Friday, September 05, 2003

Witnesses aren't worth much ...

In Same Case, DNA Clears Convict and Finds Suspect:
.. in a meeting this morning at a Burger King, Ms. Brobst told Mr. Bloodsworth of the DNA evidence against Mr. Ruffner and apologized for wrongly prosecuting him. Mr. Bloodsworth, who has become an outspoken advocate for reforming federal death penalty laws, said he cried and then hugged Ms. Brobst.

In an interview, the state's attorney for Baltimore County, Sandra A. O'Connor, said that the police and prosecutors had acted responsibly in the case, but that DNA technology did not exist at the time of Mr. Bloodsworth's trial. What did exist were the statements of five witnesses who said they saw him with the girl on the day she was killed.

Many of the DNA exonerations involved convictions based on witness statements, often many witnesses. Some of the witnesses were suspect (got plea bargains, etc), but many were reputable. Some of the witnesses were victims.

These stories, and a lot of cognitive science research, lead towards one important conclusion. Witnesses are unreliable; human memory simply does not work very well. A lot of what we "remember" is a "simulation" or "recreation" based on mixture of memory and clever invention. In other words, most of us routinely confabulate (there are people with exceptionally good memories who do less confabulation). Many of us are extremely suggestible, we can invent a lot of memory with just a few hints.

We need to deprecate the testimony of witnesses and rely upon them less. (PS. The death penalty is unworkable in the United States. Maybe it would work in Sweden, but they consider it barbaric.)

No comments: