Jeremy Daum and Changhao Wei are two of a handful of people on this planet most capable of explaining Chinese law to an American audience.¹ Both of them have the following set of skills:

1. Native or near-native knowledge of the Chinese language
2. Native or near-native knowledge of the English language
3. Expert knowledge of Chinese legalese
4. Expert knowledge of American legalese
5. Expert knowledge of Chinese political culture (including many years of lived experience), which helps them interpret Chinese laws and gauge the level of enforcement
6. Expert knowledge of American political culture (including many years of lived experience), which helps them find the right analogies

Yet despite the fact that they regularly provide research and analysis (for free!), they are not the most widely read, consulted, or cited commentators when issues such as the “social credit score” comes up in the news. What explains the discrepancy between the quality of analysis and the reach of commentators?

In Akerlof’s market for lemons, buyers of used cars have less information on the quality of cars compared to sellers, leading to adverse selection and even market collapse. In the market for U.S.-China commentary, buyers (e.g. readers, many journalists, most newsroom editors) have almost no information on product quality because they don’t have the aforementioned set of skills – language, domain knowledge, and understanding of the target country’s political culture.

Suppose your own level of understanding is {20/100 on language, 30/100 on law, and 10/100 on political culture}. You won’t be able to tell the difference in product quality between commentators whose levels of understanding are, respectively, {60, 30, 20} and {90, 80, 70}.

In such a market, what types of product (commentary) dominate? Those that appeal to consumers' fears and anxieties.

The Example of “Social Credit”

Take social credit as an example. Commentators who actually read the law and studied its implementaion have said from the beginning that it’s a rather boring regulatory credit check system where administrative and criminal punishments are better recorded and better shared between departments.² It’s aimed primarily at businesses, and the input is largely information the government already collects in its normal course of business (failures in food safety checks, financial delinquencies, etc.).

But media coverage of social credit, at least up until very recently, vastly overstates the program’s impact. If you search “social credit China," you will find many mainstream outlets saying that by 2020, “everyone in China” will receive an “algorithmically generated” score that takes into account all of their “social behavior” (e.g. who they are friends with, what they eat) and returns a “rank” that “determines their place in society.” Even today, this idea of a universal scoring system is still well alive in the American public imagination.

So how did the social credit myth persist for so long?

Here’s Jeremy Daum:

I think we managed to connect [social credit] to a floating anxiety we had about a digital caste system being created. People love to compare it to an episode of Black Mirror which predates most of China’s system. And there are other examples: The Good Place had a tallying system; other pop culture and novels have dealt with this. This idea was out there without a name. And I think, when we heard social credit, we were like, “Ah, that’s the name we’re looking for to put to this idea.”

In other words, consumers of commentary and analysis on China, limited by their understanding of the language, domain, and culture, respond more strongly to information that appeals to their exisitng fears and anxieties: The fear of “big tech” or “big government” at home creates demand for an image/meme of a scoring system in some distant foreign land. The fear of “socialism/communism” or “government overreach” at home also led many to paint a simplistic picture of China’s tech crackdown. (It’s easy to think of the crackdown as “communism strikes again,” but in reality it’s a mix of political power consolidation and attempts to address issues regulators in other countries are also concerned about: antitrust, privacy, and national security.) That China is large in size and alien in ideology certainly makes simple projections of our own fears and anxieties even easier.

On the other side of the Pacfic, consumers of Chinese-language commentary and analysis on the U.S. suffer from the same information problem. The winning strategy for an aspiring commentator there – partly thanks to the closed media environment – is to appeal to nationalism. What dominates the market is commentary (with some truths and some falsehoods) that paints the U.S. as a declining power and China as a rising giant. (See this paper on how much Chinese people overestimate their country’s popularity in the world.)

Emotions and Reason

Based on the observations I presented above, an alterantive model of the world is one where everyone engages in motivated reasoning: We (Americans and Chinese) demand commentary that appeals to our fears and anxieties not because we are incapable of evaluating its quality but because we prefer those that appeal to our emotional biases. In other words, even in a market where buyers can distinguish between peaches and lemons, lemons still drive out peaches because lemons make the buyers feel good.

This model of the world is certainly plausible, but what I hope to show in this post is that even in a world where consumers demand quality goods, inferior ones may still come to dominate because information on product quality has to cross linguistic, cultural lines.

The hurdle to information is present in lots of situations involving “translation.” Here’s a more literal example: I don’t speak Russian. If I need to decide between different translations of Anna Karenina, the only criterion I’m capable of using is whether the English sounds nice to an American ear. But of course, faithfulness to the Russian original ought to be far more important.

Limited by my understanding of the Russian language, I would end up picking the translator who over-optimizes on the beauty of the English rather than the translator who finds the right balance between faithfulness to the Russian and beauty of the English.

Given that finding this balance 1) takes much more effort than simply optimizing on the English and 2) generates similar if not less revenue than other versions, the market would be dominated by translations that over-optimize on beauty of the English.³

To some extent, the market for expert commentary on any issue experiences adverse selection. But what’s unique about the market for U.S.-China commentary is that 1) it has much fewer consumers with adequate knowledge (you need language + domain + culture, not just domain knowledge); 2) emotions run high even relative to hot-button issues in other domains such as vaccines, Bitcoin, or feminism, so it’s harder for truth-seeking consumers to overcome biases.

Possible Solutions

What are some possible solutions, both at the individual level and for the market as a whole?

1. Tying more of one’s payoffs to what is happening in the target country as opposed to how news from the target country makes you feel would incentive you to form more accurate beliefs. Participating in online prediction markets or having some exposure to the target country’s financial markets would be a concrete example.

2. The ultimate solution is to expand your knowledge to as close to {100, 100, 100} as you can so that you are qualified to judge a wider pool of sellers (commentators). Getting to {100, 100, 100} on all issues is, of course, infeasible. A realistic approach could be talking to friends or following people with different skill profiles. Together you would be capable of evaluating commentary on a broader set of issues.

3. Give more weight to commentary that uses systemtic evidence. A good example is commentary around China’s Belt and Road Initiative (BRI). If you only watch or read five-paragraph punditry on the topic, you are left with the impression that China is engaging in “debt-trap diplomacy.” But many researchers have systemtatically collected data and analyzed where the BRI projects are, what the different funding structures look like, and how these projects are doing now. Of course, data interpretaion requires domain knowledge and knowledge of statistics, so we as consumers still face the {0, 0, 0} limitation when evaluating expert commentary. But where applicable, the quality of commentary that cites systematic evidence is generally superior to those that do not.

4. People on the knowledge frontier of any given issue bear special responsibility to amplify analyses they find reasonable, including those that reach conclusions they disagree with. On issues at the intersection of many niche areas, the average consumer has no way of distinguishing between analyses that are “reasonable but different from mine” and those that “rely on complete falsehoods.” So experts ought to share all commentary they find reasonable, regardless of how much they agree with the conclusion. The difficulty lies in calibrating where you are on the knowledge frontier relative to others.

Doing Better

The history of great power conflict is filled with misunderstandings created by false or incomplete information. So are we doomed this time? Especially given the ever increasing military capabilities on all sides?

I’d say: Not if we are deliberate about what we read and what we share. Information and transportation technologies have vastly increased human-to-human contact, and we need to turn it into greater mutual understanding. Never before have we had so many multilingual, multicultural people with expertise in so many different areas. We should not degenerate into a world where American commentary on China is “omg black mirror” and Chinese commentary on America is “down with baizuo.”

We can do better. We have to do better.