Should a dog's sniff be enough to convict a person for murder?

https://www.science.org/content/article/should-dog-s-sniff-be-enough-convict-person-murder

by pseudolus • 5 years ago

140 157 5 years ago

TLDR: No

Usually it's not a dogs sniff but the dog handlers evidence about the behaviour of the sniffer dog. And dogs are far too good at behaving to please their handler/owner sometimes and acting out of instinct or boredom in others - how many dogs that have gone on a murderous rampage have been said to be sweet, loving creatures all their lives.

It's hard to see how a dogs sniff evidence can be treated with the same level of accuracy as fingerprints/DNA evidence unless tests had proved that it (perhaps even the individual dog) was accurate to 1 in 1000 or 1 in 10,000 or more cases.

Jon_Lowtek• 5 years ago

> Corcoran had trained Molly to sit when the dog detected the odor of human remains ... the team went to Redwine’s cabin, which was unoccupied. Molly sat down in a nearby stream and indicated at five other locations outside the home, according to Corcoran. Then she saw her dog looking up at a window. An investigator climbed into the house and opened the door for Corcoran and Molly. Inside, the dog made its indication—a sit—near a wood stove, in a bedroom, on a brown sofa, by a washing machine, and elsewhere.

insert "this is worthless" spongebob meme template here. Convicting someone of murder because a dog sat in their house is worse then fake science, it is worse than superstition, that dog handler is blatantly creating fake evidence. Has that dog ever cleared someone of an accusation their handler believes in by refusing to sit near their stuff?

Edit: to clarify, a cadaver dog is useful, if one searches for a cadaver: when the dog indicates a specific place one should dig there. But that is not how the dog is used here.

h2odragon• 5 years ago

IMO, this discussion goes wrong when we involve the dog in the first place. The testimony is that of the handler; they are the party saying that they have reason to believe X, based on their interpretation of their dog's behavior... which might as well be a dowsing rod or any other scrying method for the repeatability it has.

If a dog/handler leads them to evidence, then the evidence can stand (or not) on its own just as if they were led to it by a psychic or a "hunch".

annoyingnoob5• 5 years ago

I had a security dog in an airport come and sit next to me while I was standing in line. Thought for sure my day was over, really freaked me out because of stories like this. Nothing happened, I think the dog just figured me for the dog person that I am. You are not supposed to pet working dogs while they are working but that didn't stop me.

michaelcampbell• 5 years ago

Making no opinion one way or the other, convictions are (I suspect) routinely made on eyewitness testimony which has been show many, MANY times to be horribly unreliable.

encryptluks2• 5 years ago

I was going to say the same. Not to mention how people are conditioned to trust cops and how prosecutors only need to get a person to say that they saw someone do something or heard them talk about it and a cop to say, yes this person is bad based on my interactions, in order to get conviction.. regardless if there is any real evidence, or evidence to the contrary in that matter.

ChrisKnott• 5 years ago

Techniques like this should only ever be used to direct investigations, and not be presented at court as evidence in their own right. It seems they work to an extent, but we clearly don't understand when and why they don't work.

andi9992• 5 years ago

Can we just get back the polygraph?

boomboomsubban2• 5 years ago

It's sad that a bullshit system that the accused and the accuser can easily game is one of the "successes" of forensics. At least there the accused can do something.

eftychis2• 5 years ago

Or even better uniformly at random imprison some possible suspect -- you avoid the cost of the trial and you are fairly certain that an honest person is in jail -- that seems to be the goal of the system: let the jurors, judge and cops sleep at night. Who cares about justice anyways...

andi9992• 5 years ago

No smoke without fire they say.

inglor_cz• 5 years ago

There is a huge difference between a dog finding a body (material evidence independent of the dog's behavioral characteristics) and a dog handler "reading" his dog's behavior and making various predictions out of it. The latter should absolutely not have any value in court.

And to think that we laugh at the Romans for predicting future from the flight of birds (augury) and animal liver (haruspexy)...

agys• 5 years ago

Thank you for “augury”! I looked up the etymology in Italian for auguri and it derives from the flight of birds. “Tanti auguri“ means “best wishes“, on the contrary “sciagura” (same root as “auguri”) means “misfortune“.

DaiPlusPlus2• 5 years ago

predicting future from the flight of birds

Well, to be fair, we can - to the extent that when I see birds fly-off from my back-garden I know it's about to rain...

ashtonkem• 5 years ago

And to be fair, if you’re a subsistance farmer the ability to predict rain and other weather events is a huge percentage of the future you care about. Augury would be able to predict a decent bit of the important future for those people.

jjluoma• 5 years ago

The same dog was used in all occasions: 1) sniffing of the clothes 2) around the house 3) inside the house 4) sniffing the truck.

I would question if the dog was essentially trained to recognize the smell of the suspect, and if so, did the dog really indicate it was smelling odors from a dead body or the odors from the suspect.

Also, only clothes of the suspect from an evidence room? Only one truck?

kolanos3• 5 years ago

I would question if the dog was essentially trained to recognize the smell of the suspect, and if so, did the dog really indicate it was smelling odors from a dead body or the odors from the suspect.

And upon questioning, how would you possibly be able to verify?

eftychis2• 5 years ago

You mean accept as evidence a dog's reaction which has been criticized and known to be directed by officers/handlers for illegal searches (by providing fake probable causes). A dog that you can not ask what they detected or even statistically test as we have seen in the article like you could do an ML model.

A whole field has been sprung on if and how ML inference could be used to make decisions, with lots of issues of racial profiling. And the judge and jury decided to accept a dog's reaction whose handler's new career was on the line.

I am not sure why one should trust the justice system these days.

P.S. A bunch of posts about Donziger and how UN Working Group on Arbitrary Detention has stated to the U.S. he detention and imprisonment is against international law. I just find the sequence of articles ironic.

tqi• 5 years ago

Agreed. If Ring doorbell cameras were giving bogus probable cause at the rate these dogs were, the NYT would be all over it. But because it's not novel no reporters can be bothered to write about it...

ClumsyPilot5• 5 years ago

"They asked 554 mock jurors to read a story about a dog detecting the odor of drugs on a suspect’s car even though a later search turned up cash but no evidence of illegal activity. Lit says the scenario was meant to sound preposterous. Nonetheless, two-thirds of jurors surveyed said they would have found the suspect guilty."

There is a real problem right there. Looks like you could get a conviction based on a fortune teller.

biofox• 5 years ago

To me, it just highlights that there really ought to be some basic comprehension and reasoning test for jurors.

annoyingnoob5• 5 years ago

I sat in court once for 3 days as a potential juror. I was the last person that was not selected for the jury. There is a process like what you describe called voir dire. Both sides have an opportunity to interview jurors and can ask to eliminate some. In the case I watched there was a potential juror that had recently become a citizen and was very interested in being a juror, but she was also not a native English speaker and was clearly not following everything that was going on - both sides wanted her gone right away.

You still get a jury of citizens from the local community, which is the point.

MomoXenosaga• 5 years ago

I disagree with the jury system but as I understand it American jurors are supposed to reflect society.

If you only want smart people you leave it to the judge.

boomboomsubban2• 5 years ago

So do you think it's completely impossible that you would be part of the 2/3rd that found the fake suspect guilty? I don't think basic comprehension or reasoning prevent you from falling for these kinds of arguments, and rather than try to limit possible jurors we should just bar unscientific evidence.

clairity• 5 years ago

you’ve just failed your own test, since you don’t understand how and why a jury works. if you did, you’d comprehend that the better solution is expanding juries, not testing jurors. i’ll let you work out why that would be the case, and perhaps come up with an even better option as a result.

refurb• 5 years ago

So I sat on a 6 week felony jury and had low expectations but was pleasantly surprised - fellow jurors took it very seriously, the judges instructions are clear and written at a layman’s level and the “presumption of innocence” dominated.

Anecdotal but it really gave me faith that juries do a pretty good job.

AlbertCory5• 5 years ago

To me, it just highlights that if you've never served on a jury, your opinion is merely prejudice against those who are (presumably) less credentialed than you are.

pfortuny• 5 years ago

A jury of peers is a jury of citizens with no regard to their upbringing, education, rece, religion, or any other thing. Citizens.

throwaway0a5e• 5 years ago

I would really like to see a much bigger juror pool and results sorted by on zip code.

I suspect I can do an ok job predicting relative conviction rates based on well known location and demographic based stereotypes of authoritarian tendencies and attitudes toward drug use.

bumby• 5 years ago

Would this change if instead of a drug charge it was a murder charge as in the article?

HenryKissinger• 5 years ago

People trust dogs too much.

Dogs are friendly and make good pets and all but people need to draw a line.

Ensorceled3• 5 years ago

"My dog would never bite. He just snarls and snaps like that when he's nervous."

"My dog is a good dog at heart, but she's a rescue so you have to make allowances."

"My dog is only growling because you are showing fear."

"If you got bit, you must have done something to the dog. No dog bites without good reason."

People "trust" dogs to a ludicrous degree, even dogs that clearly can't be trusted.

cm2187• 5 years ago

Though that’s without the defence having a chance to tell the jury why they should disregard that fact.

But in the end, I don’t think judges are doing any better. The only reason the justice system works is because the vast majority of cases aren’t very complicated and have physical evidence. Colombo-style crime mystery is fairly rare.

lettergram• 5 years ago

This is why they have an entire field focusing on finding unbiased juries. They conduct interviews, etc

vkou2• 5 years ago

Unbiased relative to the background level of nonsense that the general public believes?

Or unbiased relative to some platonic ideal of a jury?

pessimizer2• 5 years ago

That's not why attorneys interview jurors. They're trying to find jurors who are biased but not biased enough to retrospectively throw the entire trial into question.

taneq3• 5 years ago

If you’d spent the last 40 years telling jurors that fortune tellers were the gold standard for detecting drug mules then yeah. The issue is that we’ve been sold the idea that drug dogs are effective, not that jurors are gullible.

chmod6003• 5 years ago

And we believe that dogs are noble beyond any doubt. But maybe the dog just wants the handler to smile, take her home, and give her a steak?

ClumsyPilot5• 5 years ago

"The issue is that we’ve been sold the idea that drug dogs are effective, not that jurors are gullible."

But we know that they are gullible, they take police as a trustworthy source of information and assume police would not mislead them. Nimerous studies have shown that.

giantg22• 5 years ago

That's because society, the courts, etc believe people are guilty before it is proven. And "reasonable" doubt is basically whatever they say it is, which is often unreasonable.

Innocent until proven guilty is a cruel joke.

kingcharles6• 5 years ago

On the flip side of this, which you don't hear about very often, during my eight years locked up I saw a decent number of jury trials where the jury aquitted a clearly very, very guilty defendant.

So, I would always go with a jury trial even if you are horribly guilty as you might as well just roll the dice on the jury verdict.

One example: A guy was thrown down onto the grass in front of his house by the cops during a drug arrest. When they picked him up a gun was lying on the grass under him. He told me it fell out of the waistband of his pants. At trial his lawyer argued that the gun was already randomly lying on the ground when the police threw him down. The jury acquitted him of the gun possession charge.

ClumsyPilot5• 5 years ago

"At trial his lawyer argued that the gun was already randomly lying on the ground when the police threw him down. The jury acquitted him"

Blackstone's ratio is the commonly accepted benchmark: "It is better that ten guilty persons escape than that one innocent suffer"

It's job of the prosecution to make sure their proof is "beyond reasonable doubt", which is obviously in the eye of the beholder.

amelius3• 5 years ago

Similar question:

Should Machine Learning be allowed to drive a car?

iso1210• 5 years ago

Not at all a similar question, the outcome of machine learning of driving a car can be judged on objective results (incidents, accidents, injuries, deaths, etc)

amelius3• 5 years ago

Sorry, can't agree. You can validate the performance of dogs in a test setup, just like you can validate machine learning in a sandbox environment.

swilliamsio• 5 years ago

That similar question and this entire thread reminds me of xkcd 2173. [0] We obviously already let neural networks detect drugs and drive cars, though the former gets to wear a nice fluffy fur coat.

[0] https://xkcd.com/2173/

goodpoint• 5 years ago

If ML had well documented biases towards hitting people on purpose, then no.

amelius3• 5 years ago

https://news.ycombinator.com/item?id=28219948

sebow2• 5 years ago

No.

It should be taken into account, but it shouldn't be considered "definitive" evidence (compared to something like a video footage).

kolanos3• 5 years ago

A drug sniffing dog is a violation of the confrontation clause in the sixth amendment. It provides that " in all criminal prosecutions, the accused shall enjoy the right…to be confronted with the witnesses against him." You can't cross-examine a dog.

NovemberWhiskey• 5 years ago

Much as I think there's a lot of quackery going on here, I think it's pretty clear that the dog handler is the one giving the evidence and being subject to cross-examination. The dog is just a tool that is being used to achieve an investigative purpose, like a DNA test or radar gun.

kolanos3• 5 years ago

The dog handler's "evidence" cannot be based on the dog's behavior. We can't question the dog as to why it behaved the way it did. As such, it is a violation of the confrontation clause.

AlbertCory5• 5 years ago

No, but you can run your own test, where a neutral "handler" whom the dog has never seen runs his/her own scent tests on the dog's ability. Which should be the defendant's right if dog evidence is presented against him.

MertsA3• 5 years ago

Dogs being able to identify the scent of illicit substances is not the question at hand. The issue is that it's all too easy to manipulate the dog into signalling whatever the handler wants. A neutral handler would probably show that the dog can identify presence or absence of some drug with much better than a coin flip odds.

In practice there's a terrible accuracy problem with drug sniffing dogs. In theory there's a somewhat reasonable accuracy of drug sniffing dogs. The judicial system likes to plug their ears and pretend that drug dogs are accurate and ignore the obvious fact that dogs are fallible social creatures that first and foremost are trying to do what their handlers want them to do. If the handler wants them to check for drugs, great, but when the handler doesn't care about accuracy and just wants them to "find" drugs, it should not be shocking that the dogs behavior reflects the handler's wishes.

https://reason.com/2021/05/13/the-police-dog-who-cried-drugs...

https://en.wikipedia.org/wiki/Florida_v._Harris

kortex3• 5 years ago

This should be really easy to settle, like basic undergrad experimental design, double blind study. Hardly a thorough literature search, but I found this reference.

On average, hidden drug samples were indicated by dogs after 64 s searching time, with 87.7% indications being correct and 5.3% being false. In 7.0% of trials dogs failed to find the drug sample within 10 min. ... Dogs were equally efficient at searching in well-known vs. unknown rooms with strange (i.e., non-target novelty) odors (83.2% correct indications), but they were less accurate when searching outside or inside cars (63.5% and 57.9% correct indications respectively).

So that tells me two rough indicators:

- There's a ~5% chance a sniffer dog/handler could put away an innocent person. This is a lower bound, the risk is higher with cues/confirmation bias

- There's only about ~60-80% sensitivity in unfamiliar situations, of what I presume are pretty fresh samples.

You don't have to bust out Bayes' theorem to realize that a cadaver dog accurately hitting scents in the woods months later, stretches credibility.

https://www.sciencedirect.com/science/article/abs/pii/S03790...

Zigurd• 5 years ago

Dogs are amazingly good at detecting things like cancer in double blind experiments. There are good reasons to think dogs can detect lots of things.

That's not the problem:

1. No MD would order cancer treatment on the basis of a dog's sniff.

2. Analysis of all kinds of evidence, including the use of sniffer dogs, should not be managed by police departments, or even prosecutors. There have been many scandals of police evidence handling, even by supposedly professional evidence labs, have mishandled evidence and innocent people have been imprisoned. There needs to be independent analysis of evidence, incentivised to find the truth even if unfavorable to police and prosecutors.

3. Cops lie.

flatline• 5 years ago

Unfortunately there is more than statistics at play: http://www.wayne-county-forfeiture.com/content/drug-dogs-and...

jakelazaroff• 5 years ago

The problem with a purely quantitative analysis of a dog’s theoretical capabilities is that they can be trained to give false positives. See for example this Reason article [1] about dogs searching for drugs:

As a drug detection dog, Karma kept his nose down and treated every suspect the same. Public records show that from the time he arrived in Republic in January 2018 until his handler took a leave of absence to campaign for public office in 2020, Karma gave an "alert" indicating the presence of drugs 100 percent of the time during roadside sniffs outside vehicles.
Similar patterns abound nationwide, suggesting that Karma's career was not unusual. Lex, a drug detection dog in Illinois, alerted for narcotics 93 percent of the time during roadside sniffs, but was wrong in more than 40 percent of cases. Sella, a drug detection dog in Florida, gave false alerts 53 percent of the time. Bono, a drug detection dog in Virginia, incorrectly indicated the presence of drugs 74 percent of the time.

[1] https://reason.com/2021/05/13/the-police-dog-who-cried-drugs...

kortex3• 5 years ago

Right, that was being generous, assuming the ideal conditions, double blinded, so that any cues from the handler will only add noise and move the needle towards chance.

In actuality, the police drug dogs are basically just an excuse to get into the vehicle, so they can justify anything they find post-hoc and ignore all other intrusions.

iamthemonster• 5 years ago

If I was a dog who was given a treat by my owner every time I alerted correctly, but never received a treat any other time, then you better believe I'll be alerting every time.

6gvONxR4sf7o• 5 years ago

If you do want to pull out bayes theorem, 5% of indications being false in an experimental setting could be way worse in a situation with a lower base rate. (I couldn’t find the base rate in a quick scan of the paper)

kortex3• 5 years ago

Yeah I didn't compute the Bayes factor mostly because I'm on mobile and was heading out the door. But when the prior you are plugging in is "did this person murder their son", you need a pretty hefty likelihood ratio to push that into the territory of "lock this person up for life".

If you have p(D|H) = 0.8 (hit rate) and p(D|H') = 0.05 (false positive), that gives an LR (likelihood ratio) of 16.

But that's if you had like, a perfect detection situation. I'm not a fan, but it's probably enough to push a search into probable cause°, since if you find contraband, you now have field drug tests, which are also kinda crap°°, but it pushes your hypothesis. Then you seize the drugs and have the highly sensitive and selective lab tests of chemical analysis at your disposal. You have a chain of tests updating your prior.

We have no such chain in the cabin test. It's just "dog signaled" -> some other circumstantial evidence -> jury convicts of murder.

° - that's assuming the handler is being unbiased, vs just giving the cops an excuse to search.

°° - field drug tests basically only catch broad classes of molecules or functional groups, when performed correctly. Like drug dogs, often they too are just theater to intimidate/coerce more searching

ufmace• 5 years ago

I don't think you can really judge dogs in general based on statistics like this. We can't read a dogs mind and don't know all of their motivations. You can try and do standardized training and testing, but any particular dog might not "get" the training properly, or might decide to do something a little different one day for reasons we don't understand.

I'm thinking, it's okay to have a dog to guide you to things, but ultimately, the thing is either there, or it's not, and that's what any convictions should be based upon. If a dog guides you to an actual body, fine, do all your analysis and charges based on that. This whole business of the dog claiming that a body was there months or years ago but we have no idea where it is now seems like bullshit, and I don't think anyone should ever be convicted on that alone. Ditto drugs - search dogs may be imperfect, but ultimately the drugs are either there or they're not. If the dog guides you to actual drugs, charge away I guess. If there aren't any drugs and the dog was wrong, then no charges. It's a little bad that someone got searched based on a false indication, but not nearly as bad as an actual conviction and sentence.

elmolino89• 5 years ago

Imagine a drug detecting dog going through 100 apartments. No drugs were found in any of them but say in 5 of them the dog sat on his bottom /barked/did whatever constitutes something resembling "drug odor detected". I hope no judge will put the people who did live in these apartments up to half a year before in jail.

alphabettsy• 5 years ago

Not just that, but the dog’s indication is used to do incredibly invasive searches that the government does not have to incur the costs of the damage done.

They will essentially toss a home or vehicle like a burglar, even going as far as cutting open things like seats and removing parts.

MertsA3• 5 years ago

Not only will police occasionally destroy vehicles in a search without being liable for the damages, often any insurance policies also do not cover it. You can do absolutely nothing wrong and obey the law and buy comprehensive coverage and gap insurance and still have your vehicle destroyed by the cops because they managed to convince a dog to scratch your paint, and you're the one that's stuck with a totalled vehicle and a huge loan to pay off.

https://reason.com/2021/04/23/the-cops-trashed-her-house-she... https://jalopnik.com/new-jersey-cops-destroy-bmw-searching-f...

FpUser2• 5 years ago

Whatever the fuck happened to "beyond the reasonable doubts". Any reasonable person should have such in this case.

vkou2• 5 years ago

You'd be surprised by what reasonable jurists believe is damning evidence.

FpUser2• 5 years ago

Well, not really. After reading about some cases I've long lost any trust I may have had in the system. It might be ok in general (do not have enough real stats) but when it comes to fucking up particular person's life without any consequences the government / their institutions shine.

dzink• 5 years ago

There was a news mention months ago of a dog at a police department that “detected” drugs 100% of the time ot was deployed - the running joke was that the dog was brought in whenever they wanted to detain anybody. A dog is only as good as it’s training and there is plenty of room for abuse in the system.

Buttons8403• 5 years ago

Mine as well just let the officer sniff. Yeah, human smell isn't up to the task and will be wildly inaccurate, but that's not a problems for dogs, so why should it be a problem for humans?

grae_QED• 5 years ago

Fun fact: The Stasi had a scent library[0]. They would regularly use dogs to identify and arrest people.

[0] https://theresagameinthat.blogspot.com/2011/06/stasi-scent-l...

adam-p• 5 years ago

This seems analogous to using an AI. We trained it, it seems to mostly do what we want it to do, but we don't entirely understand how (or what cues and confounding factors might be in play that wouldn't like if we did know about).

Do jurors respond the same way to a black-box "expert" if it's not furry?

deepsun3• 5 years ago

Unlike AI, sniff dogs are trained to bark at a sign from it's owner. So whenever police wants to search you, they make a sign to the dog, and it barks.

betwixthewires3• 5 years ago

This is patently false, you get your information from movies and not real life. Sniff dogs sit when they detect what they're looking for, they do not bark.

dehrmann• 5 years ago

It's worse than AI. At least with AI, we know the inputs, can get glimpses into how it works, and was build with a specific intent. Fido just wants his treat.

dymk• 5 years ago

Nah it's much more similar than you're implying. Fido is doing stochastic descent to get his treat. Humans are determine the loss function for both AI training and dog training.

harha• 5 years ago

This shows a greater dilemma: we don’t want a caste of legal experts making fundamental decisions for an individuals life, but the alternative of getting a jury consisting of the average population who lack sufficient education also doesn’t lead to an outcome one would want when on trial.

I wonder if there is a middle ground solution where the jury decides on parts of a trial independently without knowledge of other parts, e.g. only on admitting a piece of evidence.

LightG• 5 years ago

No. I've not read tfa but being stopped under suspicion of carrying drugs at an aiport because a sniffer-dog got a bit too excited about a humous sandwich I'd just bought tells me answer to this question.

Also raising the question of whether a sniffer-dog can be manipulated by its handler.

Questons, questions ...

renewiltord• 5 years ago

Dogs are cuddly AI. The first time we make a Convolutional Neural Network warm, cute, and fuzzy we will have created the perfect tool of the police state.

Which naturally means that we already have created the perfect tool. It’s a dog.

Nothing wrong with police in general. They’re just like dogs in their own way. Dogs get treats when they say there’s a drug, irrespective of the actual presence. Cops get treats when they a guy gets arrested or convicted, irrespective of guilt.

marsdepinski• 5 years ago

It's science!

NHQ• 5 years ago

A dog cannot take the stand and be questioned, so how can it provide witness?

chmod6003• 5 years ago

We celebrate "science", but we skip the scientific method. Observational studies, unreprodcible experiments, and non-falsifiable models have taken over.

Not just taken over in pop science, but actual scientific fields.

So can we really blame judges and juries for taking a few intellectual short cuts? Yes, I suppose we can, because an individual is at risk and we should expect better.

But the scientific community should take some blame for compromising on the core of science while continuing to fly the banner of science.

chmod6003• 5 years ago

Where can I find I good test for scientific and statistical reasoning? Something a layperson could easily take, like a bunch of word problems that aren't numbers-heavy?

AlbertCory5• 5 years ago

I dismiss anything Corcoran says after I read this:

There’s so many missing people that need to be found. You know, it’s just sad. And that’s really what it’s all about: If we can recover somebody, so their family can know the ending. That’s really what it should be about.

So this is totally about the problem, and not the "solution." If you're attacking her, you're against families finding out what they need to know. BS.

Maybe if she said something about the reliability of the method, and not the need for ANY method, I'd pay more attention.

ghufran_syed• 5 years ago

Seems like the appropriate defense strategy should be to arrange their own cadaver dog analysis which shows completely contradictory evidence to that of the the prosecution dog.

AlbertCory5• 5 years ago

After reading all the comments (a dirty job, but someone has to do it), I reach this conclusion, which doesn't require any massive changes to our justice system:

The constitutional right to "confront" a witness against you should be interpreted as the right to run your own unbiased tests on any dog whose results are used. This would mean a neutral handler whom the dog doesn't know conducts double-blind tests on the dog. If the dog is found to be unreliable by the judge, then the evidence is discarded, the same as if a warrantless search had been conducted.

bayesian_horse• 5 years ago

Dogs learn to please their handlers and react to their cues.

Any reliable use of dogs (or any animal for that matter) must have the paranoid kind of quality control known from machine learning. Worse, actually, because an animal is a learning agent, shifting its model.

Should a dog's sniff be enough to convict a person for murder?

Related Stories