This makes me uneasy.

On several occasions I've had need to record eye-witness accounts of incidents (mostly accidents of varying sorts). There is always a degree of variance between the accounts, sometimes inconsequential, sometimes dramatic. In all the cases, the accounts were coming from people who were mature, trustworthy and doing their best to be helpful. They just had a significant degree of variance. I talked this over with a LEO friend of mine and he suggested that the police use absolute agreement on a story as a sign of prior collusion and discounted sets of eyewitness reports unless they showed a degree of variance.

I've also had eyewitness accounts that quite simply could not be true even though the authors were quite dogmatic about what they'd seen (usually any attempt to find out what actually happened ended up in an "I was there and you weren't type of discussion). Another friend of mine who works on air crash investigation once told me that the best eye witness accounts come from children because they tell you what they saw not what they think they saw (an adult will say there was an explosion in an engine, a child that they saw a big puff of smoke from an engine). The problem there is that children don't speak the same language as adults so the person taking the statement had to listen carefully to understand what thechild is saying. Its also very, very important not to "lead" the child into making statements they didn't mean to say. My ACI friend says that interviewing children is a very specialized art that should only be undertaken by experts.

I agree that true stories have a high degree of informational congruence but they are still often contradictory. I'm dealing with a case now where we have two incidents very close together in space and time watched by hundreds of skilled witnesses and we still don't know what happened.