EXPLOITATIVE INFORMING

Suppose you are a democratically elected despot. As despots are wont to do, you would like to start a war of questionable legitimacy. However, you cannot entirely ignore the will of the people. The people will only endorse the war if they are at least 50% confident that the war is just. You would not like to risk prosecuting a war that isn’t endorsed by those who will fight in it.

Unfortunately, your people are on to you. They are only 30% confident that the proposed war is just, and frankly so are you. You don’t want to start a war under such conditions.

As you are about to abandon your plans, your minister of philosophy proposes an experiment. The experiment must, she explains, be conducted in an unimpeachably honest manner. Otherwise, some citizens might ignore it due to your chequered history with the truth.

However, your minister arranges that the statistical properties of the experiment will be known to all, and its results will be certain to be reported accurately throughout the population, no matter what the experiment finds. The statistical properties of this experiment take the following form. In the case that the war is just, the experiment is certain to return a positive signal. However, if the war is unjust, the experiment has a 3/7 chance of returning a positive signal and a 4/7 chance of returning a negative signal.

Surely, your citizens think, there could be no harm in such an experiment. However, your minister points out an interesting pattern. On your shared credences, there is a 40% chance that the experiment returns a negative signal because there is a 70% chance that the war is unjust, and in 4/7 of these cases a negative signal will be sent. In this case, you and your citizens update to zero credence that the war is just and your citizens refuse to endorse the war. The other 60% of the time, the experiment returns a positive signal. A bit of algebra shows that, as you are good Bayesians, both you and your populace update in this case to 50% credence that the war is just, and the war is endorsed.

To see this, note that by Bayes’s theorem your credence in the justice of the war given a positive signal is equal to your credence in a positive signal given the war’s justice, multiplied by your credence in the war’s justice, divided by your credence that a positive signal will be sent. These quantities are, respectively, 1, 0.3, and 0.6, so your credence in the justice of the war given a positive signal is exactly 0.5.

There must be some devilry here. Although your populace is 30% confident that the war is just, by their own lights they will endorse 60% of wars proposed in this way based on nothing more than an honest, publicly observable experiment that gathers relevant information about the justice of the war.

I argue in my BJPS article that cases of this sort have a good claim to be classified as exploitative acts of information provision—that is, exploitative informing. And because they are exploitative, it may be permissible for their targets to resist them.

It might seem odd to say that your despotic experiment is exploitative. You have,  after all, done nothing more than to gather relevant evidence, pre-commit to sharing the results with your people, then honoured your commitments. What more, you might ask, must you give them? Your second-best pair of boots?

It might also seem odd to say that citizens may permissibly resist this act of exploitative informing. Orthodoxy has it that neither the expected accuracy of citizens’ credences nor the expected utility of their acts can decrease from any act of evidence-gathering, including listening to exploitative testimony. If this is true, then what cause have they to resist?

Nevertheless, I argue that cases such as this one are exploitative in five ways. Let us consider three of them here.

First, they exploit asymmetric constraining power over deliberative outcomes. You, but not your citizens, have the power to gather and share relevant evidence. The statistical properties of the experiment you perform constrain the probabilities of final deliberative outcomes, in the form of credences about the justice of the war. Because you have the unilateral power to choose which, if any, experiment is performed, you have the unilateral power to push deliberative outcomes in the direction that you choose.

Second, you have engaged in act manipulation. You gather evidence with the sole purpose of shifting the probability distribution over acts that your interlocutors will take. Your citizens begin  with 30% credence that the war is just, and hence zero chance of endorsing the war. So long as citizens obey a minimal reflection principle, that 30% credence cannot be destroyed, only spread out. You want to push your citizens to 50% credence in the justice of the war, and this can be done a maximum of 60% of the time, provided that the other 40% of the time their confidence drops to zero. You choose an experiment with exactly these properties to manipulate actions to be maximally favourable, with a 60% chance that citizens will endorse the war.

Third, you are a surplus hog. Information-gathering typically generates a utility surplus, captured by the expected value of information. But we can show that you have helped yourself to the surplus. The expected value of information from this experiment is strictly zero for the citizens. The only person who reaps a surplus is you. The same is true across all cases of exploitative informing.

Exploitative informing shows many hallmarks of exploitation. There is, for example, an old Marxist refrain linking exploitation to the appropriation of surplus products from labour. Not coincidentally, surplus hogs appropriate the entire surplus product of information gathering. All of this is done with the sole purpose of manipulating citizens’ acts to promote your ends, with no concern for what your citizens value. And underlying this all is the open exercise of power. Your unilateral power to gather and share evidence gives you asymmetric power to constrain the outcomes of deliberation and manipulate the acts that result.

Suppose this is right. Exploitative informing is genuinely exploitative. What follows? Many applications are possible, but perhaps it will be interesting to think about how the phenomenon of exploitative informing yields an unconventional argument against state monopolies on media, and in favour of free and open information markets.

In light of her success, you promote your minister of philosophy to the directorship of Ravda, a state media outlet with a monopoly on large-scale information dissemination. At Ravda, the philosopher does some surprising things. First, she fires all dishonest reporters and replaces them with citizens of impeccable honesty. She then makes a credible commitment that all reporting on Ravda will be truthful and the manner in which information is gathered will be accessible to all.

Just as you are about to sack her, your philosopher explains that she is engaged in but one more act of exploitative informing. Her monopoly on information provision gives her the asymmetric ability to constrain deliberative outcomes, thereby manipulating the acts that citizens will take. In doing so, she can maximize the value of information accruing to the state by monopolizing the entire surplus value of information.

The old guard at Ravda manipulated citizens through lies and deceit. But the philosopher’s intervention works if and only if, and because, it is open, honest, and truthful. If citizens retained some doubt about the honesty of Ravda, they would update less strongly on information provided and thereby become more difficult to manipulate. In this way, your resident philosopher has discovered a surprising way for state media to manipulate through honesty and truthfulness rather than dishonesty and evasion.

How can Ravda be stopped? So long as Ravda maintains its monopoly on information provision, there is not much that can be done. But suppose that many other media companies were competing for citizens’ attention. Ravda would then lose its monopoly on information provision, and thereby lose the ability to asymmetrically constrain deliberative outcomes through exploitative informing. This decreased ability to constrain deliberative outcomes would hamper Ravda’s ability to manipulate citizens’ acts and monopolize the surplus product of information provision.

It is widely agreed that free and open information markets convey many benefits. However, many existing arguments for free information markets rely on the need to combat deceptive sources of information. The phenomenon of exploitative informing reveals how closed information markets allow for a pernicious form of exploitation in plain sight, in which honest reporting is used to manipulate the beliefs and actions of a captive audience. This application of exploitative informing may help us to understand some of the pressures for even monopolistic state-run media outlets to provide more credible reporting and it certainly gives us another reason to celebrate free and open information markets.