Clever Hans Speaks*

A federal judge in Utah has ruled that drug sniffing dogs are unreliable as a source of probable cause because they cue into their handlers:

For as long as people have been driving, cops have been imagining reasons to pull them over and coerce them into "voluntary" searches. The Supreme Court's Rodriguez decision (sort of) put an end to extended stops -- the ones that start with a perceived violation that's dragged out until a drug dog arrives. Unfortunately, that decision only removed part of the equation. The Supreme Court's Heien decision made it possible for cops to rely entirely on pretext to engage in fishing expeditions by saying cops only had to think they witnessed a traffic violation, rather than actually be accurate about the laws they're tasked with enforcing.

Cops are still trying to bring drug dogs to routine traffic stops. The Rodriguez decision is generally taken to mean cops just need to be quicker about rustling up a K-9 unit. Cops love drug dogs because they allow cops to perform the warrantless searches they want to perform. The drug dog's handler can call literally any movement by the dog an "alert," turning normal dog behavior into "probable cause" for a search. It doesn't help that the dogs are rewarded for every alert and given no positive reinforcement for failing to find anything interesting.

Courts have historically been willing to cut drug dogs as much slack as they cut their law enforcement officer handlers. Subjective interpretations of anything an animal does to please its master is considered close enough to Fourth Amendment compliance to justify warrantless searches. Every so often, a court will question the reliability of the dog or the intent of its handler, but those are anomalies.

This case, via FourthAmendment.com, is an amazing anomaly. Not only did the court choose to hear from experts on drug dog training and handling, it actually went so far as to call into question the reliability of every drug dog in the state.

………

The defense brought in an expert witness, Dr. Mary Cablik, who has two decades of drug dog training experience working with POST units in Nevada and California. Cablik said the absence of "blind" training is a real problem. If the dog is only tested in areas where the handler knows drugs will be found, the dog carries this knowledge on to the real world and will continue to search for nonexistent drugs until it gives its handler what they want: an "alert."

………

Utah's training does not produce reliable drug dogs. Officer Moore's drug dog is possibly more unreliable than most, but this order makes it clear everyone who's been subjected to a drug dog sniff should challenge it. The state POST training has produced little more than handy Fourth Amendment circumvention tools for officers to use at will. This court is having none of this and refuses to condone the deployment of dogs that are basically trained to please their handlers, rather than actually detect narcotics.

This is a feature, not a bug.

Dogs, in traffic stops at least, are not intended to detect drugs, they are intended to provide a corrupt pretext for a search.

Dogs are eager to please their handlers, and when the only training that a dog receives, as was the case here, was where the trainer knows what they have to find, it does not train a dog as an impartial detector, it makes the dog into a fraud routine.

This continues, because this is what the police want.

^*The Clever Hans Effect, named after a horse that cued into its handler's subconscious body to create the illusion that it could do complex math.