The Violence of Crowds

Perhaps the most compelling reason for skepticism concerning the "wisdom of crowds" is the strong case of historical evidence that, with the reinforcement of a crowd, the individual is more inclined towards violent action than towards considered reflection. This is particularly evident in Gustave Le Bon's "study of the popular mind," The Crowd (originally published in French as La Psychologie des Foules). By the time we get to the fourth chapter of this book, the author has begun a laundry list of crowds swayed to commit atrocities through group psychology. That list includes all the blood shed over the Reformation, the Bartholomew's Day Massacre, and the Reign of Terror. (I find it particularly interesting that Le Bon associates such violence with religious conviction under the proposition that the "Jacobins of the Reign of Terror were at bottom as religious as the Catholics of the Inquisition.")

Recently I have wondered whether or not such violent thoughts may be surfacing in the more innocuous instrument of "crowd wisdom," the ability to have comments on news reports subjected to a thumbs-up/down vote. The San Francisco Chronicle does this; and while, from time to time, one may encounter some useful insights coming from vox populi, that voice is just as likely to be downright scary, particularly when it involves a response to a report of a violent act. It may even be that the disquieting nature of the response is directly proportional to the senselessness of the violence being reported.

Consider, as a case study, a story filed for this morning's Chronicle by Staff Writer Jaxon Van Derbeken. Let us begin with the basic events in the narrative:

Matthew A. Adams, 38, was found dead Saturday night in his room at 1169 Market St. by his girlfriend.

The woman told police that a man attacked Adams without provocation as the couple were walking near Seventh and Market streets at 1:30 a.m. Saturday, said Lt. Mike Stasko of the police homicide detail.

Adams refused medical treatment at the scene, police said.

"He said he was OK," Stasko said, "and he walked home from where he was assaulted."

Adams' girlfriend left later that morning. When she returned to his room about 8:30 p.m., Adams was dead.

On Sunday, police arrested Edward W. Holloway, 54, of San Francisco on suspicion of assault with a deadly weapon and battery, and for 11 outstanding warrants for alleged quality-of-life crimes such as public drunkenness.

Now the police have made it clear that they are not yet certain whether or not Adams' death was related to the attack. Nevertheless, the remainder of the "report" (now using scare quotes) provides "corroborative detail" (as William Schwenck Gilbert put it) about Holloway:

Stasko said Holloway has a history of attacking people on the street without provocation. He also has a long criminal record in Los Angeles and San Francisco for theft, drug offenses and public intoxication, the lieutenant said.

In March, police arrested Holloway on suspicion of felony assault after he allegedly hit another homeless man in the leg with a baseball bat and slammed a can of beer into the side of his head, authorities said. The district attorney's office dismissed the case because the victim was unavailable to testify against Holloway, records show.

Holloway was arrested again later in March on Sixth Street for allegedly carrying a concealed weapon, but prosecutors discharged the case "in the interest of justice," records show.

In May, Holloway was arrested on a domestic violence charge stemming from an incident at Turk and Taylor streets in which he allegedly stabbed a former girlfriend in the hand in a dispute over $30, records show.

A month later, the district attorney's office dropped that case on the day of the preliminary hearing, records show. Prosecutors said the woman was unavailable to testify.

This left me with the feeling that the whole story had been framed in such a way as to draw violent reaction; and, when I scrolled down to the Comments section, I was not surprised with what I saw. First of all the Chronicle has a policy of providing the total number of comments received, followed by the three "most recommended" of these comments, where a recommendation consists of clicking on a thumbs-up icon. The total number of votes both up and down are also provided.

The most recommended comment for this story was:

The wrong person died.

The comment was submitted last night at 5:29 PM; and, by the time I read the story this morning, it had harvested 975 thumbs-up recommendations (against only 15 thumbs-down votes). Now this kind of promotion of journalism [sic] through a variant on Internet gaming may not be as threatening as the lynch mob in Walter Van Tilburg Clark's Ox-Bow Incident; but it may still be an indicator of popular reaction. Furthermore, it may illustrate the extent to which that reaction can be manipulated by the way in which the "report" has been framed.

Manipulation is probably the key factor in this case study, even if the story had not been deliberately written to be manipulative. Many of the anecdotes told to support the "wisdom of crowds" premise tend to take place in an objective world, or at least one in which subjective and social influences appear to be relatively minor. Such a world offers little leverage for manipulation; and one may then encounter reasonably "clean" supporting data. Unfortunately, we cannot live strictly in the objective world; and, if we try to apply objective instruments when the objective world "bumps into" (as Ken Auletta put it in his recent talk at Google) the subjective and social worlds, the consequences may well turn out as ugly as they did in this minor example of mass voting, if not uglier.