AI screw-up results in man being fined $400 for scratching his head

Right, but we don’t need more information, we need to not issue tickets in situations when we can’t know if a violation was committed.

And any system where false positives generate income for the operator needs a LOT of extra scrutiny.

5 Likes

“It may well be that the training dataset contains few or no photos of people sitting with an empty hand on their ear. In that case, it becomes less important for the algorithm whether a phone is actually held in the hand, but it is sufficient if the hand is close to the ear. To improve this, more photos should be added where the hand is empty.”

It’s a lot of fun finding all of those and then trying to plug the gaps.

There was also a lot of this discussion when people wanted to define driver drowsiness as X seconds of eye-closure. Had to provide that most eye closures are, in-fact, not sleepiness related.

Well, the Dutch Authorities have issued Mr. Hansen a ticket, so at this point it would be really good to get information that would at least determine if his phone was being used-sending or receiving a call or was on an app of some sort- at the time in question. One thing we learned from watching Jan6 happen is that it came out that the phone company can figure out where you are when you use a mobile phone (within a certain number of feet, it’s not pinpoint accurate but of course we have a picture of him where he was,) so what the Dutch Authorities want is for the Dutch phone company to tell them whether his phone was interacting with the nearest phone tower, because then we know his phone was “being used”.

As for someguy supposing several scenarios, that:
His passenger was using the phone, while he scratched his head.
He was using the phone legally hands free, while he scratched his head.
He was listening to a podcast with data turned off, not scratching his head.
Mr. Hansen hasn’t proposed any of this. He hasn’t mounted a defense based on there being another person in the car and his passenger was using his (Hansen’s) phone or that he was using the phone hands free or that he was listening a to a podcast. He is disputing that he had his hand up to his face for the purpose of holding a phone. No, he says, I was scratching my face.

His defense is that the image taken of him in his car was not interpreted correctly. He goes on at length about how the programs used to analyze photos taken of drivers do not use training pictures of people just having their hand near their faces, scratching or not, so as to recognize this as a possible situation and interpret it as such. He goes on at length about how this could be corrected. When I read this part of the report I get a picture in my head of a team of Bloodhounds following a trail but when they get to the end of the trail there is a stinky fish there and the handlers conclude that the suspect has put the hounds on the wrong scent while he went off in another direction.

The good thing to do in this case is to ask the phone company to provide the information about whether the phone was being used. When they get this information from the phone company, one thing that could come out is that there was no phone activity by his phone. No phone activity? Then he didn’t have it up at his face to be using it. No case. Easy verdict. If there was phone activity, and if he was letting a previously unmentioned passenger use his phone, and they testify that they had the phone at the time, No case. Easy verdict. If there was phone activity, and Hansen doesn’t remember using his phone at this time (which is what he says) then his phone history will be a great help to remind him of what he was doing. The court needs to let some experts examine his phone history.

My point here is that there we don’t have to rely only on the picture. The phone was either on or off, and it keeps a history. The Dutch Judicial System can find out what it needs to know to make a defensible judgment., which is all anyone can reasonably expect.[quote=“someguy, post:11, topic:268195”]
His passenger was using the phone, while he scratched his head.
He was using the phone legally hands free, while he scratched his head.
He was listening to a podcast with data turned off, not scratching his head.
[/quote]

He’s being fined based on the picture. If you agree that the picture doesn’t necessarily show anything, why is it his responsibility to prove that he wasn’t doing the thing that the photo doesn’t show him doing?

Why not just issue a ticket to the owner of every phone that is in use on a roadway. If someone wants to contest the fine, they can explain who or what or how the phone was being used.

On a more practical level, you say his phone history could show what someone with the phone was doing, and that person could be brought in to testify. (Again, at whose expense?)

So, here it goes:
State: You were on the phone.
Tim: No, it looks like I was scratching my head.

State: Your phone was in your car, and a call was in progress to Paul at the time the photo was taken.
Tim: Hmm. Maybe it was my wife, or my friend Ted calling Paul.

State: Wife of Tim, were you on the phone with Paul while in the car with Tim on January 6th, at 4:19pm?
Wife of Tim: Could be. I do ride with Tim, and I do talk to Paul.

State: (same interaction with Ted)

State: Paul, who were you speaking on the phone with on January 6th, at 4:19pm?
Paul: Um, I don’t remember. I could have handed my phone to my husband, or to our kid, or to my friend Ken who was visiting that week.
State: WE WILL ROUND THEM UP!

Tim: By the way, my car also allows connects to the phone hands-free, so it could have been me talking on the phone legally, while also scratching my head.

I mean, the fact that the guy actually understands the tech that is being used against him shouldn’t count against him.

If he had just said, “nope, doesn’t look like I have a phone to me” would you be more convinced?

1 Like

Someguy said, “He’s being fined based on the picture.”

Yes. He’s being fined because they base it on the picture. They think they have proof he he was using a phone while driving. They think the picture shows him with a phone. It doesn’t matter if you or I think the picture is too indistinct. The proper authorities, the jack-booted Dutch Masters of thuggish oppression, say it looks like he has a phone. If asked, and I’ve not been asked, I’d probably say pay the fine, and then, Mr. Hansen, if this still bothers you, find out who you have to see to change it. I mean it’s in his field, apparently. He has ideas about how to improve it. Maybe this experience makes him uniquely qualified to bring about needed change.

If Mr. Hansen manages not to be dragged off to a Gulag this time around, he should consider doing that Gandhi thingy about being the change he wishes to see in the world. I’m only seeing a problem with where it stops. For instance, maybe when the government has designed an image to take into account that a person could be scratching their face, and it is recognized as “not a phone” maybe we can’t stop there. We’re gonna need a bigger picture file. I guess this could expand to just about anything that can be done with a hand: wiping? tapping? massaging? or held in a hand: blusher brush? candy bar? key fob? baby hamster? or not held in a hand: Halloween mask? Face-hugger? or something that looks like a hand: reflection of a tree on the windshield? or someone else’s hand: Ted? or something that doesn’t look like a hand but no one wants to go on record saying, “It doesn’t look like a hand” because it might be someone’s hand but they have an ugly hand, and that might hurt someone’s feelings about their hand’s appearance, so we have to have an image of an ugly hand, so it is recognized as “not a phone.”

I’m interested in seeing if Mr. Hansen can get his way in court, but I don’t think that’s going to happen. Compensation for Mr. Hansen’s time is definitely not going to happen. The method the authorities use to analyze the traffic photos is an established methodology here… and it’s not being used in a discriminatory or predatory way, so Mr. Hansen can’t complain that he’s being singled out. He may not get any preferential treatment either. But it’s an interesting case, so please update us, Mr. Pescowitz, as this proceeds? Thanks!

If a false positive are profitable to the people who run the system, it’s predatory unless there is sufficient counterbalance.

3 Likes

We knew that before Jan 6th, it wasn’t really a secret. The question is more whether Dutch and EU law would allow such privacy violation

2 Likes

Well, someguy, first to answer your question ”If he had just said, ‘nope, doesn’t look like I have a phone to me’ would you be more convinced?”, I would say, “No, I’m thinking he has to present evidence in a court of law. “

And also you asked: ”If you agree that the picture doesn’t necessarily show anything, why is it his responsibility to prove that he wasn’t doing the thing that the photo doesn’t show him doing?”

And I say, “It’s his responsibility to defend himself. This isn’t an ‘argument’. Courts work this way. Or he can plead nolo contendere.”

When an image in a traffic photo is identified as “using a phone” and a citation is issued, the judge is very likely to side with the system where there are trained analysts making the judgment. That Mr. Hansen says there was a false positive is not evidence. He has to prove he was not driving and using his phone. What is his proof? If he can’t show that then he’ll be found guilty and fined.

Maybe he’ll suddenly discover that there is a way to prove his case or maybe the judge will surprise everyone and take the word of a stranger who has motivation to not pay a fine over the people who work for the police and have been analyzing traffic photos without incident for years, and y’know, everyone considers they are experts.

I’m interested in how this turns out and hope to hear about the resolution. It would be especially nice to hear that Mr. Hansen plans to take this up at the level where he can effect the changes he thinks should be done. Maybe Tony Stark would give him a suit.

And, a third response to someguy:

Someguy said,”If a false positive are profitable to the people who run the system, it’s predatory unless there is sufficient counterbalance.”

No-wait!-you mean the city here makes MONEY off these fines???!!!??? Well, NOW there are pearls just ALL OVER my lovely divan. And (sigh!) of course: /s

@Doctor_Faustus I started to look around for information on whether the Dutch Court System admits phone tower ping information, but then I found several articles that say it wouldn’t make any difference. It turns out experts are all over the place on whether courts anywhere should admit cell phone tower ping information.

I kinda thought Mr. Hansen here could, if he knew that he had not used his phone at all during the time in question, he could ask the Court to obtain the phone tower ping records and that would make his case. But no. It turns out if the nearest phone tower is all busy the system will send your call to another tower, so it can always be argued that maybe it was the next tower, farther away, or maybe the next one, ad infinitum… So, it’s nowhere near the type of clear hard evidence a Judge wants. Huh.

Here’s three articles:

Cell Site Evidence is Junk Science (and a Recent DC Opinion Could Help Keep It Out of Court)

https://kmlawfirm.com/2020/03/10/cell-site-evidence-is-junk-science-and-a-recent-dc-opinion-could-help-keep-it-out-of-court/

March 10, 2020 By Sara Kropf (attny)

“A cell tower in a crowded area may be too busy to handle a call or text message. A phone will automatically find the closest available cell tower to use. The next closest one may also be too busy to handle the signal. The phone will keep looking for a site that can handle the signal. And that cell tower could be a bit of a distance away.”

Are There Really Flaws in Cell Phone Location Evidence?

07 October 2019

Larry Daniel, EnCE, DFCP, BCE, CTNS, CWA, CTA, CCO, CCPA, CASA Spencer McInvaille, CWA, CTNS, CCO, CCPA Eric Grabski

The Guardian further discusses this issue in a recent article, stating "The system has also linked phones to the wrong masts, connected them to several towers at once, sometimes hundreds of kilometres apart, recorded the origins of text messages incorrectly and got the location of specific towers wrong."2

Saint Louis University Public Law ReviewSaint Louis University Public Law Review

Volume 33, Number 2, The 2013 National Conference on Prison Higher Education Hosted by the Saint Louis University Program (Volume XXXIII, No. 2) Article 17 2014

Ping! The Admissibility of Cellular Records to Track Criminal Defendants

Alexandra Wells awells8@slu.edu

https://scholarship.law.slu.edu/cgi/viewcontent.cgi?article=1074&context=plr

“While triangulation is common, it is not the only method of determining

the location of a phone. What is most often the source of contention, and the

focus of this article, is a simple mapping system. 51 This is where someone,

often the detective in criminal trials52 or a records custodian from the cellular

company, 53 takes historical cell tower records and makes a map of the calls and

the towers that received those calls.54 This is all based on a belief that a call

will ping to the tower that is closest to the cell phone.55 Thus, by mapping the

towers which accepted the specific calls, one can locate the general vicinity of

the caller.56 However, this belief has one large fundamental flaw: cell signals

go to the tower with the strongest signal, which is not always the cell tower

geographically closest to the cell phone.”

And here I thought the authorities could easily put people in the Capitol Building on Jan 6 based on phone tower ping records of all those people sending phone selfies of themselves in the Capitol on Jan 6. But I read that they were able to use the actual selfies, though, and that was easier than the ping stuff.

1 Like

Sounds like a Babylon 5 episode:

“The truth is sometimes what you believe it to be, and other times what you decide it to be. My task is to make you decide to believe differently.”

The court already has pictorial evidence from authorities that they-the authorities- base their decision on. Mr. Hansen will have the opportunity to present evidence that contradicts that evidence. I’m interested in hearing what happens. I hope there is a follow-up.

This whole thing is a larger waste trying to dredge up every single data point when they could… just drop it.

None of this makes the public any safer. We’re well into the land of diminishing returns.

Not since Antoninus Pious.

Ei incumbit probatio qui dicit, non qui negat, etc.

1 Like

Nope. Wiki History shows no record that Antonious Pious was ever issued a citation for driving while using a cell phone. Mr. Hanson was issued a citation for driving while using a cell phone, though, and the authorities think they have pictorial evidence of it. They say they see a phone. If Mr. Hansen shows up with no acceptable proof that he was not on the phone, he will likely be judged guilty and the fine imposed. No one has given me any good reason why the judge would accept: “I didn’t do it! There was no phone!” as reason to dismiss the case. A judge sees people all the time who claim, “I didn’t do it! There was/were no-----!”(fill in contested item here- some things judges have probably heard: gun, knife, money, drugs, secret meeting, documents, bribe, dead body, stop sign, pee tape, telltale heart, open window, nooky, poison, contrails, bone saw, person, woman, man, camera, TV) and they say, “Guilty! (Bang!) Bailiff!”

Once again, It would be a wonderful contribution to furthering justice if Mr. Hansen could make it his goal to fix the system. He could find himself the hero of folk songs.

Is anybody going to do a follow-up on this?

… and yet here we are, in a forum where people post their opinions :thinking:

3 Likes

Have you really never learned that people are presumed innocent until proven guilty, and that the burden of proof is on the state not the accused?

If all the state has is a picture they say shows Hansen holding a phone, and it isn’t clear that he is holding a phone, that should be the end of the case.

Well, this is probably my last post as this discussion will close soon. So here goes again:

I keep having people post to me that the authorities must “prove” he was holding a phone. They think they have already done that. They say they have a picture of him holding a phone. He says he plans to contest this proof by saying it is a False Positive based on inadequate training photos of simulated photos showing people scratching their head. The problem here is that there is a shadow across his cheek which could reasonably come from an object he’s holding. The AI says it’s a phone and the human analyst concurs.

If I were advising him I would say to go with the idea that he’s holding something. But here’s where he contests it. He has something in his hand but it’s a key fob (or some other plausible thing he could be holding in his hand ) He could actually have a friend take pictures with several objects and go with the one that looks closest to the police photo. Let’s say it’s the key fob. He could bring that photo and the key fob to court for the judge to look at. “Your Honor, I remember it now-I had my keyless entry fob ring hooked over my thumb and it was in my hand…like this!” (shows photo and demonstrates with a key fob.) If the photo looks as indistinct as the police picture, he could get the case dismissed.

The judge still might need more, and if Mr. Hansen knows for sure that he did not have a phone on, he might ask the court (ahead of his court date) to get him the cell tower records from several nearby towers as a second proof. The judge might be willing to consider two forms of proof together as sufficient to contend with the police evidence.

What I don’t think is going to work is a statement that, “I didn’t do it ”, along with a lengthy theoretical discussion of why everyone but him is wrong.

I notice that I’m chiefly arguing with guys here. I notice that when the story was presented no one asked if the original headline “AI screw-up results in man being fined $400 for scratching his head might not be re-written as “man claims (bolded emphasis for purposes of showing a difference) AI screw-up results in man being fined $400 for scratching his head” to avoid bias in reporting. Nah. Everyone is in love: he’s a tech guy, and these are valentines. One guy fesses up: “The fact that this guy is an ML engineer is borderline Black Mirror stuff. I secretly love this.” Whoa.

I remain eternally grateful to my husband’s two older, very sharp sisters for providing the necessary rebuffs during his formative years that enable my husband to consider that he’s not right all the time just because he’s a guy with an engineering degree.

This topic was automatically closed after 5 days. New replies are no longer allowed.