The Role of Information Cascade in the Acceptance of Unintelligible Texts
In the above article, “The Intelligibility of Unintelligible Texts: Authorial Status in Text Evaluation,” Robb Willer explores the effect of the reputation of an author on the judgment of the comprehensibility of texts.
An interesting phenomenon has shown up in publications. Journals have published articles that are relatively or incompletely incomprehensible, even to those who are experts in the field. In 1996, Alan Sokal was able to take advantage of this fact by writing a paper imitating post-modern critiques of the natural sciences, but his paper was intentionally full of contradictions and false facts. After Sokal’s paper was accepted by the journal Social Text, he published a reply to his own publication and exposed the hoax he had played.
Information cascades, or herding behavior, can explain why this would happen. At the time of the hoax, Social Text was not peer-reviewed. The article seemed to fit the preconceived notions of the postmodern publishers, but was not actually based on reason. Those who did read the article undoubtedly found it incomprehensible, but did not wish to admit this to the other reviewers, and thus the readers supported the article because they assumed that the other readers understood it. As we discussed about information cascades, if you have private information about something (this text is incomprehensible), but everyone else is making a choice (supporting the article) that is different than what your private information suggests you should choose (not support the article), you assume that these other people have their own private information (they must understand the article) that is more accurate than your own.
Willer did his own experiments with unintelligible texts and found that if the author is of very high status and writes an unintelligible text, the readers will rate the paper as relatively comprehensible. If the author is of low status, then the article will be rated as less comprehensible. People assume that if the author is of high status that he or she must write comprehensible texts, which leads to positive text evaluations. This is because if a reader assumes that the article will be comprehensible based on the author’s reputation, then he or she will not want to admit to being unable to understand the text since it is assumed that the other readers will be able to understand it. This is herding behavior in that the acceptance of an article by one or a few readers will lead to other readers accepting the article even though it is incomprehensible because the readers assume that private information held by the other readers (i.e. understanding of the text) is sufficient to validate the text.