We don’t like to change or update our beliefs, even when presented with evidence that contradicts them. However, there are differences on how costly it is to do so for each person. Whatever the fundamental reason, there seems to be a lot of heterogeneity in how costly it is for each one of us to update. It is likely that extreme experiences increase the cost of updating. For example, the direct involvement of the US government in establishing right-wing military juntas in Latin America may have (rightly, at least at the time) formed a strong belief among left-leaning people that any US involvement is in itself undesirable. Or, conversely, that any system that is opposed to the US is in itself good and desirable.

One example of this is the refusal, for the past decade, by a portion of the LATAM left, to admit that Venezuela is a failed state and a dictatorship. Whether it involves Venezuela or Cuba, some people go to great lengths to find rationalizations for the clearly bloody and repressive regimes that rule those countries. One would think that this is an extreme position, something that you wouldn’t hear from the average Chilean, or even the average Latin American, and that view may actually be right. But I find that there is a strong refusal, by part of the Latin American left, to update their prior beliefs in the face of overwhelming evidence.

While this may be true in the West, the DSA for example is almost a caricature of this, I find that in Latin America this is an old phenomenon and those who hold these views are much closer to power than anywhere else. In Chile for example, Frente Amplio, a broad coalition of leftist parties that currently holds the presidency, is composed of a group of people who wrote this. It is a tract on Chilean foreign policy and how to evaluate democracy in the dictatorship triumvirate (Venezuela, Nicaragua and Cuba). Let me highlight some passages from the text that I found particularly telling (my translation):

Can we question democracy in Venezuela? Obviously, just like we can question North American democracy, which is unparalleled in its violation of the human rights of its indigenous community and the rest of the international community through armed military intervention in areas of the Middle East and even Latin America.

The “tu quoque” or “whataboutism” is a common fallacy used in Chile to distract from the disastrous human rights record in Venezuela. Reading further:

But despite all these achievements, Cuba requires more democracy and citizen participation, which is why they themselves, without the interference of international organizations that receive orders from the IMF or interventionist deputies, have initiated a constitutional reform since July to strengthen their democracy, through a constituent, binding and participatory process – an issue that in Chile we have not been able to solve. –
What a glorious nation! Without putting its people at risk, it democratically seeks to advance and improve, because without a doubt, they have also stated that Cuban democracy must improve.

One can argue that this view may not be representative, and that those who wrote this piece are to the extreme left of the coalition, something that may be true. However I would have to insist that the burden of proof lies on the other side of the argument; if members of your coalition keep making the same claims over and over again, the argument that they are not representative gets thinner and thinner.

The point of this post is not to dredge up extreme positions and point fingers, it is to quantify the resistance that people have to updating their political beliefs. These reactions, besides being baffling and prone to ridicule, beg a deeper question. At what point in time should people finally abandon their prior beliefs? And, if they do not, how does one go about measuring the obstinance of said beliefs and the inability of those holding them to update? These are useful questions in the sense that they allow us to properly understand how difficult it would be for someone to change their mind when faced with contradictory evidence, a useful trait for a person who has decision making power.

The index

To answer these questions, I am going to use the Venezuela example as a benchmark. Following the country’s timeline, one can pinpoint some key dates where the necessity of updating would be obvious, but I will allow for some degree of flexibility. The idea is simple; suppose we denote by the probability that an individual assigns to the bad outcome. What is the bad outcome? Well the bad outcome is Venezuela today; a dictatorship that violates human rights, has left its economy in shambles and where at least one fifth of the population has had to flee to feed itself and their families. At what point in the Venezuela timeline should that probability be 1? Let’s call that date and let’s suppose, for the sake of simplicity, that at the beginning of Chavez’ rule it is reasonable to have ¹. One can go from 0 to 1 in an infinity of ways, but, to simplify, let’s suppose that a completely rational, machine like creature, a Martian if you will, would do so linearly. That is to say, if we put years on the axis and on the axis, the proper updating function would look like this:

The value of can be discussed, but, it seems to me that , that is, after 20 years of Chavismo (2018), it should be clear that . I argue below that , but we will return to this in a bit. For now, let’s keep it as a variable.

Naturally, human beings don’t update like machines, we have biases and strong priors. Some of us resist updating a lot, and some are fast in updating. Either way, our updating function is slow at the beginning and eventually changes very rapidly. We resist and resist, if we have strong priors that Chavez is going to be great for Venezuela, and once things look pretty bad, we eventually update quickly². So the human updating function should be some sort of convex function, one that starts slowly and then updates quickly. It would look something like this:

where is the eventual time when an actual human would update. For this I have used the functional form:

where is the time dimension and is the parameter that measures the graduality with which we update; a very large means that when we do update, we do so almost immediately. measures how fast we get to the right conclusion, while measures how drastic our eventual change is. The following figures show the difference (the left figure varies keeping constant while the right figure does the opposite):

I have taken a 20 year timeline. As goes to infinity, a person never updates, and as goes to infinity, a person updates instantaneously, that is to say, they just wake up one day () and decide, ok, today . Most people are not that extreme, but some may be.

Now we’re ready to construct the index. We can measure how ideological a person is, or how sticky their prior beliefs are, by measuring the difference below the martian curve and below the human curve. I have depicted this below for two different levels of (on the left and on the right) :

If we divide this area by the total area of the trapezoid to the right of the linear curve, we get an index that should be between zero and one³. If or and then the index should give us 1, perfect ideological purity, a true believer⁴. And if and , the index is zero, the perfect machine, the unfeeling martian who cares only about facts. Note that this is not some sort of re-invention of the wheel, it is just an application of the GINI measure, where instead of the Lorenz curve we are using our updating function.

Finally, you may have noticed that I have put the maximum of the axis at 20. This is deliberate; Chavez took over in 1998 and by 2018 it was clear that Venezuela was a basket case. Besides RT and the odd North Korean broadcast, no one was arguing that it was doing well. Now you may argue that different people assign different reasons for this, and you would be right. There are those that argue that this is all due to US sanctions and Maduro is god’s gift to bookkeeping, fine, those people would have a , and they are still included in the index, their curve is just a lot flatter. But I think 20 years is a pretty good cutoff for how long you want to keep the axis going before you start ordering straight jackets.

Here is the final index (vui) for two individuals with different and with on the horizontal axis ( calculations below).

Why t₁=10

Chavez won his first election in 1998. By 2008, 10 years after his first victory, the following had happened:

• Created a civilian militia
• Nationalized the telecom and electricity industries
• Nationalized the central bank
• Cancelled the license of the private media company RCTV
• Proposed an act that allowed him to rule by decree for 18 months
• Nationalized the oil industry
• Held (and lost) a referendum to abolish presidential term limits and to suspend media rights and hold citizens without declaring charges during a state of emergency. (he held the referendum again in 2009 and won).
• Human Rights Watch issued a very critical report of his regime’s human rights record

By 2008 or, at most, 2009, if one is unencumbered by the chains of ideological affinity, the only logical conclusion is that Chavez is going to be a run-of-the mill populist Latin American caudillo. In my view, this is the cutoff that gives you the lowest ideological purity score, which, while not zero, allows you to get away with just being a human and not wanting your beliefs to be wrong. if and you are the least ideological approximation to a human alive. Otherwise you’re a mortal like the rest of us.

Now I understand that the choice of may be controversial, I stand by my assessment that it is 10, but some people, with a stronger bias than mine would argue that it is the arrival of Maduro that messed things up. I disagree, but I understand the urge.

Application

Now consider the hypothetical case of a leader who supports the Venezuelan government well into 2013, 15 years after the first Chavez win. This hypothetical leader, acknowledges that there are problems, but that the way forward continues to be “el Chavismo”. Clearly, this person’s . Let us be generous and assign . Suppose also that by 2018 this leader begrudgingly admits that Venezuela is a basket case. Under the functional form above, and so their index would be .

Suppose also, that a journalist goes to Venezuela in 2014, in the middle of shortages so severe that there is even a wikipedia page on the subject. Suppose that said journalist visits a supermarket next to a military base, where shelves are full of the same product, and claims that the problem isn’t that there are shortages, it’s that the price of things is too high. Clearly this person is avoiding readily available information so as to not change his mind. He acknowledges there are problems, but they are exaggerated by adversaries. Let’s say that our confused journalist finally admits, in 2019, that Venezuela is a failed state. This person’s and (assuming ). This hypothetical journalist has an index of 0.67.

In the following figure I plot the index for these two hypothetical people, where the x-axis is the assumed value of ( for our leader and for the journalist). The value of varies between 0 and 0.9.

The argument that I hear often in Chile is that those assessments were made at a particular point in time and that, given the eventual acceptance of the state of things in Venezuela, they mean very little in the current context. This is where I fundamentally disagree with that conclusion. Updating is a form of wisdom, something that measures a person’s ability to change their mind in the face of evidence that contradicts his or her beliefs. The longer the person takes to do so, the more difficult it is to trust them to make decisions of national importance. This is not a trivial matter, it is essential. At some level our minds are information processing machines, and, if they do not process information fast enough because of pre-existing biases, they are not reliable in the long run. To me, the fact that someone finally admits in 2019 that the situation in Venezuela is a disaster is a terrible signal. It suggests that their model of the world is broken and that they are an unreliable conduit of reality.

Calculations

Suppose that the cutoff time is denoted by instead of 20 as above. Then we can construct two indexes so that the final index is a step-wise function. We have:

(1)  

(2)  

where .

Footnotes

1. This is a simplifying but not too realistic of an assumption. The man tried to take over by means of a coup twice, you could not be blamed if you were skeptical about his democratic credentials.
2. This says nothing about the reasons why we are like this. This is also a heuristic, I have no empirical proof for the functional form that our updating has. There is plenty of work, starting with Kahnemann and Tversky that suggests our priors are sticky, but there will be enough flexibility for this input.
3. A keen observer would argue that if the index could be negative, and that would be correct. I am assuming here that the correct updating time is , and that having any earlier is still a from of bias. So one could argue that these people are also ideological, just in the opposite direction
4. There is room for complaints here. I am treating and individuals the same way, as ideological purists, which is not quite fair. individuals eventually update if . Given the way I have set up the problem, as long as there are no discernible differences. But the complaint would be valid.

[email protected]