Hacker News Reader

Comment by bb88 | Hacker News Reader

You may find it interesting the NHTSA is pushing for breathalizers in cars starting in 2026.

Personally I'm more in favor of smart road like technologies where a highway has a mechanical or physical properties to assist self-driving vehicles stay in line with the road. Mostly because a 100% software solution will likely not work, particularly in the midwest and great lakes states during the wintertime.

It need not be expensive either. Magnetic nails embedded in the road would maybe be the cheapest way, and would be easy for both city and rural populations to add, and be detectable with icy and snowy conditions.

View on HN

Quality steel nails are expensive, roads are very plentiful, and sensors have noise. Also, staying in a lane is not the problem. Teslas and pretty much all other quality brands sold today do it already. The problem is complex behavior like merging and routing around obstacles (like a person changing their tire.)

Cheaper and easier would be surveying and improving road markings. As for snow, well, my Tesla maintains lane position in snow fairly well, and gets better all the time. No magic nails required.

What a surreal read... where are all the cases of Tesla hitting stuff it clearly shouldn't, that no real sane driver would ever do? Also, 'maintaining position fairly well' is too low bar in situation where mistakes can and do cost lives.

I am all up for people willingly do beta testing for Tesla they even have to pay for, but it should be clearly marked and explained as so in contracts. The thing is, most folks are not aware of this at all. As things are now, Tesla's and Musk's lies (this and auto drive) cast a pretty bad picture on morals of that company.

Bb8 proposes that we should install magnetic nails in our roads.

Bdamm counters that magnetic nails would help with lane keeping, but that is clearly not the part self driving cars have the most trouble with.

As you astutely observe “ hitting stuff it clearly shouldn't” is the problem. Which magnetic nails in the road would not help even a bit.

> 'maintaining position fairly well' is too low bar in situation where mistakes can and do cost lives

Yes! And again. Maintaining position is the only thing magnetic nails in the road would help with. Bdamm was not arguing that the car keep lane position therefore it is good. Bdamm was arguing that magnetic nails would not help with the real problems.

> What a surreal read

That is exactly my opinion about your comment. Can you follow two sentences stringed together?

> where are all the cases of Tesla hitting stuff it clearly shouldn't, that no real sane driver would ever do?

Real sane drivers hit stuff all the time. Miscalculated takeovers ending up in frontal collisions (these can be really REALLY fucking nasty, two 3-ton SUVs @ 100km/h have a lot of potential energy), trucks crashing into traffic jam rear ends and squashing together a dozen cars in front of them, people mistaking stairs to a train station as a car parking ramp (this is a common occurrence here in Munich [1])... the simple thing is, as long as an instance of an object is on a road, one of the instances will eventually get hit by a stupid mistake of someone.

[1] https://www.sueddeutsche.de/muenchen/muenchen-marienhof-u-ba...

Lane following isn't the hard part of self driving cars. I wouldn't say it's trivial, but it's a minor detail compared to everything else that also needs to work reliably 99.9999% of the time.

Such infrastructure might help make cheaper lane-keep assist as standard options I guess, but at the speeds where it matters for safety, you need significant look-ahead on the trajectory rather than just lateral position error.

I suppose you could go crazy and embed some sort of NFC into the road, so that it can communicate the lane geometry. It could probably even be done cleverly, perhaps by embedding multiple nails and encoding curvature in their spacing, or maybe in the orientation of the magnetic flux itself. All that to avoid some front facing cameras and image processing, though, that can also be used to detect other cars and pedestrians.

Given that road departments often have a hard time keeping paint fresh I have low hopes of nails faring much better. Some roads get their next resurfacing in timespans measuring decades.

I've found the folks forget just how HUGE the USA is. It might be possible to do this in a few core urban areas, but out in the hinterlands where there's nothing but cropland that feeds the core many roads are maintained only when they condition of the road causes serious damage.

Urban areas are not hot either. Roads are pretty consistently underfunded in most of them, and are in disrepair due to heavy usage, frost issues, etc. none of which is very kept up with due to the sheer size of urban road networks.

Do you need everything to work reliably 99.9999% of the time? Is that the reliability rating of human drivers?

Depnds on how you count. If you consider every ~5 seconds you could have an accident, and a minor or worse accident happens once every 2,000 to 20,000 hours of driving that’s 99.99993% to 99.999993% reliability.

On the other hand if you start talking per trio then we aren’t even close to that.

Why in the world would you count in 5 second intervals? If the measure is accidents per hour you don't then divide it by 5s.

Reliability with 1 accident per 2k hours would be 99.95%.

Reliability for 1 accident per 20k hours is 99.995%.

Five seconds seems roughly how long it would take for really bad choices to cause an accident. In heavy traffic you can cause an accident by steering into a car in under a second. I am sure someone has had two independent accidents in under a minute.

An hour on the other hand is completely arbitrary, why not a day or a minute?

Any fixed timespan chosen to analyze risk is arbitrary, and you can convert between timespans using probability math. I'd say it's typical to choose the average "mission" length, because that's the most relevant to the purpose. For a self driving car, an hour would make sense. For the first stage of a rocket, 5 minutes. You don't ride in a car appreciating every 5 seconds that you haven't crashed yet, do you? Maybe if it's a taxi ride in Los Angeles!

For a system that isn't "mission" oriented, you would probably think more in terms of mean time between failure.

You're trying to talk about system reliability as a whole, while the others are talking about reliability of individual decisions or event handling.

It's like the difference between talking about all-causes mortality versus the reliability of a bullet proof vest. Some don't buy that the vest is more reliable when we are not being shot.

The average American only drives ~35 miles per day. 1 hour may represent the average time driven per day but it’s much longer than the average trip length. Which doesn’t mean it’s a bad point of comparison but distance seems like the more common metric.

Failure intervals generally use multiple different metrics depending on context. You see statistics for fatalities per year or per passenger mile as both being relevant depending on the context. A life insurance company and a car insurance company would focus on different statistics.

The natural comparison for AI seems to be each individual situation. With City and Highway driving being different enough to be considered separately. In that context ~5 second intervals seem fairly natural because the car who accurately identified everything 0.001 seconds ago is probably going to avoid maying a major mistake for a few seconds. In much the same way a human driver can blink or look at a reed view mirror without significant risk.

Decision events probably correlate better with distance than time.

Like your self driving car shouldn't be meaningfully more reliable if it slows down by 5 mph (because if it is the operating parameters are poorly chosen!).

5MPH makes a dramatic difference in stopping distance so that alone is going to make a car noticeably safer.

However, safety isn’t the only thing people care about or it would be illegal to sell cars with the downright ludicrous top speeds currently available.

A single person might die, go to the hospital, or even go to jail depending on the accident. But if it's the car's software fault... Who gets blamed? Bear in mind we are trying to make this economically sustainable.

Presumably the company making the car/software/hardware – whichever part is at fault. If one company is making all three (Tesla) it's pretty easy.

Even assuming one "event" per mile that needs to be handled satisfactorily (pothole, pedestrian, inconsistent embedded reflectors, chipped or graffitied sign, ....), 99.9999% is about the same reliability as the rate that human drivers crash in the US, and fatal crashes are around 100x less common. If, as in this comment thread, self-driving cars are being proposed as a solution to the deaths caused by human ineptitude then 99.9999% seems like an absolute bare minimum for reliability.

Well in terms of fatalities, FSD actually has a 100% reliability rating, so presumably it's ok for it to be deployed, right?

Not really:

- All FSD miles put together are still only 1/3 of the average miles needed to generate a fatal crash using a human driver.

- The vast majority of those miles are in ideal conditions.

- A cursory glance indicates there have been 15 FSD-induced fatalities.

- Fair point, guess we will see in the next 60m miles, which by the sounds of it will be in the next few months.

- Tesla's FSD in particular (being an entirely vision-based system) likely operates in less-than-ideal conditions much more often than competitors.

- Where are you finding the "15 FSD-induced fatalities" stat?

- Another 60m miles will be suggestive, but you need at least 2x-3x the ordinary fatality mile count to even have a 90% confidence that the results weren't a fluke.

- Tesla's FSD might operate in less-than-ideal conditions more often than competitors, but I'd posit that the typical human driver operates in much less ideal conditions than FSD. Humans are explicitly instructed to not enable it in sketchy scenarios and to be fully attentive when it's in control, and performance in that environment is at best an upper bound on what you might expect if deployed more widely.

- I skimmed the stats from https://www.tesladeaths.com and was hoping Cunningham's law would kick in if that was too far off base.

Autopilot is not the same as FSD. There are no recorded deaths related to FSD.

Magnetic nails embedded in the road

Volvo wanted to do that, in Sweden. They pointed out that it would also be useful for snowplow guidance. That's often done by putting up poles at the side of the road.

Volvo are just trying to get someone else to provide the infrastructure for self driving.

I live in Norway, the poles at the side of the road are not just for the snowploughs, they are very helpful for human beings too. Magnetic nails embedded in the road are not at all useful to humans and are just more stuff to maintain, especially in an environment where simply maintaining centre and edge markings is a difficult and time consuming job.

Roads should provide the infrastructure for driving.

Of course they're doing something dystopian instead of something sensible.

But yes, I bet even better standardized road markings would be a huge help for self driving cars. I wonder if there's something stupid simple like a paint additive that would make it easy for machines to track. Magnetic beads?

Then any asshole could mix up their own magnetic bead paint and cause havoc.

Probably no more than someone painting their own road lines or putting up fake signage does now [1], but if you want something that's better than what we currently have it would need to not only be both machine and human readable, but also authenticated.

I always thought it would be quite neat to have those portable roadside displays you see for roadworks and so on also communicate map and driving data (cryptographically signed by your local council or dept of transport). So you might get a sign that says "roadworks ahead, please slow down" for human drivers, and the radio beacon would communicate a change of speed limits inside a certain geo-bounded area that would cause self-driving vehicles to slow down. Or similar stuff for "bridge out ahead, turn around now" causing navigation to reroute.

[1] https://techcrunch.com/2017/03/17/laying-a-trap-for-self-dri...

Just have the human readable signs be reasonably standardized, and the self driving car computer can then read them too.

It's a lot easier and more future proof to side-channel map updates to cars through a cell network, and allows for some curation. Also, improved cell network coverage seems like a more worthwhile capital expenditure for a local government than a bunch of smart signs that only help 0.001% of drivers. Cynically, I'm sure some politician's spouse's cousin would be happy to win the contract to slap a BLE beacon into a box and bolt it onto a sign for $150k a unit, though...

There is an ongoing effort to introduce infrastructure free direct communication for vehicles, using 5.9ghz spectrum. The spectrum was previously allocated to DSRC radio technology the automotive industry pushed for, but it's been 15 years and it hasn't gone anywhere. The FCC is now letting the cell network folk take a crack at solving the problem instead.

> Just have the human readable signs be reasonably standardized, and the self driving car computer can then read them too.

This is already the case in Europe.

Are there self-driving car tests in Europe, and are they doing any differently? Seems like a potential way to test the hypothesis. OTOH I bet they could still have a lot more signs that would help. The Europe/America difference might be pretty marginal...

I've wondered about the possibility of such a nail mechanism being tampered with. Sounds a bit dangerous to blindly trust such a mechanism

Caltrans had pilot program in the mid-1990's running self driving cars that used magnets embedded in the road. They would run "caravans" of five-ish vehicles driving almost bumper to bumper. This was tested on the car pool lanes on sections of a highway between los angeles and san diego.

This was a partnership between caltrans, the college of engineering at uc berkeley amongst others.

Unfortunately I can find any readily available links, but all of the data should be public. I'm only aware of this because I worked w/ some of the technical groups on the project (very tangentially involved).

I know the magnets were always considered a problem (cost/effort), but there weren't other options at the time, as computer vision was too computationally expensive at the time.

Any solution to self-driving cars (that purports to be a universal solution) that requires specific modifications to the roads is doomed to failure.

Even leaving aside all the gravel and dirt roads, and the many places where, at any given time, there's construction moving a lane 5 to 15 feet to one side temporarily, do you know how many miles of road there are in the US alone? I sure don't, but I know it's a staggering amount, a hell of a lot of it is rural (and some seasonal), and getting all of it retrofitted with even the most simple and low-tech self-driving guidance assist tech like your magnetic nails would cost vast amounts of money and either take a decade or three or cost about ten times as much.

Nice, I hadn't heard about that. Apparently there's a company working on using light to detect blood alcohol level so you can use a touch sensor. https://www.npr.org/2022/09/20/1124171320/autos-drunk-drivin...

They also need to enforce the speed. In that story the drunk driver was driving 90 mph.

People drink and drive in America because they have no choice. You can’t walk him from a bar, and even if it’s available nobody is paying $40 for an Uber when they can drive their car.

If we want to lower drunk driving incidence rates we need people walking to the bar or taking public transit. It requires us to fix our outdated and asinine transit culture.

All of the technical solutions in the world can’t solve this problem because they’re solving for the wrong thing (safer cars) when we should be solving for reduction in driving.

>People drink and drive in America because they have no choice. You can’t walk him from a bar, and even if it’s available nobody is paying $40 for an Uber when they can drive their car.

List of choices:

1. Not drink alcohol

2. Drink less alcohol, and switch to water at least 1 hour before driving.

3. Pay for a driver

4. Live where you can walk to an establishment that serves alcohol

5. Rent a hotel room nearby

And so on and so forth. Drunk driving is certainly a choice, one made because the perceived or true cost of drunk driving is less than the cost of not driving drunk.

6. Sleep in your car

I think it's important to add this one because it's free and doesn't even require previous planning, so it's very hard to argue that you couldn't do it instead of driving drunk.

(Although keeping a blanket in your trunk makes it a much better experience.)

You definitely will run into legal issues depending on the state.

Some states like California you can only get a DUI if there’s evidence you moved your car, but in other states it’s a lot less clear - and you can be charged with any appearance of intent to drive, like sleeping in the driver’s seat.

Not to mention the possibility of getting a trespassing charge in a private lot overnight.

Which just results in you getting a DUI even though you chose not to drive.

Well, even so, getting a DUI while sleeping is strictly better than getting a DUI after a car crash.

Plus, I don't know how American cops work, but hopefully they're less likely to stop and breathalyze a dude sleeping in a parking lot than a guy on the road at 4 AM.

Also, now that I think about it, American cars are huge, if you put the seat all the way back is your body you going to be even visible from the road?

> body you going to be even visible from the road?

No but your car will be, which will probably be investigated. If you are on the side of the road you'll definitely be investigated and for sure get a DUI because they'll say you drove there and parked. (Also definitely a no-go on highways/freeways).

If you are in a parking lot cops will usually scope out a car, especially if you are parked away from other cars and then they'll wake up and tell you to move. Then you'll say "I'm drunk I can't drive" and then that's where all of the problems start. If you park near other cars (perhaps late night workers) they'll probably call the cops or call a towing company because you aren't allowed to park there and sleep overnight (another reason all these gigantic stupid surface parking lots are so useless).

There's just way too many problems with this strategy.

Stopping on the side of the road should get you a DUI, for the reason you listed. I didn't suggest that - although I guess stopping too late is still better than staying on the road!

I've slept in my car in night clubs' parking lots several times and have never been bothered, either by the owners or by the police. I've also had friends follow my recommendation (well, more like pleading/pushing, since they were insisting 'noo I'm fine to drive') without trouble. Likely a difference between laws/customs in different countries.

From an individual's perspective, this is strictly worse than driving drunk. Not only can you not sleep in your car legally, you also can be charged with drunk driving while sleeping in your car unless you go out of your way to dispossess the keys. You also certainly can't sleep in your car with the engine running, which is pretty important in the winter time.

There are better alternatives than drunk driving, but sleeping in your car isnt one of them.

  1. Don't drink alcohol

Well yea but we're talking about people who do drink alcohol

  2. Drink less alcohol, and switch to water at least 1 hour before driving.

I think this is a good solution, though of course different factors can and do easily disrupt this.

  3./5. Pay for a driver/hotel

Hmm, I've only had a few and I feel pretty good. And an Uber is $50 just one way and I still have to pick my car up in the morning (if it doesn't get towed). Vast majority of people (hence why we have DUI and drunk driving problems) will choose to risk it.

  4. Live where you can walk to an establishment that serves alcohol

This is tautological w.r.t the OP. If you could walk to an establishment that serves alcohol you'd live in an area where you could walk, which would mean you didn't really need a car. But that's not how we build in America which is what precipitated my comment.

Related to #6 from the poster below - you can get a DUI for sleeping in your car while intoxicated, you can get woken up because you aren't allowed to sleep where you are parked, you can get towed, etc.

6. Ride a bicycle

If the drunk driver ends up in a collision with a motorist, then the damage to the other party would be minimal. The drunk person also still has personal transportation as well.

Against the law and you can get a DUI for this too. It’s actually pretty dangerous. This includes other things too like scooters. E-Bikes can go pretty fast and cause serious injury too.

If you were riding a bike drunk and drove into or were hit by a car you would probably be hurt pretty badly. If you ran into something else, or another person, same deal.

I was thinking more along the lines of a conventional bicycle, not an e-bike. The primary problem with DUI is the injury and property damage that others suffer as a result of a crash. This is because of the mass of the vehicle that the DUI uses. If they were on a bicycle, the danger would be minimized for others, and they would be the ones who are injured or killed in the crash.

Based on the remark about individuals making a choice to operate a vehicle while under the influence, having them use a bicycle instead of a vehicle that's significantly heavier, the choice changes from do they drive under the influence and take the chance on injuring or killing people to do they drive under the influence and take a chance on injuring or killing themselves. Whether that change influences their choice remains to be seen.

I added the e-bike as an extra factor not focusing on it there. You can still get a DUI.

But the point of the bike comment largely doesn’t matter since people have to drive to the bar or wherever and there wouldn’t be a bike to get a DUI on.

I’m actually not even sure that if given equal consideration a bike causes more drunken damage (although probably less lethal) but we won’t know because so few people ride bikes since they have to drive a car so they just get DUIs and kill people and themselves that way since we force that on people.

"People drink and drive in America because they have no choice."

They always could choose not to drink if they have no choice but to drive. How is that not a choice?

Because you have the wrong premise. It's not "do I drink or not", it's "I have drank, do I drive or not?"

So we seem to agree -- there is a choice.

I think we disagree because you're saying you can choose to drink or not, but the choice we're talking about is now that you have drunk, do you drive or not? Because that's the actual choice that Americans face.

If you're saying that "drive or not" is the choice, then sure we agree. But that's not really a choice because every incentive would be to drive and that's the structure we've built with car-only culture.

It's a train. You're inventing trains but much worse and not available to the poor.

Just build a train.

If everyone lives right by the train station that’s fine.

What about people in wheelchairs who live in the countryside and prefer to be dropped off at their doorstep? Or someone who can’t drive themself but could use an autonomous car or taxi to go to the hospital in an emergency situation (without waiting a while for the next open train)?

Most people already live near a train, tram or bus station. Or somewhere it would be worth building one anyway.

Taxis already exist, just like trains they work fine.

Personal cars will likely always exist, but only a small minority absolutely needs one. The bulk of human transportation doesn’t need cars.

Trains are great. They’re safer, more efficient, and often more convenient than cars. But it’s not either-or.

My comment was emphasizing that smart road technologies would be useful to many people also. There are many places where it’s not possible or desirable to build train infrastructure—these are areas that can be served by autonomous vehicles.

> What about people in wheelchairs who live in the countryside and prefer to be dropped off at their doorstep?

The important thing is to build a system that works for wide masses - even replacing half of individual car transit by rail and bus works wonders for space consumption [1]. The more complex cases - rural people, mostly - can be dealt with later on, e.g. by providing something like a "phone bus" service. Here in Germany, we're experimenting with on-demand buses [3]: they serve the existing station network, but unlike the usual village buses that run every two hours or worse, you can order them like a taxi and use your existing public transport ticket.

Saying "no" by default because a system is not entirely perfect at the beginning is a destructive discussion tactic that only serves those in power.

> Or someone who can’t drive themself but could use an autonomous car or taxi to go to the hospital in an emergency situation (without waiting a while for the next open train)?

Someone in an emergency situation should just be able to call an ambulance. That's what they exist for. Ideally, someone in an emergency situation should also not have to fear going bankrupt (to the tune that this scenario has become a meme in itself [2]).

[1] https://www.researchgate.net/figure/Comparative-efficiency-o...

[2] https://www.reddit.com/r/ThatsInsane/comments/utbn5z/a_man_k...

[3] https://www.sueddeutsche.de/muenchen/landkreismuenchen/landk...

> Saying "no" by default because a system is not entirely perfect at the beginning is a destructive discussion tactic that only serves those in power.

Who are these people in power, and what are their goals? I don’t see what this has to do with my comment—I’m not arguing against trains, merely in support of autonomous vehicles, in areas where they might be useful, like far away from cities.

Mining, auto, and energy stocks go vroom vroom. Put tax money in roads and stocks go up. Also when you have a multi trillion dollar military to steal oil for chevron it's a handy way of knowing when the stocks are going to change.

https://investinganswers.com/articles/10-most-popular-stocks...

Then there's good old fashioned malice. Look at who used to live or run businesses where highways are now.

Times are a changing a bit these days. Weirdly a bunch of companies on that list are the ones that would love having a 24/7 video and lidar feed of everyone anywhere near a public space.

> You may find it interesting the NHTSA is pushing for breathalizers in cars starting in 2026.

Yeah, another idiocy. They aren't reliable and need constant calibration, good luck driving a car full of drunks as designated driver...

Reply on news.ycombinator.com