the market to function. If as a society we care about ethical and moral goals that the market cannot achieve by itself, it is the government’s role to step in. Laissez-faire liberals (I love the word miniarchist, btw) tend to vastly overestimate the power of the market. Collectivists, on the other hand, tend to massively underestimate the problem of state failure.
Taking politics out of the equation, I would say we have a fairly good understanding of the range of market failures that occur and where they occur. There are plenty of very reasonably sounding estimates of how much carbon emissions, for instance, should be taxed. Do it, Watson. And while you’re at it, implement a progressive consumption tax. Get rid of all the arbitrary distortions created by our political system and implement an unconditional minimum income to act as a safety net. You’re the boss. There is little doubt in my mind that such an arrangement would greatly increase personal freedom while at the same time contributing to economic efficiency. Since as a society we rightly care about inequality, have Watson take that into consideration and, if required, adjust marginal tax rates accordingly on a year-to-year basis. Inevitably, the question from where we are supposed to get all the required information needed to run the system in such an optimal way arises. But there is little diffuculty imagining that it would be theoretically possible.
“Maximizing social welfare” is a goal that seems pretty straight-forward. It is the technicalities where this notion seems to run into trouble. What is “social welfare”? How do you measure it? I personally am not sure the issue is as unsolvable as often claimed. I see no reason why, along the lines of the theory of revealed preference, we cannot assume the monetary value of a transaction to accurately reflect the utility gained from that transaction. As long as the transaction is bi-laterally voluntary and informed (and it would be Watsons role to make it so), clearly the amount of money I would want to spend on something has to reflect how much utility I expect to gain from it. In the case of non-observable (i.e. non-monetary, e.g. leisure time) arrangements, just take the opportunity cost of the action. Plug the sum of these utilities into a function which takes into account the very reasonable notion of decreasing marginal utility, add it up for everyone in the population and you’ve got total social welfare. Make sure the outcome is consistent with whatever notion of equality we agreed on and we’re done. Yet much more intelligent people than me have taken a crack at the issue and apparently come up short, so I’m sure I’m missing something. And even if you manage to measure current social welfare and compare it to other points in the dataset – it all relies on observed transactions. I see no way to construct a function with an essentially unlimited domain that is to be maximised, since you simply cannot know how much “utility” a person gets from alternative scenarios.
A lot of other important questions also remain: how much power should such a benevolent dictator have? How far should he be able to go to actually and actively “maximize social welfare”? What would the checks and balances to such a supercomputer be? What should stop it from going all V.I.K.I. on us, deciding we’re too stupid as a race and that the only way to effectively maximize our social welfare is to get rid of us all-together? How are we even supposed to get to the point where a supercomputer can take over in the first place? “Less power to the people!” hardly seems like a compelling call for revolution.
If we want our supercomputer to play a mostly passive role, making sure there’s a level playing field, that the market works wherever it can and government operations run as efficiently as possible, I can find little objection to such an arrangement. It might be possible to determine social welfare, as outlined above, but what if the value arrived at falls short of what could be theoretically achieved? Should the benevolent dictator be allowed to do what his name would indicate – dictate the actions of people? If not, would it even make sense to refer to him as a “dictator”? Should he be able to prevent someone from, just to name a clearly made-up example, rent out a Ferrari and speed along the Champs-Élysées after recently having had over a thousand Euros worth of electronic equipment stolen from him? I’m not sure Watson would approve. And that’s a beautiful thing, because it should be none of his business. For better or for worse.