Live data from Hacker News

Self-driving trucks begin mail delivery test for U.S. Postal Service

reuters.com

121–130 of 169 posts

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#121
post #119

Earlier quoted context omitted.

>the USPS has not actually made the payments they were supposed to That's so 100% disprovable, so easily disprovable, disproven by the first link in the comment you directly replied to, that I'm have a hard time assuming good faith in your argument as the HN guidelines tell me to do.

I'm reading the first link and I see language like "83 percent," "50 percent funded," and "unfunded by $86.6 billion." Are you interpreting treis' comment as "hasn't made any of the payments?" Because I read it as, "hasn't made all of the payments," which seems accurate.

"Hasn't made any of the payments" is the only way to read it when they're saying "it's hypothetical". Once the first payment is made, it's not hypothetical any more. It's real money. The USPS only stopped paying because they were more than $8b in debt and could no longer afford the $5.7b per year they needed to pay. Which goes to prove the point that Congress is intentionally taxing them out of business.

If I owe you money and I stop paying and my argument for stopping paying you is that I don't have any money left, do you claim that my money problems are hypothetical because I haven't paid it all? No. They're very real, highlighted by the fact that I can't pay the rest.

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#122
post #82

Earlier quoted context omitted.

>Wrong. How petulant. > Car cuts you off unexpectedly? Easy, slam the brakes in a Tesla. Truck? Crash and kill the people in front of you. Again, how is this different between a computer and a human driver? As far as I can tell, the hardest part of self-driving is accurately seeing the world. Once you see a car, it's fairly easy to make the algorithm pessimistically assume that any car that can cut it off is likely t…

> Once you see a car, it's fairly easy to make the algorithm pessimistically assume that any car that can cut it off is likely to do so. Well, this just doesn't match reality. Uber showed us that these algorithms are full of false positives and have to be disregarded to make the vehicle actually drive itself (Remember when they killed that pedestrian?). What a computer would need is 'awareness,' and that awareness is…

Yes, I do remember when Uber killed a pedestrian. A pedestrian that the car didn't see. Yes the computer needs awareness, but it needs awareness no matter the kind of vehicle it is controlling. The hard part of awareness is the ability to see and correctly categorize potential hazards. That part is absolutely no different for any motor vehicle. The way the vehicle reacts to potential hazards is absolutely going to change depending on the vehicle's characteristics, but that's one of the least hard parts of the whole system.

>Given the current landscape, I'd say with enough data you could more or less brute-force an AI driven car and be somewhere on the spectrum of safer than the worst human drivers.

What's your definition of "enough data"? Tesla seems about as close are you could practically get, and they don't have a great safety history (although they don't have enough history period to really say whether they're safer than humans). The specific issues that Tesla and Uber have faced are issues with seeing and correctly interpreting the world around them. Those issues are largely the same in either case. If you assume you can see and follow lane markings and it turns out you don't do so well when the lane markings are damaged, you're going to have a bad time no matter what size vehicle you're in.

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#123
post #118

Earlier quoted context omitted.

>the USPS has not actually made the payments they were supposed to That's so 100% disprovable, so easily disprovable, disproven by the first link in the comment you directly replied to, that I'm have a hard time assuming good faith in your argument as the HN guidelines tell me to do.

>I'm have a hard time assuming good faith in your argument as the HN guidelines tell me to do. Seriously? You should Google before becoming quite so self righteous: >It is not the first time the Postal Service has missed its required payments to prefund retirees’ health benefits. From 2012 through 2016, the agency failed to deliver nearly $34 billion toward its pool for retirees’ health care. https://www.govexec.com/…

Please explain your use of the word "hypothetical". That might be where I am getting confused. Because the only way I'm interpreting that statement is "the USPS can't blame this law for their money problems because they haven't even made any payments"

At best, I might be able to twist the words to say "the USPS's money problems aren't real because they can't afford to pay their debt", which makes no sense.

The reality of the situation is the $5.7b they're required to pay every year is more than they can afford to pay, so they paid it until they reached $8b in debt and then stopped paying. None of that is hypothetical in any way.

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#124

Earlier quoted context omitted.

And it has a name as well : Starve the beast.

Like every political push, the intent was noble. But it only gains legs when some entrenched interest can profit.

Noble? Maybe... I'd say it's dishonest at best. If the USPS failed on its own, there might be a point to the argument that the government is inefficient. But when lawmakers destroy something and then claim that it destroyed itself and therefore we must destroy everything else, it doesn't quite seem right.

It's like smashing a window with a baseball bat to prove glass is easily breakable. Yes the glass broke, that's why you don't hit it with a baseball bat. Rather than replacing the windows with brick, how about just not hitting it with a baseball bat? If you want to prove how easily breakable the glass is, let is break (or not) on its own.

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#125

Back in 1987, at my first job at Hewlett-Packard, they had a little robot that would deliver mail. This was their plant in Loveland, CO, I believe it was the first non-California site HP opened. Along the main hallways there was some sort of retroreflective tape laid down, and the robot, which looked like and equipment cart, would follow that tape. At every stop it would sound a chime and the secretary for that secti…

A robot like this was a central plot device in The Americans, too.

https://youtu.be/XEQ5mCMO1a8

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#126

Earlier quoted context omitted.

Like every political push, the intent was noble. But it only gains legs when some entrenched interest can profit.

Noble? Maybe... I'd say it's dishonest at best. If the USPS failed on its own, there might be a point to the argument that the government is inefficient. But when lawmakers destroy something and then claim that it destroyed itself and therefore we must destroy everything else, it doesn't quite seem right. It's like smashing a window with a baseball bat to prove glass is easily breakable. Yes the glass broke, that's w…

You're responding on a completely different level than what I actually said.

> If the USPS failed on its own, there might be a point to the argument that the government is inefficient. But when lawmakers destroy something and then claim that it destroyed itself and therefore we must destroy everything else, it doesn't quite seem right.

To be clear, I agree with this. I like the USPS, and think they do a great job.

The point is that grassroots support for "starve the beast" didn't start out focused on the USPS. The USPS was just one of the more vulnerable targets, plus some business stood to gain, so politicians directed the ire there. It's similar dynamic to how in these "government shutdowns", national parks close but the TSA keeps right on going. Or locally, how trying to keep real estate taxes down ends putting the screws on teachers and firefighters, rather than shrinking the town bureaucracy.

The larger point is to avoid simplistically ascribing the worst intent to the "other team". It's an easy answer, but prevents understanding that they actually have legitimate points. And it drives adherence to the agenda of your familiar team, which is also busy manufacturing consent for corporate agendas in a similar manner.

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#127
post #119

Earlier quoted context omitted.

>the USPS has not actually made the payments they were supposed to That's so 100% disprovable, so easily disprovable, disproven by the first link in the comment you directly replied to, that I'm have a hard time assuming good faith in your argument as the HN guidelines tell me to do.

I'm reading the first link and I see language like "83 percent," "50 percent funded," and "unfunded by $86.6 billion." Are you interpreting treis' comment as "hasn't made any of the payments?" Because I read it as, "hasn't made all of the payments," which seems accurate.

The USPS has missed most of they payments, not making a full one since 2010. The health benefits fund (the 50% one) was the only one that needed to be funded and much of that funding came from transfers from the other, overfunded, pension accounts.

Anyway, the point is moot, since the prefunding period is over and the payments are now on the permanent schedule. Because the Post Office couldn’t fully find the account, the fund is now running deficits and will be depleted in a decade.

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#128
post #118

Earlier quoted context omitted.

>I'm have a hard time assuming good faith in your argument as the HN guidelines tell me to do. Seriously? You should Google before becoming quite so self righteous: >It is not the first time the Postal Service has missed its required payments to prefund retirees’ health benefits. From 2012 through 2016, the agency failed to deliver nearly $34 billion toward its pool for retirees’ health care. https://www.govexec.com/…

Please explain your use of the word "hypothetical". That might be where I am getting confused. Because the only way I'm interpreting that statement is "the USPS can't blame this law for their money problems because they haven't even made any payments" At best, I might be able to twist the words to say "the USPS's money problems aren't real because they can't afford to pay their debt", which makes no sense. The realit…

I don't know where you got your numbers but they certainly don't reflect reality. As my cite says, they missed ~34 billion worth of payments. As your cite says they're missing 86 billion dollars that they need to pay their pension obligations.

>Please explain your use of the word "hypothetical".

Hypothetical as in they are not actual real outflows of cash from the USPS. They can't be blamed for the USPS' financial woes since they haven't made a payment in 7 years now.

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#129

Earlier quoted context omitted.

> Once you see a car, it's fairly easy to make the algorithm pessimistically assume that any car that can cut it off is likely to do so. Well, this just doesn't match reality. Uber showed us that these algorithms are full of false positives and have to be disregarded to make the vehicle actually drive itself (Remember when they killed that pedestrian?). What a computer would need is 'awareness,' and that awareness is…

Yes, I do remember when Uber killed a pedestrian. A pedestrian that the car didn't see . Yes the computer needs awareness, but it needs awareness no matter the kind of vehicle it is controlling. The hard part of awareness is the ability to see and correctly categorize potential hazards. That part is absolutely no different for any motor vehicle. The way the vehicle reacts to potential hazards is absolutely going to c…

> Yes, I do remember when Uber killed a pedestrian. A pedestrian that the car didn't see.

The car did see the pedestrian. It was unable to correctly identify it as pedestrian initially. And instead of taking the pessimistic path of 'something unknown moving toward roadway, so I brake now' it had that portion disabled in favor of 'safety driver take over.' Reason being, as it came out, that these types of events are so common it would be wholly infeasible to attempt a 'driverless' solution without disabling the emergency braking routine.

> lane markings are damaged, you're going to have a bad time no matter what size vehicle you're in.

That's true. But it's not a problem humans have. In a Tesla (or whatever, not picking on them specifically), you can probably stop in a reasonable distance, and if you don't, you probably aren't going to kill scores of people. With a semi-truck, stopping distance is greatly increased. So, you need the ability to see farther up the road. If there's traffic in front of you, you obviously can't see lane markings 200 yards up the street. There's just so many variables. What if the road's wet and the lane markings end and the computer thinks it needs to slam on the brakes? I guess jack-knifing the trailer in the middle of busy highway is NBD, because that's what the computer was supposed to do.

I won't share the road with such a contraption.

Re: Self-driving trucks begin mail delivery test for U.S. Postal Service

#130
post #111

Earlier quoted context omitted.

>It started with the mandate to pre-fund pension and healthcare benefits for the next 75 years, funding pensions for retired employees who aren't even born yet. This is wholly inaccurate. All that the USPS is being required to do is set aside money to pay for pension liabilities that they have incurred. Nothing close to funding pensions for employees that have yet to be born. Besides, this is all hypothetical as the…

>the USPS has not actually made the payments they were supposed to That's so 100% disprovable, so easily disprovable, disproven by the first link in the comment you directly replied to, that I'm have a hard time assuming good faith in your argument as the HN guidelines tell me to do.

He’s correct. If they had made the payments they were suppose to, the health benefits account would have been fully funded by 2017. Because they didn’t, it is only half funded and is running deficits.
Post reply on HN