Hacker News Reader

Comment by camel_gopher | Hacker News Reader

I’m seeing more of them with trash. Last one I took had a rolled up bundle of used bandages.

People are excited by driverless cars but it also means a car with no social barriers and no person who considers the cars condition important. For now they’re well surveilled and a premium vehicle. Soon they will be filthy pods in a race to the bottom with all the charm of a public bathroom. They’ll be cheap, but you’ll get what you pay for. Private driverless cars will be the premium alternative.

Why do you assume the surveillance will go away as they become cheaper? The taxi company know who is in their car and they have access to interior cameras if something happens. In many respects, it is going to be even more difficult to take a dump in one and get away with it than if a human was driving it. They have your credit card number and visual evidence of what you did, they will just charge your card automatically for things like puking.

Because if they ever become super popular, it will be prohibitively expensive to actually police all the surveillance. Likely to store / process it all. AI does an alright job summarizing videos today (Gemini Flash) but it has to get a lot better if they're going to actually police at scale.

The next rider will alert if something outrageous has happened.

No, most people would just move on and stop using the service. A very small % of users report issues.

Incentivize reporting with a discount, combine that with review of video to verify it’s not a false report. Maybe not scalable, but could work.

Image recognition is getting better. I rarely have to get my receipt checked at Sam’s Club anymore now that their camera exit thing is in full swing.

You think? Checking for significant differences between before and after images doesn’t even need AI. Have a triaging system, something like: image diff -> AI assessment -> human assessment -> car drives to human cleaner. In time you can streamline significantly.

I think simple solutions exist. For example the waymo could just ask the rider at the beginning of the ride “Is your car smelly? Any trash in the car?”

If they reply YES to any of those questions, then google keeps the recording of the previous ride for later review. Simple, no?

If the car actually was dirty, give the person that said “car was dirty” a refund of 20% on their ride. Then charge the party that made the car dirty a few. If a person is found to consistently say cars are dirty when they are not, give them a warning and consider kicking them off the platform (abuse).

Alternatively google can switch from recording the first minute of the ride to the last minute of the ride. In other words, only look at the “delta” to compare the car state. That would require a lot less video storage and less effort to review.

You realize that this will just cause people to answer, "yes, it smells like urine" and then pee in the car?

Do you think that many people pre-plan peeing in the car?

it's actually pretty easy to automate the state of the vehicle. if anything is left in the car the video is sent to an operator. The service instructs a cleaner to check the car and the passenger is sent a warning and eventually is banned from using the service.

And when margins compress the first thing that will go is the threshold for cleaning. Cleaning is not just expensive in labor but also taking a car out of circulation. All incentives point towards lowering standards to the absolute minimum people will tolerate.

The car drives itself to a cleaning station, or it is scheduled to be cleaned while charging. I don’t get why you think it will just be a race to the bottom and all common sense will be thrown out the window.

Margins aren't going to compress for some time considering that Waymo's only competition is human-operated driving services.

Competitors that require a $10k car and minimum wage, not a $500k car. That's a lot of margin to make up.

No it won’t. The problem is already solved, event detection is AI from 10 years ago, let alone today. Rider puked in the car isn’t that hard, they can add smell sensors also.

I assume it’ll become less manned and less action taken because manning and taking action costs money, reduces margin, and removes cars from circulation. Certainly for crimes they’ll go after you. But leaving litter, spilling stuff, even leaving bodily fluids, I suspect will get through the cracks more and more as margins compress.

This doesn’t even happen for cheaper mass transit, why would it happen for more expensive automated taxis? Cleaning a car isn’t hard, they can centralize that fairly easy, Waymo has already talked about adding e-nose sensors to the interior eventually, then we don’t even need AI to flag a strong odor.

Have you ever ridden the NYC subway? I did for 10 years and let me tell you, it’s happened on mass transit.

> For now they’re well surveilled and a premium vehicle. Soon they will be filthy pods in a race to the bottom with all the charm of a public bathroom.

So, like transit?

I will likely have my own personal self-driving vehicle. And I'm 100% sure that there'll be an upmarket segment with slightly more expensive cars that are kept more clean than the rest.

That's what transit is like in your country? How unfortunate.

Yep. That's the state of the US transit.

> I will likely have my own personal self-driving vehicle.

The self-driving car ”utopia” (or rather moderate improvement) very much hinges on the space savings on roadways and parking, to increase utilization, reduce congestion and allow dead space to be reclaimed. If people think like you (no value judgment, I suspect this might be the future norm), then you’ll see almost no change to the urban landscape as a whole. It’ll continue to be a one-flesh-body-per-2-tonne vehicle utilization, a ~5:1 provisioning of parking spaces, 25-50% of urban areas being roadways+parking, and a double-digit productivity loss from commuting and running simple errands.

That leaves you with an individual comfort improvement (allowing you to be on your phone while in the car) for a premium price, and increased surveillance tech on personal vehicles. (And, to be fair, it can still be huge for drunk driving deaths, access for elderly & disabled, once costs come down). Overall, very mediocre imo.

Controversial take: the US has painted itself into a corner, where by ignoring the well being of people in their own communities, they need so many workarounds to prevent space sharing between the ~2-3 social groups where intermingling means friction and fear. There are very real logistical challenges to a gated community segmentation of the physical world. This paints the resistance to public transit in a different light: it’s not so much about being public, but rather being shared with strangers, especially of different social cohorts. It also explains the sacred status of air travel which mainly has been left outside the debate: imo because of the higher socio-economic average clientele. Now that cost has come down and low-cost airlines like Spirit share the same airports, the friction has come there as well.

> It’ll continue to be a one-flesh-body-per-2-tonne vehicle utilization

My personal vehicle can fuck off to a distant parking lot way more conveniently than me taking a shuttle there.

Nah. Transit is already a bankrupt transportation mode, cars and a good city design are already superior to transit.

The US is far ahead of the curve, it has not fucked itself like Europe with insane housing density. This is clearly seen in the birth rates. Compare the US, Europe, and Japan.

Do you know how many cities and towns in the US and worldwide have been bankrupted by road maintenance costs? It's car-centric design that is not economically viable, but it's subsidised to the hilt by governments, at the expense of far more cost effective transit methods, to the benefit of... Car and oil companies.

Actually this is untrue. The congestion improvement comes from cooperation between vehicles. Basic control theory applied to cooperative vehicles found 5% of cars participating can eliminate stop and go traffic conditions, and in general as you approach 100% penetration all traffic moves at maximum rate effectively like a self assembling mass transit system. Only if you effectively replaced human drivers as adversarial agents with AI adversarial agents would things stay the same. Even at lower levels of 30% penetration you see most congestion resolve even in urban roads.

https://www.sciencedirect.com/science/article/abs/pii/S09680...

If you built a lightrail where only 3 people could be in each section, that would be a close equivilant to driver density. At anything close to rush hour, you couldn't put enough trains on the line to keep up with demand, because trying to transport that many people, at that low density, all to the same area just doesn't work. It is so crazy to me the lengths people will go to justify car culture over walkability and public transit.

> because trying to transport that many people, at that low density, all to the same area just doesn't work.

This is being solved (to the chagrin of urbanists) by companies moving out of downtown high rises and into suburban office parks. Remote work has helped as well.

> It is so crazy to me the lengths people will go to justify car culture over walkability and public transit.

You aren’t going to make the entirety of the LA metro area walkable and connected by public transit. Sure you can make a neighborhood a little more walkable and you can connect some denser areas with light rail but that just doesn’t solve the problem that you seem to have issue with.

You make it sound like a bad thing that I can avoid sharing space with people who smoke on the train, people who blast music on the train, people who do gymnastics on the train, people who make death threats on the train, people on the train who accost my girlfriend, people in the JetBlue security line who don’t know how to expediently unpack their luggage at the security line (got around this with precheck), and so forth

All of these were actual situations I encountered in NYC. The only reasons I don’t move somewhere car dependent are because I enjoy cycling and I hate parking even more than sharing space with the dregs of society.

Before anyone gets on their high horse about taking care of the lowest members of society, I pay enough taxes to completely absolve myself of any additional responsibility to them (NYC spends around 60000/annum/homeless person).

> prevent space sharing between the ~2-3 social groups where intermingling means friction and fear.

The issue is not space sharing between social groups. I'm happy to share space with people of any ethnicity, skin color, sexual orientation, political leanings, dress code, economic status, educational level or "social cohorts". The issue is civil, polite people being forced to share space with rude, entitled, violent individuals.

This is true. Public spaces require conformity. If the US wants a hyper-individualistic society, then the end result is everybody living in their own isolated pocket dimension simulation.

People adapt. Hopefully it will be more like elevators and less like public toilets.

And then there will be cameras

In public spaces specifically, at least in Sweden, like train stations, the only difference between an elevator and a toilet is that one has to pay to enter the toilet.

> one has to pay to enter the toilet.

I’m not sure what people were expecting when they implemented this policy. There’s plenty of places to pee in the world, a restroom is a request it happens in one particular place.

I would pay for restrooms if they were regularly cleaned and excluded homeless by virtue of their price

The challenge of excluding homeless from bathrooms is they still have to go to the bathroom. I’d rather them use a toilet than on the sidewalk. In SF they had self cleaning bathrooms that would cycle cleaning to ensure it was always sanitary and people couldn’t stay in the bathrooms (I.e., sleep in them).

There often seems to be the idea that if you exclude homeless people from services and accommodation they disappear. They don’t - their impacts on the city become worse.

With per-per-minute sharing cars having existed in many cities in Europe 10+ years, this concept is not new.

People will adapt to the level of cleanliness in the car the get into, so it's a slippery slope. Users will behave respectfully in the early days (maybe because they are first-movers), and then it deteriorates long term.

My own experience is that people used to not even leave an empty soda bottle in the cars and now I see remains from take-out in the floor, coffee cups, chewing gum left around the dashboard etc. You can report this to the car service, but they won't be able to take any meaningful action on it.

Nah, cameras work as long as people are watching and infractions are punished. You need a high conviction rate. Without that people realize that theres a good chance nothing happens, so they stop caring.

Forcing people to pay, and then using that payment to ensure upkeep, is what makes a difference.

Waymo will be making way for cheap versions, where costs are even lower and upkeep a suggestion.

Were you able to flag it to Waymo?

did you report it? Ideally the person that left the bandages in the car would get flagged. They get flagged a few more times for littering in the car they get banned.

Yes, you don't know if it was the previous person, previous previous, etc but if they are a repeat litterer it won't take long to figure out who it is and warn them they'll lose their privilege to use the service if they continue to abuse it.

Reply on news.ycombinator.com