THE WITNESS / WHAT THE FAMOUS CASE GOT WRONG
In 1964 a woman was murdered in Queens, New York. A New York Times report stated that thirty-eight respectable, law-abiding citizens watched and none called the police. The account entered textbooks, gave its name to a syndrome, and became the standard illustration of moral collapse in crowds.
It was substantially inaccurate. Later examination of the trial record and the original reporting — including work by Manning, Levine and Collins published in 2007 — established that the number was wrong, that most of those present could not have seen the attack, that the assault was not one continuous observable event, and that at least one person did call the police.
The Archive includes this record because the correction matters more than the case.
A story about human indifference was accepted for four decades by researchers, editors and educators, largely without checking, because it confirmed something the audience already believed about crowds and cities. That is the same mechanism this collection documents in more serious settings, operating on people whose profession is verification.
THE CORRECTED FINDING IS STILL REAL. Laboratory work by Bibb Latané and John Darley in the late 1960s established diffusion of responsibility as a genuine effect: the presence of other potential responders reduces the probability that any given individual acts, because responsibility is felt to be distributed and because each person reads the inaction of others as evidence that no response is required.
The second half is the useful half. Inaction is contagious not because people do not care but because each of them is using the others as evidence. A room of people all waiting for a signal will produce a room with no signal in it.
THE RELATION TO THIS COLLECTION. The chain divides responsibility deliberately. The witness position divides it accidentally, and the result is similar: an event that everyone can see and no one owns.
THE PRACTICAL FORM. The finding that has survived replication best is that specificity defeats diffusion. A request addressed to one identified person is acted on far more often than the same request addressed to a group. This is the rare record in this collection with an instruction attached: name someone.