Information Cascades in the spread of Gossip
https://today.uic.edu/study-explores-how-gossip-spreads-in-social-networks
Gossip and its spread is something we have all dealt with at some point in our lives. The spread of information about people, typically unconfirmed information, has the potential to lead to many people believing false things about others. In a paper, Brian Flood and Co-author Rinni Bhansali discuss a paper/study from Laura P. Schaposnik (associate professor of mathematics at the University of Illinois at Chicago) that explores how gossip spreads in social networks, particularly in large populations. I hope to add to this discussion by drilling in on the information cascade aspect of the spread of gossip.
Flood initially brings up the question “how many different types of people… should we hear a piece of information from before we start sharing it as a true fact? Most importantly, he discusses the study’s developed model that “elucidates the reasons why some news propagates through social networks before there is time to corroborate the facts” (which in itself is a major reason for the formation of an information cascade), and how that model was applied to a society of 10,000 people where, for example, everyone is either a Republican, Democrat, of Independent. In the end, Flood discusses how the conclusion to this study was that the quickest way gossip spreads was if a person heard and believed gossip from at least one of these three types of people.
Diving deeper into exactly how this fast gossip spreads, and how this developed model works we have to further analyze the information cascade structure behind it. Gossip tends to start with 1 individual observing a piece of information about a particular person. For example, let’s say that this piece of information is that Johnny likes Sally (a CLASSIC middle school case). Because they are the first person with this knowledge, with no other person before them to influence them, the only factor in their decision is their own observation (or “signal”), so they will naturally believe it, or decide to “accept” it.
Person 1, due to human nature, naturally wants to share this information with someone else: Person 2. Person two observes Person 1’s accepting that Johnny likes Sally, but not Person 1’s actual observation, or (“signal”). Person 2’s previous knowledge about this Johnny and Sally gives Person 2 a signal on whether or not Johnny likes Sally. This could be because Person 2 actually thinks Johnny likes Sally from seeing him talk to her in the hallway at school, or for another reason. Person 2 is influenced by Person 1’s decision to accept this information. However, he will follow his own observed signal regardless. Person 2 will either (1) gets a signal matching the action of Person 1, that Johnny likes Sally (let’s call this a “high” signal), only confirming his own signal and causing him to decide to accept as well, or (2) gets a signal opposing the action of Person 1, that Johnny doesn’t actually like Sally (“low” signal), but due to the tie-breaking rule that if the person is indifferent between A and R with conflicting observations and information, they then they will follow their own signal, causing Person 2 to believe Johnny doesn’t actually like Sally. For the sake of this case, assume Person 2 received a high signal that Johnny likes sally.
Person 3 now, through word of mouth, observes Person 1 and Person 2’s accepting that Johnny likes Sally. Person 3 also observes their own signal about the Relationship between Johnny and Sally. Person 3 can also infer the signals from the two people before him, who both accepted this information. Using this information of these 3 signals, regardless of Person 3’s own observed signal, the inferred signals of Person 1 and Person 2’s decisions before him to accept that Johnny likes Sally will outweigh Person 3’s own observed signal of Johnny and Sally 2-1; he knows that the observations of the two people before him must have pointed to the idea that Johnny likes Sally, so their quantity simply outweighs his own signal, causing him to accept that Johnny likes Sally against his own signal. This is where the information cascade starts.
This dynamic will repeat for all people in the gossip chain. The next person, Person 4 observes their own signal on the relationship between Johnny and Sally and whether or not Johnny likes Sally, but also knows that due to the dynamic of this cascade that Person 1 and Person 2’s decision to accept that Johnny likes Sally is reflective of the information, or observed signal, that they had. Because of this, like Person 3, will follow the herd and, regardless of her own signal, choose to accept that Johnny likes Sally because the others’ signals outweigh her own. This will be the same thought process for Person 5, 6, 7, and so on.
As we can see here, it is very easy to create a cascade, regardless of whether the information, or gossip, is correct or not. For all we know, the signals, or observations of Johnny and Sally by Persons 1 and 2 could have been false (something very common with predicting romantic connections between people), or Persons 1 and 2 could have even formulated false observations and therefore false acceptances of Johnny liking Sally in order to purposely spread a fake rumor about Johnny because Persons 1 and 2 simply don’t like Johnny. The point is, the spread of this information occurred because of a disconnect in information between people, and had if the people following Person 1 and Person 2 actually shared their real observations/signals on Johnny and Sally, this gossip would not have spread as far. It circles back to the concept mentioned in the resource that “some news propagates through social networks before there is time to corroborate the facts”, causing information cascades.
This resource and it’s concept of a lack of sharing facts/information/observations/signals when between connect back the topic of information cascades discussed in the course as it gives a real-word example of how information can be spread in a cascade or herd effect regardless of it’s truth. It discusses signals, and their weight in the decisions of future people in the chain in comparison to the actions of the (only) two people who began the cascade.