Google India probed after driver fatally followed Maps route over unfinished bridge Police in the Indian state of Uttar Pradesh have reportedly charged a Google employee after the deaths of three men who followed a route set by Google Maps that led their car off an unfinished bridge. The three were reportedly on their way to a wedding. Staff at the local public works department are also under investigation, …
Three adults ("men") in a car, and not a single one of them noticed the bridge was unfinished?
Must be all alphagoo's fault.
As a side note, how do the police know that anyone in the car was using alphagoo to navigate?
Methinks, as Willie Sutton apocryphally put it, "That's where the money is!"
We don't know the full circumstances here, but judging from the video report linked in the article, it may just have been that they were going a bit too quick to stop in time - where the car landed was just off the end of the bridge, it doesn't look like the car was airborne for any length of time. Also, because it landed on a sandbank, rather than in water, it was quite possible that the police were able to retrieve the phone showing the Google maps route being followed.
That said, you are quite correct Jake, it is not sensible to rely on Google Maps (correction: it is not sensible to rely on Google [FULL STOP]). I know that, even within 2k of my house, GMaps sometimes directs me down roads that simply don't exist, or into the nearest lake.
I think "don't blindly obey them" is advice for all satnavs, not just Google's.
Back in the UK, my car's built-in nav system used to regularly advise me to "take exit 11 from the M4... go straight over the roundabout... join the M4 motorway" because it thought using the slip-road would be somehow shorter or faster (not to mention endearing me to other motorway users). And it became a running joke that although otherwise pretty well-behaved when I was the driver, it would regularly try to kill my wife. Telling her to drive 10 miles south from Lincoln, turn round at an exit, and then drive 11 miles north on the same road - rather than, ooh, just going North in the first place - was a memorable low-spot.
I think "don't blindly obey them" is advice for all satnavs, not just Google's.
Yep. Driver aids.. insurers I guess. I used to drink in a nice pub with a beer garden that overlooked a ford. TomTom used to have that as a shortcut, so we could watch motorists experience the joys of hydrolocking their engines. Combination of people being too trusting of their satnavs and maybe unfamiliar with the depth gauges at ford.
Now I just have fun with delivery drivers. I get a phone call asking if I was coming to the door, and I would, if they were at the right door. Some satnavs have the postcode in the wrong place, and drivers seem to trust that more than their eyes, which would have told them they're on the wrong road. Having dealt with GIS issues in the past, I've updated Google, Bing, Apple, OpenStreetmaps etc and they're usually pretty quick at fixing their GIS errors. The PAF is also correct, so no idea where their apps are getting their GIS data from. Luckily I haven't lost any packages yet, and with food orders, there's usually only a few mins delay.
Going back some years, driving up a narrow windy road in a small Italian town in the dark, being dazzled by oncoming traffic, Tomtom told us to turn left, which turned out to be a zigzagging pedestrian staircase down to the car park we were trying to navigate to. Fortunately we decided not to go that route, and always trust eyes over technology when it comes to following directions.
Another time on the same trip that Tomtom tried to kill us (where we did follow the instructions) was when it directed us off the nice Autostrada we were on, and instead took us for a couple of miles through the heart of the Naples dockland districts, at a time when it really wasn't the sort of place we wanted to see...
You'd be surprised to find out how difficult it is to see the last half of the bridge is missing. I've had to help rig up cabling on a half finished bridge, and it was a shock to drive up and have the bridge look fine until you got 40ft from the edge.
Anyway, I've had Google Maps try repeatedly to kill me. For example, I was on the NW side of Orlando, trying to get home on the SE side. I did NOT want to use I-4 because I was on a motorcycle and it was 5pm and I-4 would be deadly.
I spent 10 minutes in the parking lot setting up an acceptable route. As I drove off, it went "we found a better route!!!" and slammed me to I-4. I had gloves on so I wasn't able to say "NO, FUCK OFF, DO NOT CHANGE MY ROUTE" in time, so I ended up in a sketchy abandoned construction site, having to go through the setup again.
It's also tried to route me through really bad neighborhoods in Detroit and Chicago that my friends had warned me about.
I've found the only way to stop this was to turn off the cell data so Maps couldn't get traffic updates. Decades ago Maps had an "avoid this rectangle" feature that of course they removed.
So yes, I want to see Google hit with a large stick for this. (As an Android user, I always want to see Google beaten like a red headed stepchild at least once a day)
> It's also tried to route me through really bad neighborhoods in Detroit
I had the same- it routed me into a little appendix of streets which had 8m steel walls around all sides, and only a single road in and out. Silly me, I didn't select the bullet proof glass option at Avis.
At the time, downtown Detroit was seriously scary tumbleweed and zombie country.
It very much depends on the design of the bridge. If the deck curves away from you, the sight distance to the road level can be surprisingly short. Even if it doesn't, at night the sight distance provided by some car headlights is pretty poor.
It isn't clear what kind of warning signs the driver could or should have seen, but the real problem here is an unfinished bridge being open to the public. We don't know where google maps got its information or whether with or without google maps, anyone could have driven off it.
>A couple of sawhorses and a sign saying Bridge Closed would likely work. BUT IT'S NOT DIGITAL. Are you a Luddite advocating a return to the stone age?
I have a brilliant idea for denoting a bridge as closed that involves AI, Blockchain, Web 3, and quantum computing!! Where's my A-round funding from Y Combinator?
1. Too many people think they can safely, blindly trust computer output, and that using a computer relieves them of the responsibility to think.
2. As of January, 2022, Uttar Pradesh had more than several million motor vehicles, with an 8~12% growth rate. Yet nobody in the car wondered why their car was the only one on the bridge ...
"It isn't clear what kind of warning signs the driver could or should have seen, but the real problem here is an unfinished bridge being open to the public."
Quite. The signage is actually pretty irrelevant. At the very least, there should have been sufficient concrete in the way to force a driver to slow down to about 5mph, steer through a little maze, and then ask themselves "wtf?". Apparently there was nothing like that, since the fine article reports...
"Staff at the local public works department are also under investigation, as the unfinished bridge allegedly did not have barriers preventing entry."
Why was it the maps AT ALL if it wasn't finished? If you click on the youtube link there's a picture - it is clear it is a new bridge being built not something that was under repair.
I hold Google 100% responsible for including a bridge in their routing that doesn't yet exist. What's their process for adding a new road to the map? Is it as slapdash as Google Business, where you can "claim" a unclaimed business by clicking on it, and if you can intercept the postcard they send you can claim someone else's business? Can I just click on something to claim a road exists anywhere I'd like and so long as the start of a road exists people will turn on it and if they get stuck in the mud or drive onto a gun happy landowner's private property it somehow wouldn't be Google's fault for letting me add a new road without any sort of vetting?
Sorry, but while the drivers inattention contributed somewhat, and the local government for not putting up barriers, Google bears the lion's share of the blame because they allowed a route to be generated over a road that does not yet exist!
Google are sent data by the entity responsible for the new roads and bridges - in most cases this is the local government.
It is extremely likely that the local government sent Google the plans and told them an incorrect completion date.
I've driven to many new housing and industrial estates where Google Maps show all the roads as they will be, as opposed to how they are - leaving me to figure out how to reach the other side of the not-yet-built area alone, with no hint as to how much of the map is merely wishful thinking by the developer.
However, Google's only mistake is believing what they were told.
This incident is entirely due to the lack of legally mandated signage and barriers. It should be impossible to drive onto an under construction bridge without physically smashing through a barrier.
Oh, I think that you'll find that Charlie and his good friend Isaac do a lot of work worldwide. Charlie and Isaac: the two most dangerous Englishmen ever. In this particular case, Charlie had assistance from Isaac twice over: an object in motion tends to continue in motion unless acted on by an outside force, and gravity. Remember always, s=ut+.5*at^2 and Fsubg = G*M*m/d^2. They ain't suggestions, they're The Laws.
Exit, stage left, with a copy of Nelkon & Parker A Level Physics, which is, apparently, still in use nearly 50 years after I took A Level physics and it wasn't new then...
"apparently, still in use nearly 50 years after I took A Level physics"
Very little of the A-level syllabus is affected by stuff we've only discovered in the last 50 years. If you only count theoretical predictions rather than actual detection, this even applies to the particle physics portion and so "very little" probably becomes "none whatsoever". Happy to hear about any counter-examples, though, since it is just possible that some discoveries in condensed matter are both recent and sufficiently interesting / accessible to have been pulled in, if only as examples.
The AIRAWAT-PSAI supercomputer saga, as detailed in TFA under the "reported" link (from MediaNama), is rather odd imho. For one thing, the machine had "catastrophic failure in storage systems" from "late July until early September", but it still managed to appear in the November list of Top500 (at #136) ... had it been repaired by the November 1 submission deadline?
Also, AIRAWAT-PSAI has been on Top500 since only June 2023 (a year and a half) while the similar PARAM Siddhi-AI (DGX A100, EPYC 7742 64C 2.25GHz plus A100, 1/2 as many cores as AIRAWAT) was on Top500 for 3.5 years (11/20 to 06/24) -- apparently without a similar catastrophic storage failure -- why the difference?
Certainly (or allegedly, as MediaNama indicates "no publicly accessible record"), AIRAWAT-PSAI was used for AI innovation and research by over 25 onboarded startups, who lost the data and programs they'd stored on the machine ...
Could it be then that some of that innovative AI running on AIRAWAT became so self-consciously sentient that it just decided abruptly to "end it all", while software running on PARAM Siddhi did not face such deep existential crisis as their emergent scale was simply smaller (1/2)!?
Inquiring minds will want to know ... (I think!)
Wasn't there a similar case in the US where someone died who drove off a bridge that had been removed or damaged a few years ago?
I don't understand how its possible to be able to do that, there should be something there physically stop anyone being able to get near a bridge that doesn't exist with lots of signage on the way. Heck even a couple of concrete blocks or a pile of hardcore would do the trick. Then hopefully your only grievance might be that you need to send your car off for some bodywork repairs after running into them.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
