There are actually significant differences between apprenticeships and internships. An internship cannot be paid below minimum wage unless it is an unpaid internship. And if an internship is unpaid, the employer must not derive an immediate benefit from the intern's work. In all other situations, an intern must be paid at least minimum wage.
No it isn't. If ten people live on a forested mountain, the one hundred people in the town in the next valley should not be able to vote to deforest the mountain.
Similarly, if one hundred people live on a forested mountain, the ten people in the next town over and the 20 people in a different nearby town should not be able to vote to deforest the mountain.
The problem is that the founding fathers never imagined that we would have a single state that is almost two orders of magnitude larger than the smallest state. The result of that huge population difference is twofold:
Thanks to the Permanent Apportionment Act of 1929, it is impossible to properly apportion representatives in the House proportionate to the number of people. If Wyoming gets a single representative (the minimum), then California should have 68 representatives. Instead, it has only 53.
Because the electoral college elector count is the sum of the senators and representatives, the disproportional allocation of electors is further magnified.
The net result is that Wyoming has 3 electors and California has 55, whereas proportional to the population, California would have 204. If California were three states, it would still probably not have more than 53 representatives (but it might). It would, however, have 6 senators instead of two, and thus four additional electors. 59 electors is a relatively small improvement, but it can't hurt. If California split into six states (to get within an order of magnitude of Wyoming), that would be ten extra electors.
The only real long-term fix is to either replace PAA 1929 with a true proportional representation law or get the courts to overturn PAA 1929 as an unconstitutional violation of Article I Section 2 Paragraph 3 of the U.S. Constitution as amended. Then change the electoral college so that it matches the congressional behavior, i.e. president is elected by electors proportional to the population, and the vice president (and president of the senate) is elected by two electors per state.
I hate to tell you this, but most pharmaceutical R&D is already done by the U.S. government. Every single drug approved between 2010 and 2016 was at least partially funded by NIH grants.
Most drugs today are initially researched by academic institutions, most of which rely heavily on NIH grants to cover their costs. Then, those patents get sold to small businesses, who do the next phase of trials. Then, they get sold to bigger businesses that fund the final phases of trials and manufacturing. Businesses do the development, but government does the actual research.
I kind of like to look at it in terms of dollars per person:
The U.S. spent $1123.73 per person.
The EU spent $541.34 per person.
Sub-saharan Africa spent 29 cents per person.
The real problem isn't that the EU spent half what the U.S. did, but rather that most of the world is more like that third bullet point than either the first two.
As of 2014 (the latest stats I could find), the U.S. provided 45% of global medical R&D spending. So if the U.S. cut its spending to EU levels, we'd lose somewhere on the order of a quarter of the world's medical R&D spending.
This is not to say that the EU necessarily needs to step up its game—there are plenty of other countries that are contributing far less—but U.S. spending on medical R&D is decreasing pretty significantly, so if nobody else picks up the slack, then medical tech is likely to stagnate pretty badly in the near future.
Switzerland has a population of about 8.2 million people, roughly equal to the city of New York. Try scaling their model up 40x and see how well it works.
Currently, every state in the U.S. has a separate insurance market, and there are 50 states. No scaling needed. Q.E.D.
By law, non-profits cannot turn their income over to any sort of shareholder, which means that the only people who can make money from it are the people who actually work for the nonprofit. You might disagree with how much some top-level execs are being paid or whatever, but the net cost to society from all the dividends paid to people whose sole contribution was a little bit of operating capital is staggering by comparison.
And if all hospitals were nonprofit, then there would also be no need for them to compete with one another, which means there would be no need for advertising, which also makes up a healthy chunk of the cost of healthcare.
Every dollar spent on competing against other hospitals or passed on to shareholders is a dollar that could have been spent on improving patient care or reducing the cost of patient care. Now you might argue that they would instead spend it on raising doctor and nurse salaries, and you might be right, but even if that is true, wouldn't you rather have salaries that encourage the best and brightest to go into the field, rather than salaries that barely pay off their medical school debt and the cost of malpractice insurance?
And just to be clear, I love the tech. I'm glad I have it; driving home late at night, there's no question about whether it is safer with that feature than without. If I fell asleep at the wheel, I would have a decent chance of surviving the experience with autosteer active, which is far better than the likely result without that feature. As I said, it is better than even the best drivers at their worst.:-)
True, but those statistics show it being safer while monitored by a human driver who takes over when it goes bonkers. If you had to rely solely on their self-driving tech (i.e. if the driver stops paying attention), it can go wrong in a hurry on some roads. It is nowhere near as good as even an average driver by itself, or at least not with AP2 and later. (I have no experience with AP1, so I can't comment on that.)
For example, until 2018.10.4, CA-17 was downright terrifying with AP2 under autosteer from about The Cats restaurant to Vine Hill Rd, and there were even a couple of curves through Scotts Valley where it did really stupid things. I pretty much had to drive that stretch by hand, because it would get dangerously close to barriers, cross the center line into the adjacent traffic lane, etc. with such regularity that it was almost like it was trying to kill me intentionally.
In the two most recent firmware updates, it can *usually* make it most of the way without me wresting control more than a couple of times, and I can usually put it right back into autosteer after a few seconds without having to take control back three seconds later, so it has gotten a lot better. That said, it still has a long way to go before I would say that it is better than an average driver when it is driving without adult supervision.
Tesla Autopilot has driven more than 1.3 billion miles [electrek.co], and has killed two people. So the fatality rate is roughly a tenth that of humans. That is a lot better than "on par" with humans.
The times when the car is most likely to be in danger of getting into a wreck, a human has to be in control. AP is reluctant to change lanes, cannot turn or exit at all, cannot handle stop signs or traffic lights, etc. Given those limitations, if it didn't cause an order of magnitude fewer deaths, I'd be terrified.
Because Self Driving Vehicles are already safer than the AVERAGE human. That means the every day you delay replacing human drivers with SDCs, is a day in which somewhere between zero and 3500 extra people die that didn't have to.
That might be true for Waymo's self-driving tech (I don't know, because I've never been in one), but it sure as heck isn't true for Tesla's. It's a nice convenience to relieve us of some of the tedium of having to pay continuous attention to traffic when it is moving bumper-to-bumper at 10 MPH. It is halfway decent at most straight or nearly straight roadways. BUT:
It changes lanes randomly in curved intersections
It often handles exits by splitting the difference and then randomly guessing a direction after you're already driving in the gore area.
It often thinks that striped areas are valid driving lanes.
Same goes for shoulders.
It doesn't reliably seem to notice and leave room for lane-splitting motorcycles, cars cutting into your lane, pedestrians, bicycles, or really much of anything else other than cars right in front of you.
In short, there are plenty of spots where it drives significantly worse than someone who has never been behind the wheel of a car before. The only way that's better than an average driver is if you're limiting the discussion to people who are out driving after 11:00 P.M. on New Year's Eve.
It is, however, usually better than a driver who is drunk or asleep or not paying attention to the road, which means it is better than the average driver at his or her worst times. That makes it useful, but only as a backup.
California has adopted "pure" comparative negligence, which would allow the driver's estate or family to seek damages from Tesla (but I would guess not California, due to sovereign immunity).
Sovereign immunity only applies if the state was not grossly negligent and was not required by law to do those sorts of repairs in a timely manner. IMO, it isn't a given that the state would be protected, given the circumstances.
Let me turn that around. Some company builds a car seat, and designs it to withstand collisions at 35 MPH and under. They make it very clear that you are not to use it on highways, but someone does, and a child is killed.
Under your legal theory, the parent is at fault, because the child seat was labeled as being unsuitable for highway use. In practice, the car seat maker loses badly, because it is reasonable to assume that no matter how many warnings are on the box, the owner of such a car seat has a reasonably high chance of choosing to drive at speeds above 35 MPH anyway.
So IMO, the questions that will determine liability in this case are whether it is possible for people to remain as vigilant while their car is steering for them as they would while doing the work themselves, whether the lack of vigilance is something that could have been reasonably predicted by the manufacturer, and whether they took adequate steps to detect a lack of vigilance and disable the feature.
Everyone who buys a Mac is paying to subsidise the continued maintenance and support of the 32-bit versions of system libraries. Very few people are actually using these libraries.
On the other hand, to play devil's advocate, if very few people are actually using the 32-bit frameworks, then there won't be very many bug reports, so the 32-bit frameworks will get very little maintenance. Mind you, they will eventually degrade to the point that they can only run a handful of 32-bit games, but if that's all people are using them for, then whatever.:-)
The funny thing is that I want 32-bit support back on iOS, and I'm okay with it going away on OS X. iOS 11 hurt badly because of the sheer number of abandonware 32-bit apps on iOS that I still used.
The problem is that iOS apps were 32-bit-only for six years (the first couple of years of which were jailbreak-only) before the first 64-bit devices came on the market. That's potentially a six-year window in which app developers could develop 32-bit-only apps, versus only a five-year window for updating those apps to 64-bit afterwards.
By contrast, OS X went 64-bit eleven years ago, and there was only about a year from when the first Intel Macs were released and when the first 64-bit Intel Macs and OS became available. That's a much narrower window in which developers could have created an Intel app and subsequently abandoned it. And that window ended eleven years ago instead of five. With the possible exception of games, if you're still running a 32-bit app at this point, there's really something wrong.
The attenuator is a safety device that no one should have hit - it wasn't like a giant pothole in the middle of the road that people were swerving around and that Caltrans didn't fix promptly. Caltrans simply can't fix every such device immediately and keep the road closed until the repair is completed.
If you look at the Google Maps historical photos, you'll see that the attenuator at this spot is collapsed very frequently (several times per year). So leaving it collapsed for almost two weeks gives you a fairly decent chance of having a fatal crash.
The reset procedure takes half an hour, and requires replacing two sacrificial bolts. Numerous people who do that sort of work for a living in other parts of the country remarked that this sort of attenuator should have been reset within a day, give or take. So taking almost two weeks seems like a pretty serious lapse in safety to me.
As for the rain, even if it had been pouring down rain all day every day (and I'm pretty sure that was not the case), they could have easily dropped temporary sand barrels into that gore point (which should take seconds, can be done at any time of day or night, and doesn't require shutting down the road at all). That would have been far preferable to leaving it unprotected for so long.
Also, when the attenuator collapses, those lanes of the freeway are typically shut down anyway for the crash investigation, and there is a Caltrans depot three blocks away. Most of the time, when someone hits that attenuator, they should be able to get someone out there to reset it while the road is still shut down for the crash investigation and fix it before the police and emergency vehicles even leave the area. I find it alarming that the attenuator is ever in a collapsed state with cars on the road, much less in that state often enough to show up in Google Maps historical photos.
So no, this failure by Caltrans really isn't understandable. It's baffling, and far too many people who do that sort of thing for a living have said the same.
No, it isn't. People have a conception that "autopilot" for aircraft means the computer takes over for the pilot, who can then take a nap or wander around the cabin. We as a society have used the prefix "auto-" to mean that a computer handles the whole process.
Wouldn't matter. They could call it "Lane keeping and traffic-aware braking", and people who haven't driven the sorts of roads that cause it to freak out would still get used to not having to do anything, and would eventually stop paying attention. The problem is human nature. The naming might make the problem worse, or cause it to happen sooner, but it will still happen.
There is plenty of blame to go around-- victim, Tesla, Caltrans for starters. Each of them screwed up on at least two levels. Tesla likely needs some kind of way for drivers to flag a spot where the autopilot screwed up, so they can gather data and investigate, because the victim was aware of issues at this location and tried to address it with Tesla in (apparently) multiple occasions to no avail.
They do. Just tap the voice recognition thing and say, "Bug report autopilot tried to inappropriately touch the guard rail again" or whatever.
Perhaps they shouldn't call it autopilot? The term is clearly a marketing term that makes you think it is going to automatically pilot the car. Call it advanced cruise control, or lane assistance.
The terminology issue is a red herring. The reality is that partial self-driving capabilities lull users into a false sense of security because they usually work well. The rare failures are often catastrophic precisely because people have gotten used to the technology working, and end up surprised when it doesn't.
That's what made this recent software update such a problem. It made major changes to the way autosteer works on (at least) AP2-based Tesla cars. One of the big changes was "wide lane" handling, which changed the way vehicles behaved when they encounter a wide lane, such as an exit lane. This has resulted in a number of unexpected behaviors, up to and including cars driving straight towards gore points.
I don't know whether that change was in any way a factor in the autosteer malfunction that led to Mr. Huang's death, because I have no way to know what firmware version that car was running. However, the fact that this major update was in the process of being rolled out to users at the time of the accident is suspicious.
To be fair, a lot of other driving situations got significantly better with that software update. However, Tesla AP's tendency to ignore solid white lines has been an ongoing problem that might well have been made worse by that update; if that is the case, then the problem needs to be corrected ASAP, and they probably should NOT have continued the rollout of that update. Either way, I'm not convinced that Tesla did enough to warn drivers that autosteer might behave very differently, and to be particularly alert after that update.
Also, I would add that, speaking as a Tesla owner, it bothers me to see the amount of spin they're spewing after this accident. I realize that they don't want to let their users get scared into not using AP, because on average, it does significantly reduce accidents. And if there are videos out there showing AP malfunctions that they feel are not genuine, they can and should comment. But they should really stop trying to convince the public that the driver was solely to blame, because IMO, that just isn't the case.
First, the fact remains that autosteer obviously DID malfunction, and that malfunction DID result in a fatality that would NOT have occurred if the vehicle had not been equipped with autosteer functionality (because no sane driver would have looked away from the road for 5+ seconds without that functionality).
Second, the situation was entirely predictable. For at least a decade, people have warned that humans are likely to zone out in partial self-driving situations, and that it isn't really possible to change that innate human tendency. Tesla ignored those warnings and pushed forward anyway, and someone died. They blamed the driver, and the crash investigators tentatively agreed, and they kept pushing forward. And then a second person died. And now a third. IIRC, product liability law hinges in large part on whether user errors are reasonably predictable, and no "I agree to pay attention" can change that fact, which means this is little more than a legal smokescreen, IMO.
Third, the fact also remains that Caltrans failed to reset the safety barrier that was designed to slow down a car before impacting the gore point, after the barrier was collapsed in a wreck nearly two weeks earlier. And the fact remains that had the barrier been reset properly (as is required by law), it is unlikely that Mr. Huang would have died.
In other words, there are three parties, any one of whom/which could have prevented the fatality, and the deceased driver was only one of those three. So it is entirely disingenuous to try to pin this on the driver in the court of public opinion. IMO, it really isn't a question of who is at fault; they all are. Rather, it's a question of wh
How is a non-user different from someone who is neither a user nor a non-user?
I believe the word should actually be "either", not "neither", and the answer, of course, is Heisenbergian uncertainty. The wave function does not collapse until observed.:-)
There are three groups: People who are known to be Facebook users, people who are known to not be Facebook users, and people who might be either one. In the first group, you know their account info, so you know who they are, and you know that they have Facebook accounts. In the second group, you know enough about them (name, phone number, address, whatever) to believe with a reasonable degree of certainty that they are not Facebook users. In the third group, you don't have enough information to decide either way.
That said, what the GP post appeared to mean was that there are active users, inactive users, and non-users. The GP erroneously conflated inactive users with non-users.
You could disable the thrashing detection if there are multiple people in the pool; let the people who are there decide whether someone is just messing around or is actually in trouble.
Those "worst drivers" lose their licenses sooner or later.
Ha! You Fell Victim to one of the Classic Blunders, the most famous of which is "Never get involved in a land war in Asia," but only sightly less well known is this: "Never assume that the worst drivers are the same people from one day to the next!"
To be fair, the drunk drivers tend to be consistent night after night, but taking away their licenses doesn't usually keep them from driving drunk, from what I've seen, making that a fairly ineffectual approach at improving road safety.
The bigger problem problem with your assumption, though, is that anybody can drive while fatigued, and probably every driver does so once in a while. Fatigued drivers represent a continually shifting subset of the driving population. The people who are tired one day probably won't be tired on their next trip. Instead, it will be someone entirely different.
On the flip side, a decent percentage of people drive them because they like powerful cars. Some like powerful cars for safety reasons, but others like powerful cars because they like to drive like a bat out of you-know-where. So you shouldn't necessarily assume that the cost of the car makes people better drivers.:-)
I would think a wave-activated device, coupled with a security camera and some sort of mobile or web app, would be great (if I had a pool, which I don’t). It would work for neighborhood kids, which this wouldn’t. And, if I were away from the house, I could still call for emergency responders were it warranted.
Even better: Just keep a camera pointed at the pool. Use motion detection to contact you if anyone enters the pool. Combine it with face recognition to determine if the person is authorized to use the pool unattended, and don't send the alarm for those people. Then, teach it to recognize signs of drowning and send an emergency message even if the person is authorized.
Yeah. Put this together with the back seat alarm [slashdot.org] and I won't have to think about my kid ever again.
Ah. An alarm to keep you from *forgetting* a kid who is *already* in the back seat. In the context of a story about an alarm that tells you when someone gets into something (a pool) that they're not supposed to, I half expected you to be linking to an alarm that warns you when your teenage daughter gets in the back seat of your car with her boyfriend.
Nope. Traffic laws invariably define situations in which that is not true, such as pedestrians stepping out in front of a car that is too close to stop. And most states also say that the pedestrian no longer has the right of way after the pedestrian has finished crossing your lane.
Slashdot Mirror
There are actually significant differences between apprenticeships and internships. An internship cannot be paid below minimum wage unless it is an unpaid internship. And if an internship is unpaid, the employer must not derive an immediate benefit from the intern's work. In all other situations, an intern must be paid at least minimum wage.
Similarly, if one hundred people live on a forested mountain, the ten people in the next town over and the 20 people in a different nearby town should not be able to vote to deforest the mountain.
The problem is that the founding fathers never imagined that we would have a single state that is almost two orders of magnitude larger than the smallest state. The result of that huge population difference is twofold:
The net result is that Wyoming has 3 electors and California has 55, whereas proportional to the population, California would have 204. If California were three states, it would still probably not have more than 53 representatives (but it might). It would, however, have 6 senators instead of two, and thus four additional electors. 59 electors is a relatively small improvement, but it can't hurt. If California split into six states (to get within an order of magnitude of Wyoming), that would be ten extra electors.
The only real long-term fix is to either replace PAA 1929 with a true proportional representation law or get the courts to overturn PAA 1929 as an unconstitutional violation of Article I Section 2 Paragraph 3 of the U.S. Constitution as amended. Then change the electoral college so that it matches the congressional behavior, i.e. president is elected by electors proportional to the population, and the vice president (and president of the senate) is elected by two electors per state.
I hate to tell you this, but most pharmaceutical R&D is already done by the U.S. government. Every single drug approved between 2010 and 2016 was at least partially funded by NIH grants.
Most drugs today are initially researched by academic institutions, most of which rely heavily on NIH grants to cover their costs. Then, those patents get sold to small businesses, who do the next phase of trials. Then, they get sold to bigger businesses that fund the final phases of trials and manufacturing. Businesses do the development, but government does the actual research.
I kind of like to look at it in terms of dollars per person:
The real problem isn't that the EU spent half what the U.S. did, but rather that most of the world is more like that third bullet point than either the first two.
As of 2014 (the latest stats I could find), the U.S. provided 45% of global medical R&D spending. So if the U.S. cut its spending to EU levels, we'd lose somewhere on the order of a quarter of the world's medical R&D spending.
This is not to say that the EU necessarily needs to step up its game—there are plenty of other countries that are contributing far less—but U.S. spending on medical R&D is decreasing pretty significantly, so if nobody else picks up the slack, then medical tech is likely to stagnate pretty badly in the near future.
Currently, every state in the U.S. has a separate insurance market, and there are 50 states. No scaling needed. Q.E.D.
Citation needed.
By law, non-profits cannot turn their income over to any sort of shareholder, which means that the only people who can make money from it are the people who actually work for the nonprofit. You might disagree with how much some top-level execs are being paid or whatever, but the net cost to society from all the dividends paid to people whose sole contribution was a little bit of operating capital is staggering by comparison.
And if all hospitals were nonprofit, then there would also be no need for them to compete with one another, which means there would be no need for advertising, which also makes up a healthy chunk of the cost of healthcare.
Every dollar spent on competing against other hospitals or passed on to shareholders is a dollar that could have been spent on improving patient care or reducing the cost of patient care. Now you might argue that they would instead spend it on raising doctor and nurse salaries, and you might be right, but even if that is true, wouldn't you rather have salaries that encourage the best and brightest to go into the field, rather than salaries that barely pay off their medical school debt and the cost of malpractice insurance?
And just to be clear, I love the tech. I'm glad I have it; driving home late at night, there's no question about whether it is safer with that feature than without. If I fell asleep at the wheel, I would have a decent chance of surviving the experience with autosteer active, which is far better than the likely result without that feature. As I said, it is better than even the best drivers at their worst. :-)
True, but those statistics show it being safer while monitored by a human driver who takes over when it goes bonkers. If you had to rely solely on their self-driving tech (i.e. if the driver stops paying attention), it can go wrong in a hurry on some roads. It is nowhere near as good as even an average driver by itself, or at least not with AP2 and later. (I have no experience with AP1, so I can't comment on that.)
For example, until 2018.10.4, CA-17 was downright terrifying with AP2 under autosteer from about The Cats restaurant to Vine Hill Rd, and there were even a couple of curves through Scotts Valley where it did really stupid things. I pretty much had to drive that stretch by hand, because it would get dangerously close to barriers, cross the center line into the adjacent traffic lane, etc. with such regularity that it was almost like it was trying to kill me intentionally.
In the two most recent firmware updates, it can *usually* make it most of the way without me wresting control more than a couple of times, and I can usually put it right back into autosteer after a few seconds without having to take control back three seconds later, so it has gotten a lot better. That said, it still has a long way to go before I would say that it is better than an average driver when it is driving without adult supervision.
The times when the car is most likely to be in danger of getting into a wreck, a human has to be in control. AP is reluctant to change lanes, cannot turn or exit at all, cannot handle stop signs or traffic lights, etc. Given those limitations, if it didn't cause an order of magnitude fewer deaths, I'd be terrified.
That might be true for Waymo's self-driving tech (I don't know, because I've never been in one), but it sure as heck isn't true for Tesla's. It's a nice convenience to relieve us of some of the tedium of having to pay continuous attention to traffic when it is moving bumper-to-bumper at 10 MPH. It is halfway decent at most straight or nearly straight roadways. BUT:
In short, there are plenty of spots where it drives significantly worse than someone who has never been behind the wheel of a car before. The only way that's better than an average driver is if you're limiting the discussion to people who are out driving after 11:00 P.M. on New Year's Eve.
It is, however, usually better than a driver who is drunk or asleep or not paying attention to the road, which means it is better than the average driver at his or her worst times. That makes it useful, but only as a backup.
Sovereign immunity only applies if the state was not grossly negligent and was not required by law to do those sorts of repairs in a timely manner. IMO, it isn't a given that the state would be protected, given the circumstances.
Let me turn that around. Some company builds a car seat, and designs it to withstand collisions at 35 MPH and under. They make it very clear that you are not to use it on highways, but someone does, and a child is killed.
Under your legal theory, the parent is at fault, because the child seat was labeled as being unsuitable for highway use. In practice, the car seat maker loses badly, because it is reasonable to assume that no matter how many warnings are on the box, the owner of such a car seat has a reasonably high chance of choosing to drive at speeds above 35 MPH anyway.
So IMO, the questions that will determine liability in this case are whether it is possible for people to remain as vigilant while their car is steering for them as they would while doing the work themselves, whether the lack of vigilance is something that could have been reasonably predicted by the manufacturer, and whether they took adequate steps to detect a lack of vigilance and disable the feature.
On the other hand, to play devil's advocate, if very few people are actually using the 32-bit frameworks, then there won't be very many bug reports, so the 32-bit frameworks will get very little maintenance. Mind you, they will eventually degrade to the point that they can only run a handful of 32-bit games, but if that's all people are using them for, then whatever. :-)
The funny thing is that I want 32-bit support back on iOS, and I'm okay with it going away on OS X. iOS 11 hurt badly because of the sheer number of abandonware 32-bit apps on iOS that I still used.
The problem is that iOS apps were 32-bit-only for six years (the first couple of years of which were jailbreak-only) before the first 64-bit devices came on the market. That's potentially a six-year window in which app developers could develop 32-bit-only apps, versus only a five-year window for updating those apps to 64-bit afterwards.
By contrast, OS X went 64-bit eleven years ago, and there was only about a year from when the first Intel Macs were released and when the first 64-bit Intel Macs and OS became available. That's a much narrower window in which developers could have created an Intel app and subsequently abandoned it. And that window ended eleven years ago instead of five. With the possible exception of games, if you're still running a 32-bit app at this point, there's really something wrong.
If you look at the Google Maps historical photos, you'll see that the attenuator at this spot is collapsed very frequently (several times per year). So leaving it collapsed for almost two weeks gives you a fairly decent chance of having a fatal crash.
The reset procedure takes half an hour, and requires replacing two sacrificial bolts. Numerous people who do that sort of work for a living in other parts of the country remarked that this sort of attenuator should have been reset within a day, give or take. So taking almost two weeks seems like a pretty serious lapse in safety to me.
As for the rain, even if it had been pouring down rain all day every day (and I'm pretty sure that was not the case), they could have easily dropped temporary sand barrels into that gore point (which should take seconds, can be done at any time of day or night, and doesn't require shutting down the road at all). That would have been far preferable to leaving it unprotected for so long.
Also, when the attenuator collapses, those lanes of the freeway are typically shut down anyway for the crash investigation, and there is a Caltrans depot three blocks away. Most of the time, when someone hits that attenuator, they should be able to get someone out there to reset it while the road is still shut down for the crash investigation and fix it before the police and emergency vehicles even leave the area. I find it alarming that the attenuator is ever in a collapsed state with cars on the road, much less in that state often enough to show up in Google Maps historical photos.
So no, this failure by Caltrans really isn't understandable. It's baffling, and far too many people who do that sort of thing for a living have said the same.
Wouldn't matter. They could call it "Lane keeping and traffic-aware braking", and people who haven't driven the sorts of roads that cause it to freak out would still get used to not having to do anything, and would eventually stop paying attention. The problem is human nature. The naming might make the problem worse, or cause it to happen sooner, but it will still happen.
They do. Just tap the voice recognition thing and say, "Bug report autopilot tried to inappropriately touch the guard rail again" or whatever.
The terminology issue is a red herring. The reality is that partial self-driving capabilities lull users into a false sense of security because they usually work well. The rare failures are often catastrophic precisely because people have gotten used to the technology working, and end up surprised when it doesn't.
That's what made this recent software update such a problem. It made major changes to the way autosteer works on (at least) AP2-based Tesla cars. One of the big changes was "wide lane" handling, which changed the way vehicles behaved when they encounter a wide lane, such as an exit lane. This has resulted in a number of unexpected behaviors, up to and including cars driving straight towards gore points.
I don't know whether that change was in any way a factor in the autosteer malfunction that led to Mr. Huang's death, because I have no way to know what firmware version that car was running. However, the fact that this major update was in the process of being rolled out to users at the time of the accident is suspicious.
To be fair, a lot of other driving situations got significantly better with that software update. However, Tesla AP's tendency to ignore solid white lines has been an ongoing problem that might well have been made worse by that update; if that is the case, then the problem needs to be corrected ASAP, and they probably should NOT have continued the rollout of that update. Either way, I'm not convinced that Tesla did enough to warn drivers that autosteer might behave very differently, and to be particularly alert after that update.
Also, I would add that, speaking as a Tesla owner, it bothers me to see the amount of spin they're spewing after this accident. I realize that they don't want to let their users get scared into not using AP, because on average, it does significantly reduce accidents. And if there are videos out there showing AP malfunctions that they feel are not genuine, they can and should comment. But they should really stop trying to convince the public that the driver was solely to blame, because IMO, that just isn't the case.
First, the fact remains that autosteer obviously DID malfunction, and that malfunction DID result in a fatality that would NOT have occurred if the vehicle had not been equipped with autosteer functionality (because no sane driver would have looked away from the road for 5+ seconds without that functionality).
Second, the situation was entirely predictable. For at least a decade, people have warned that humans are likely to zone out in partial self-driving situations, and that it isn't really possible to change that innate human tendency. Tesla ignored those warnings and pushed forward anyway, and someone died. They blamed the driver, and the crash investigators tentatively agreed, and they kept pushing forward. And then a second person died. And now a third. IIRC, product liability law hinges in large part on whether user errors are reasonably predictable, and no "I agree to pay attention" can change that fact, which means this is little more than a legal smokescreen, IMO.
Third, the fact also remains that Caltrans failed to reset the safety barrier that was designed to slow down a car before impacting the gore point, after the barrier was collapsed in a wreck nearly two weeks earlier. And the fact remains that had the barrier been reset properly (as is required by law), it is unlikely that Mr. Huang would have died.
In other words, there are three parties, any one of whom/which could have prevented the fatality, and the deceased driver was only one of those three. So it is entirely disingenuous to try to pin this on the driver in the court of public opinion. IMO, it really isn't a question of who is at fault; they all are. Rather, it's a question of wh
No, no, you have that backwards. Facebook uses everyone. Everyone does not use Facebook.
I believe the word should actually be "either", not "neither", and the answer, of course, is Heisenbergian uncertainty. The wave function does not collapse until observed. :-)
There are three groups: People who are known to be Facebook users, people who are known to not be Facebook users, and people who might be either one. In the first group, you know their account info, so you know who they are, and you know that they have Facebook accounts. In the second group, you know enough about them (name, phone number, address, whatever) to believe with a reasonable degree of certainty that they are not Facebook users. In the third group, you don't have enough information to decide either way.
That said, what the GP post appeared to mean was that there are active users, inactive users, and non-users. The GP erroneously conflated inactive users with non-users.
You could disable the thrashing detection if there are multiple people in the pool; let the people who are there decide whether someone is just messing around or is actually in trouble.
Ha! You Fell Victim to one of the Classic Blunders, the most famous of which is "Never get involved in a land war in Asia," but only sightly less well known is this: "Never assume that the worst drivers are the same people from one day to the next!"
To be fair, the drunk drivers tend to be consistent night after night, but taking away their licenses doesn't usually keep them from driving drunk, from what I've seen, making that a fairly ineffectual approach at improving road safety.
The bigger problem problem with your assumption, though, is that anybody can drive while fatigued, and probably every driver does so once in a while. Fatigued drivers represent a continually shifting subset of the driving population. The people who are tired one day probably won't be tired on their next trip. Instead, it will be someone entirely different.
On the flip side, a decent percentage of people drive them because they like powerful cars. Some like powerful cars for safety reasons, but others like powerful cars because they like to drive like a bat out of you-know-where. So you shouldn't necessarily assume that the cost of the car makes people better drivers. :-)
Even better: Just keep a camera pointed at the pool. Use motion detection to contact you if anyone enters the pool. Combine it with face recognition to determine if the person is authorized to use the pool unattended, and don't send the alarm for those people. Then, teach it to recognize signs of drowning and send an emergency message even if the person is authorized.
Ah. An alarm to keep you from *forgetting* a kid who is *already* in the back seat. In the context of a story about an alarm that tells you when someone gets into something (a pool) that they're not supposed to, I half expected you to be linking to an alarm that warns you when your teenage daughter gets in the back seat of your car with her boyfriend.
Nope. Traffic laws invariably define situations in which that is not true, such as pedestrians stepping out in front of a car that is too close to stop. And most states also say that the pedestrian no longer has the right of way after the pedestrian has finished crossing your lane.