Live data from Hacker News

Disrupting Uber

jacobinmag.com

211–220 of 230 posts

Re: Disrupting Uber

#211
post #199

Earlier quoted context omitted.

I'd also like unicorns towing magic carpets and appreciate the sarcasm, self driving cars are not the 'unicorn' anymore. Uber is actively investing in prepping for a self-driving world.[0][1] https://newsroom.uber.com/mapping-ubers-future/ http://www.theverge.com/2016/7/31/12338268/uber-maps-investm...

When a ~$100k Tesla can't successfully avoid a stationary guardrail or a turning semi, even under the limited conditions that it's autopilot actually attempts to run under, I'd say it's massively premature to suggest that there's anything like a driverless future coming in even the medium-term for affordable vehicles. There's just too many variables on our roads and the stakes are too high. If they are coming, it's m…

It isn't about single accidents; its about statistics. Autodriven cars don't get tired, or drunk, or fool around with the playlist. There is so much low-hanging fruit that its entirely conceivable that autodrive is already safer (statistically) than human drivers.

Add to this, that each time a special case comes along, the software can be patched and it won't happen again. Something humans are terrible at - learning from others' mistakes.

Re: Disrupting Uber

#212
post #199

Earlier quoted context omitted.

I'd also like unicorns towing magic carpets and appreciate the sarcasm, self driving cars are not the 'unicorn' anymore. Uber is actively investing in prepping for a self-driving world.[0][1] https://newsroom.uber.com/mapping-ubers-future/ http://www.theverge.com/2016/7/31/12338268/uber-maps-investm...

When a ~$100k Tesla can't successfully avoid a stationary guardrail or a turning semi, even under the limited conditions that it's autopilot actually attempts to run under, I'd say it's massively premature to suggest that there's anything like a driverless future coming in even the medium-term for affordable vehicles. There's just too many variables on our roads and the stakes are too high. If they are coming, it's m…

> so there's bound to be unpredictable manually-driven cars still on the road for a long time that any driverless cars would still have to contend with as well.

Luckily unicorn-driven magic carpets don't face this problem.

Re: Disrupting Uber

#213

Earlier quoted context omitted.

When a ~$100k Tesla can't successfully avoid a stationary guardrail or a turning semi, even under the limited conditions that it's autopilot actually attempts to run under, I'd say it's massively premature to suggest that there's anything like a driverless future coming in even the medium-term for affordable vehicles. There's just too many variables on our roads and the stakes are too high. If they are coming, it's m…

It isn't about single accidents; its about statistics. Autodriven cars don't get tired, or drunk, or fool around with the playlist. There is so much low-hanging fruit that its entirely conceivable that autodrive is already safer (statistically) than human drivers. Add to this, that each time a special case comes along, the software can be patched and it won't happen again. Something humans are terrible at - learning…

I think parent comment was suggesting that human-driven cars that behave erratically are better handled by opposing humans rather than software (that has not been exposed to dealing with the novel situation).

(Not that auto-driven cars don't produce less accidents when other human-driven cars behave expectedly.)

Re: Disrupting Uber

#214

Good luck with the co-op idea. The problem is that drivers feel like their services are worth a lot, and passengers simply don't believe it. The reason why Uber is a success is that they created new demand by dropping prices. Everyone I know, including myself, would never have taken an Uber as frequently as we do if the prices were as high as taxis. You can't replicate this unless you keep pushing prices down, and th…

Why Cant there be a "meta" ride-app giving you the best rides from a whole host of different services big and small

The only thing preventing that is each service's terms, assuming the aggregator uses official API access.

In some big cities, Google Maps can provide this, sort of, Uber vs bus fare, for instance.

Re: Disrupting Uber

#215
post #102

Earlier quoted context omitted.

That's how economic reality should be treated. Our own reality is primary. We will take back our lives. What Jacobin is working towards is admirable. We believe in progress, that a better way is possible, and that the current state of affairs is suboptimal. We should encourage alternative economic thinking.

I am in favor of encouraging plausible, workable economic thinking. Jacobin is better described as leftist fever-dreams with roughly that level of connection to any reality. Could things work differently? Absolutely. Are they going to work differently because some hack gets off fantasizing about how in abstract theory, the services Uber performs could be better performed by a co-op that pays workers more? No. All the…

United Dairy Farmers based around Cincinnati is one co-op you might find interesting. They are an ice cream shop / convenience store / occasional gas station (I know it sounds strange) that has grown to ~200 stores in our tristate area. I wish there more info about them online.

http://www.udfinc.com/about-udf

The founder and family made Forbes' billionaire list.

http://www.forbes.com/lists/2006/10/S22J.html

Re: Disrupting Uber

#216

Earlier quoted context omitted.

When a ~$100k Tesla can't successfully avoid a stationary guardrail or a turning semi, even under the limited conditions that it's autopilot actually attempts to run under, I'd say it's massively premature to suggest that there's anything like a driverless future coming in even the medium-term for affordable vehicles. There's just too many variables on our roads and the stakes are too high. If they are coming, it's m…

It isn't about single accidents; its about statistics. Autodriven cars don't get tired, or drunk, or fool around with the playlist. There is so much low-hanging fruit that its entirely conceivable that autodrive is already safer (statistically) than human drivers. Add to this, that each time a special case comes along, the software can be patched and it won't happen again. Something humans are terrible at - learning…

So when an autodriven car maims little Johnny, before they patch the bug to not run over future infants, then who's responsible and has to pay? What if other cars maim more infants before they get around to patching what is now a known bug?

You can't blame the driver, they weren't driving. The car manufacturers can't accept blame, as it costs a lot if you have to factor in the cost of a lifetime of support for a maimed infant. Even ignoring the fact that it only works on highways and not normal surface roads, Tesla's "beta" Autopilot feature takes no responsibility because the driver is still supposed to be holding the wheel.

Also as any software developer would tell you, software updates would also be a problem in themselves. If you propose that cars be auto-updated automatically, I wonder how insurance would work when the car is running different software to the software it was originally insured with. Critical saftey systems can't get auto-updated without a lot of testing. What if the auto-update mechanism gets hacked, or maliciously changed by a disgruntled self-driving car employee? What if a legitimate update causes another bug where it maims the elderly now instead of infants? Do they roll back to the infant-maiming software now? I'm being silly but updates on critical safety stuff isn't so simple, especially with the legal ramifications.

I think legal issues will be a problem for self-driving cars, maybe even a larger problem than the actual designing a working self-driving car problem. Someone has to be responsible for inevitable accidents (bugs?). After a century of being able to blame the driver, car companies won't want to change that status quo.

Maybe insurance will decide that they'll cover self-driving cars if they prove themselves to be safe, but we'll see.

Re: Disrupting Uber

#217

Earlier quoted context omitted.

It isn't about single accidents; its about statistics. Autodriven cars don't get tired, or drunk, or fool around with the playlist. There is so much low-hanging fruit that its entirely conceivable that autodrive is already safer (statistically) than human drivers. Add to this, that each time a special case comes along, the software can be patched and it won't happen again. Something humans are terrible at - learning…

I think parent comment was suggesting that human-driven cars that behave erratically are better handled by opposing humans rather than software (that has not been exposed to dealing with the novel situation). (Not that auto-driven cars don't produce less accidents when other human-driven cars behave expectedly.)

I'm imagining that driverless cars could work well on very predictable roads that are exclusively for driverless cars.

But on real roads with questionable maintenance and other unpredictable cars on it, I just can't see it happening unless there's some major change to the roads infrastructure.

Re: Disrupting Uber

#218
post #199

Earlier quoted context omitted.

I'd also like unicorns towing magic carpets and appreciate the sarcasm, self driving cars are not the 'unicorn' anymore. Uber is actively investing in prepping for a self-driving world.[0][1] https://newsroom.uber.com/mapping-ubers-future/ http://www.theverge.com/2016/7/31/12338268/uber-maps-investm...

When a ~$100k Tesla can't successfully avoid a stationary guardrail or a turning semi, even under the limited conditions that it's autopilot actually attempts to run under, I'd say it's massively premature to suggest that there's anything like a driverless future coming in even the medium-term for affordable vehicles. There's just too many variables on our roads and the stakes are too high. If they are coming, it's m…

>Tesla can't successfully avoid a stationary guardrail or a turning semi

It can. The problem wasn't that. The problem was that it wasn't able to differentiate the color properly and thought nothing was there. There's more info in Tesla's blog post. And it's not like it is an unsurmountable problem that is leading to accidents everywhere. It was an edge case and will probably be sorted out.

>massively premature to suggest that there's anything like a driverless future

I really think that's a very harsh opinion. And you're asserting this opinion on many intelligent people working towards a difficult solution. Thousands of people have used auto pilot without any problems and many lives were probably saved (consider if they were driving normal cars instead of self driving ones?)

>There's just too many variables on our roads and the stakes are too high.

Nobody is disagreeing with you on that. That's the reason it already took so long. Self driving cars first appeared in 2005 if not earlier [0]. And remember all of self driving is mostly a software problem, which greatly reduced iteration time for new developments unlike electronic or mechanical revisions.

>25 years away IMO

Funny how people predict things. Ray Kurzweil predicts AGI (general Artificial Intelligence) to arrive before the year 2045, 4 years after you say we get self driving cars. I'm not going to argue with you on that number since we probably have different opinions on what constitute self driving cars. Just so you know, I predict we'll have fully self driving cars on the road (with humans in them) in 5 years.

>Remember 25 years ago, cars weren't super different from today

Please tell me you're kidding. [1][2]

>Self driving cars are pretty much unicorns

At least those unicorns are achievable and people are trying towards achieving it rather than give up.

[0]https://en.wikipedia.org/wiki/DARPA_Grand_Challenge#2005_Gra...

[1]http://www.popularmechanics.com/cars/a11201/why-cars-are-saf...

[2]http://www.iihs.org/iihs/sr/statusreport/article/50/1/1

Re: Disrupting Uber

#219

Earlier quoted context omitted.

It isn't about single accidents; its about statistics. Autodriven cars don't get tired, or drunk, or fool around with the playlist. There is so much low-hanging fruit that its entirely conceivable that autodrive is already safer (statistically) than human drivers. Add to this, that each time a special case comes along, the software can be patched and it won't happen again. Something humans are terrible at - learning…

So when an autodriven car maims little Johnny, before they patch the bug to not run over future infants, then who's responsible and has to pay? What if other cars maim more infants before they get around to patching what is now a known bug? You can't blame the driver, they weren't driving. The car manufacturers can't accept blame, as it costs a lot if you have to factor in the cost of a lifetime of support for a maim…

This is the biggest problem that's really holding back self driving cars right now, and it isn't really new. It is called the trolley problem [0]. Wired has an article or two about this [1]

[0]https://en.wikipedia.org/wiki/Trolley_problem

[1] https://www.wired.com/2016/06/people-want-self-driving-cars-...

Re: Disrupting Uber

#220

Earlier quoted context omitted.

> If you get ten people together on the same street corner and everyone calls for an UberX, you'll get ten different prices. That would be a huge story and would be much more interesting than the typical Uber hit job. If it were true, I have to imagine that it would have been covered by a journalist intent on attacking Uber. Care to share some actual evidence?

Wouldn't it be simple to claim that more demand (for any given corner or area) is immediately reflected in the price? That doesn't sound like a scandal; it's harder to arrange for 10 cars to arrive at a spot than 1.

> Wouldn't it be simple to claim that more demand (for any given corner or area) is immediately reflected in the price? That doesn't sound like a scandal;

Absolutely not. This is why price-discrimination always happens jumping through a weird hoop or other. Imagine the typical example of price discrimination, S/M/L coffee for widely spread prices that don't have much to do with the cost for the amount of coffee with prices that wildly change up and down depending how busy it is, customers would be rightly pissed for getting screwed because they're obviously not getting a fair price. Price differentiation needs some sort of trick to obscure this fact.

Otherwise it's like saying "getting away with this is perfectly fine because capitalism", which is motivated purely by greed, not the possible benefits of a free market.

> it's harder to arrange for 10 cars to arrive at a spot than 1.

I think you forgot to divide by ten, there. Try this statement for a fairer comparison:

to arrange for 10 cars to arrive at 1 spot vs 10 cars at 10 spots

Seems to me that the 10 cars at 1 spot should be cheaper.

The only reason they get away with this price differentiation going the other way is because they are able to hide these prices from the customers. Driving up profits from unequal information availability is also known as "screwing people over", so yeah that'd be a scandal (except for all the HNers that will go "I am genuinely curious why this is so *surprising" to anyone" -- because getting screwed should be a surprise, otherwise you should've gotten mad sooner, like when we said there's no proof of getting screwed)

Post reply on HN