Last week, I wrote a post looking at how the criminal justice system operates in an alternate reality, one in which truth isn’t dictated by facts or data, but by precedent and case law. Today, I want to look at a case pending before the Supreme Court that is a great example of the problem.

At issue in Edstrom v. Minnesota is whether a drug dog’s sniff outside an apartment door constitutes a lawful search under the Fourth Amendment. If it does not, the police would be required to obtain a warrant before using a narcotics-detecting dog in that manner. If it does, then the police could take their dogs up and down apartment complexes the way they sometimes do with school lockers. Over at the legal analysis site Verdict, Cornell University professor Sherry Colb runs through what’s at stake, and offers some informed speculation on what the court may do.

For the purpose of this post, though, I want to focus on what’s missing from Colb’s analysis and, should the Supreme Court decide to hear the case, will almost certainly also be missing from oral arguments, the court’s ruling and most discussion of the case: that narcotics-detecting dogs and their handlers aren’t very good at discerning the presence of illegal drugs. Multiple analyses of drug-dog alerts have consistently shown alarmingly high error rates — with some close to and exceeding 50 percent. In effect, some of these K-9 units are worse than a coin flip.

For some units, the reason may be sinister — the police handler may have trained the dog to alert on command. I’ve asked dog trainers to look at videos of roadside searches in the past, and, on more than one occasion, they said they saw clear indications that a dog was being cued to alert.

But it needn’t be so malicious. While dogs are indeed capable of sniffing out illicit drugs, we’ve bred into them another overriding trait: the desire to please. Even drug dogs with conscientious handlers will read their handlers' unintentional body language and alert accordingly. A 2010 study found that packages designed to trick handlers into thinking there were drugs inside them were much more likely to trigger false alerts than packages designed to trick the dogs. (Police-dog handlers and trainers responded to that study by refusing to cooperate with further research.) Many drug dogs, then, are not alerting to the presence of drugs, but to their handlers' suspicions about the presence of drugs. And searches based on little more than law enforcement’s suspicions are exactly what the Fourth Amendment is supposed to prevent. (Tracking dogs that pick suspects out of “scent lineups” have had similar problems, and have led to numerous wrongful convictions.)

If the Fourth Amendment rights of drug suspects hinge on drug dogs, one would think the accuracy of those dogs would be something the Supreme Court might want to investigate. You’d be wrong. In the 2005 case Illinois v. Caballes, the court ruled that a dog sniff after a traffic stop is not an unreasonable search under the Fourth Amendment. In his majority opinion, Justice John Paul Stevens explained: “A dog sniff conducted during a concededly lawful traffic stop that reveals no information other than the location of a substance that no individual has any right to possess does not violate the Fourth Amendment.” Buried in that sentence is the presumption that drug dogs only alert when illegal drugs are present. We know that isn’t true. And we know that each time a false alert leads to a search, a motorist is subjected to a thorough, invasive, often humiliating roadside search. Police have been known to empty luggage, look through pockets, and leave belongings strewn along the side of the road.

To prevent the innocent from being subjected to unjust searches, drug-detection dogs need to be accurate. There should be checks in place to prevent both intentionally and unintentionally cued alerts — including teaching dogs to disregard the body language and hunches of handlers.

Those studies showing startlingly high error rates are pretty good evidence that this sort of training isn’t happening in large areas of the country. In fact, dog trainers have told me that not only can dogs be trained to ignore unintentional cues or body language from their handlers, they can also be trained not to alert to immeasurable quantities of illicit drugs. But police departments don’t want dogs trained that way. They want dogs that will alert often. They want dogs that will err on the side of alerting.

Thus far, it appears only one Supreme Court justice has addressed expressed concern over any of this. Justice David Souter, during his Caballes dissent in 2005, made mention of the disturbingly high error rates and concluded “the infallible dog . . . is a creature of legal fiction.” But Souter was outvoted. And in the criminal justice world, that fiction has become fact.

The Supreme Court had another opportunity to address the problem in the 2013′s Florida v. Harris. At issue in that case was whether mere certification could be enough to establish a drug dog’s reliability, or if, as the defendant argued, a judge should also consider a dog’s record in the field before ruling on the legality of a search based on the dog’s alert. In a unanimous decision, the court rejected the defendant’s argument. If a dog is certified to detect narcotics, the court ruled, its roadside alerts are sufficient to establish probable cause for a search.

Importantly, the court put no qualifiers on how reputable the certifying organization should be, or whether certification should ensure that the dog has been trained to disregard its innate desire to please its handler. The court also didn’t seem concerned with the fact that there are incentives for certifying organizations to please its customers — which, of course, are police departments. The court seemed incurious about important matters such as how often does the certification organization fail a K-9 team? Who determines whether the K-9 unit passes?

A drug dog’s actual performance in the field is a far better indication of its accuracy than whether it passed a controlled test administered by a certification group. But the Supreme Court justices just didn’t get it. During oral arguments, Justice Antonin Scalia seemed perplexed as to why a police department would want a dog prone to false alerts. “Why would a police department want to use an incompetent dog?” he asked the public defender who argued the case. “What incentive is there for a police department?”

That Scalia, who died in February 2016, would ask such questions revealed how far removed he was from day-to-day policing. The most obvious answer is civil asset forfeiture. The courts — including the Supreme Court — have made it far too easy for police officers to seize property they claim is connected in some way to a drug crime. While most courts have refused to uphold a forfeiture on the alert of a narcotics detecting dog alone, an alert coupled with a police officer’s claim that a motorist seemed nervous, agitated, or gave inconsistent answers to questions can be enough to justify a seizure. But, more generally, police want the power to search people they suspect of crimes. Even if a search doesn’t turn up drugs, it could turn up evidence of other crimes or infractions that merit a citation. In departments where officer performance is measured on arrests made or citations given, that can be a powerful incentive.

The Harris ruling made these dogs' real-world accuracy irrelevant. In the alternate reality of the criminal justice system, from 2013 on, any police dog that is “certified” is accurate and reliable — because the Supreme Court says they are.

Two years later, a ruling from the U.S. Court of Appeals for the 7th Circuit stretched the dimensions of this new reality. In United States v. Bentley, the defendant was searched after an alert by a drug dog that had alerted 93 out of every 100 times it sniffed. Why did it alert so often? Perhaps because the drug dog’s handler admitted that he rewarded the dog with a treat only when it alerted. The dog was confirming its owner’s hunches, and getting a treat each time it did. It also had a false positive error rate of 41 percent — 4 out of every 10 drivers searched because of a dog’s alert turned out to be innocent.

And yet, relying on Harris, the 7th Circuit upheld the search. The three-judge panel rejected the defendant’s challenge to the dog’s reliability. The same court had previously rejected a challenge to a dog with a false positive rate of more than 60 percent. In both cases, the dog was “certified,” and the pet’s handler testified that the dog was reliable. Under Harris, nothing else matters.

The Supreme Court got it wrong on drug dogs back in 2005. But instead of correcting itself, the court and the lower federal courts have since perpetuated a line of jurisprudence that has turned drug dogs into furry generators of probable cause — or as one defense attorney put it, “a search warrant on a leash.” It’s no wonder one police department scrapped any pretense of fairness and just named its drug dog “Guilty.”

In another ruling the same year as Harris, the Supreme Court threw out a search based on a drug dog’s sniff of a home on private property. But that ruling (Florida v. Jardines) was based on property rights — the court found that the police didn’t have probable cause to come onto the suspect’s porch with their dog, and were therefore trespassing. The issue in Edstrom is whether a sniff of a door from a public or shared hallway in a multifamily dwelling — such as an apartment complex or public housing — constitutes a search. If the justices take up the case and decide that it is not, as the Minnesota Supreme Court did, the justices will create a massive loophole allowing law enforcement to search the homes of people who live in such facilities based on little more than a hunch. (It would also mean that people who live in multifamily buildings have fewer Fourth Amendment protections than people who live in single-family residences.)

And, yet, in our criminal justice system, those searches would not violate the Fourth Amendment rights of the people inside. Because in this reality, the Supreme Court determines what is and isn’t true. And the Supreme Court has said that so long as they’re certified, drug dogs only alert when someone is guilty.