I'm not sure what DOJ's problem is, but I hope that the two companies are willing to address it. As you say, Sprint is effectively dead at this point (I say this as a 17-year customer of Sprint and Nextel), and will declare bankruptcy. They recently announced the sale of their HQ campus in Overland Park, KS, with a deal to lease it back. That's never a good sign. It's unclear how much more money Masayoshi Son is willing to pour into it, but it has to stop at some point. Then the remaining three companies will scramble to buy up parts of it at pennies on the dollar, and I doubt it will be as clean or have as even a result as the current merger could.
There's another possibility: A tech company that would benefit from owning a carrier (Apple, LG, Samsung, etc.) could buy them. Apple was reported to have been considering starting an MVNO. If Apple really wanted to get into services and reduce their ties to hardware, I can't think of a better way to make money off the Android ecosystem than to own one of the carriers that provides service to Google Fi (which currently roams on Sprint, T-Mobile, and U.S. Cellular).
And the only reason Sprint is losing money is that it doesn't have enough customers to cover the cost of upgrading their service hardware. But imagine if every iPad came with some basic level of free service, and that were used as a selling point for the devices themselves. Suddenly, that concern essentially goes away.
And the same argument could be made for any other hardware manufacturer (except perhaps Google, because it already has Fi).
A modern nuclear reactor using proven 3rd generation technology costs upwards of $16billion dollars to build. A 4th gen reactor or experimental design will likely cost significantly more.
Now ask yourself why it costs so much. In 2002, the cost of building a third-generation plant was only $2 billion. I can pretty much guarantee the actual cost of construction hasn't gone up by almost an order of magnitude in 17 years. And the design hasn't changed significantly, either. So where did that extra $14B go? Mostly defending against frivolous lawsuits by NIMBY groups and other bureaucratic red tape that has no basis in safety.
Don't get me wrong, asking questions is good, and demanding improvements to safety standards (and even upgrades to existing plants) as new threats are discovered is also good. But there's a right way to go about it, and the right way involves pushing the NRC (or the equivalent group in whatever country you live in) to raise standards and require compliance upgrades. Tying up construction in court for years just results in skyrocketing costs and results in bringing power plants online that are based on outdated technology, and doing so decades after they should have been decommissioned.
When are we going to find out that all the older reactors similar to Fukushima are shut down safely... Seriously, I have no great problem with the newer designs, but the older ones need phased out, not renewed.
When the NIMBYs stop fighting the deployment of newer designs. As long as pseudo-regulatory barriers erected by the general public make development of new plants based on new designs financially infeasible, companies will stretch the operation of their existing plants far beyond their original design lifespan. It is rather amazing just how much the anti-nuclear movement has made nuclear safety worse.
Cool story bro! You just admitted that the older reactors were unsafe and should not have been built. Even if you had to blame the so called NIMBY's Why would someone want a reactor that you admit is unsafe to be built nearby?
Nothing is perfectly safe. You can either hide in a cave and hope that it doesn't collapse on you or you can embrace technology and the benefits it provides. If you do the latter, each subsequent generation is more reliable and safer than the last, barring serious mistakes. Therefore, it is almost always better to replace existing technology with never versions. This is true whether you're talking about nuclear powerplants, cars, airplanes, etc.
A couple problems with that. First, and probably the biggest problem is that the public was told that these earlier reactors were safe. In some cases, reality proved otherwise. So you have no credibility whatsoever. Why should they continue to believe your dismissive assertions over their lying eyes?
Next up, I'll shock you by saying that a reactor can be built that will be very safe.
Nobody could have predicted a record tsunami. And yet in spite of massively exceeding the design specs, the reactors did not kill people en masse. So I would argue that it was safe. It just wasn't as safe as it would have been if the technology had been kept up-to-date and made to comply with newer safety standards. As a result, it will take years to fully clean up the site.
It won't though. Humans are in the mix.
So you get shady siting decisions, you get unfathomable decisions to build seawalls that are well known both from history and geological records that will simply be breached. It is not possible for them not to be breached eventually unless plate tectonics suddenly stops. Earthquakes will happen, and Tsunami will occur.
You get people making unauthorized experiments that fail and destroy reactors. And on and on.
You get bean counters making/overriding engineering and safety decisions. You get managers demanding schedules be met.
People are the problem. The reactor itself can be made safe. It is an exercise i knowing how much energy is in the reactor, and devising the physical plant to control it. It isn't super easy, and it is very expensive to do, but it can be done if you control both the internal effects and the externalities like siting.
But as long as present day humans are in the mix, ain't happenin' bro!
You're right. And that's why we have laws that force various types of maintenance and inspections — to ensure that even if humans do stupid things, the plants still remain mostly safe, just as Fukushima did.
Is there a worldwide list of reactors that are at risk, and that need to be replaced by newer designs ? And was the Fukushima reactor on that list prior to the incident ?
Yes. Every Generation I or II reactor (everything built before 1996) should be replaced by or upgraded in place to being a Generation III reactor as soon as it is practical to do so. The last Generation I reactor was still online for about four years after Fukushima happened. Fukushima was generation II.
Yes, but uranium mining is an automated process (by necessity), and AFAIK none of that process involves leaving uranium ore lying around to contaminate runoff.
Every single one of the reactors that got shut down was more than 30 years old at the time it failed. And prior to the NIMBY movement, that was generally considered to be the design life of reactors, IIRC. So I would argue that the Fukushima failure would probably not have occurred at all had there been less NIMBY resistance to nuclear power, because it would have been replaced by a newer plant and shut down entirely by 2011.
Now I realize that the timing was entirely random, so it *could* have happened earlier, but the fact remains that newer designs are significantly safer than older designs, and the longer they keep older reactors running, the less safe everyone is.
Also, the NIMBY movement, by making the construction of replacement nuclear plants less attractive, also makes research into nuclear plant design less attractive, which slows progress towards improved safety. So that mindset literally makes the world less safe all around. But that effect is a lot harder to quantify.
Why are most people on the internet so fucking dumb now? Nobody has enough knowledge on literally any subject to make an intelligent post. It's all pseudoscience and conspiracy tards everywhere there's a public interface.
This can be explained pretty easily with statistics. When a new technology comes into existence, assuming it is sufficiently unpolished, only smart people can figure out how to use it. So the median intelligence of users is high. As they make it easier for the masses to use, the median intelligence of users falls. Eventually, most of the smart people move on to something else, and what is left are the average people plus the folks who are just too stubborn to bother changing.:-)
That's the same reason I'm always horrified when someone releases a new app framework or language in an attempt to make it easier to write software. The end result is invariably more software, but written by people who genuinely shouldn't be doing so, with predictably bad results. (Case in point, PHP.)
When are we going to find out that all the older reactors similar to Fukushima are shut down safely... Seriously, I have no great problem with the newer designs, but the older ones need phased out, not renewed.
When the NIMBYs stop fighting the deployment of newer designs. As long as pseudo-regulatory barriers erected by the general public make development of new plants based on new designs financially infeasible, companies will stretch the operation of their existing plants far beyond their original design lifespan. It is rather amazing just how much the anti-nuclear movement has made nuclear safety worse.
Given the local climate they probably don't crack seal yearly like the northern states with freeze thaw cycles so this will likely cost the city more than it would elsewhere where annual crack sealing is common.
I doubt a year goes by in which Louisville doesn't experience at least two or three freezes. Mind you, it isn't a deep freeze, but it doesn't have to freeze below the ground's surface to cause damage to roads that are sitting on top of the surface. All it takes is water getting into the cracks.:-)
Well yes.... that's what most flat earthers have always said... that the earth is round, but like a coin or plate, and not a ball. With the north pole at the center of the plate, and the area that is supposedly Antarctica being at the perimeter.
Really? All you have to do to prove that theory wrong is measure the amount of time it takes for a boat or aircraft to travel from various locations to various other locations. The fact that it does not take a hundred times longer to sail between the tip of Africa and the tip of South America (~6743 km) than between Maine and Sicily (6691 km) proves beyond any reasonable doubt that their theory is absurd....
Then again, the fact that you can fly either direction between those locations, over entirely different parts of the world, proves that their theory is absurd, too, so I guess I shouldn't be surprised....
How many people drive after drinking? If it is 1/3, then alcohol is having no effect on accidents. "Alcohol related" doesn't mean "drunk", just some amount of alcohol.
By that, I'm assuming you meant to say that if a third of drivers on the road have recently consumed alcohol, then it is having no effect. It seems unlikely that the percentage of drivers who have recently consumed alcohol is anywhere near that high, so it is pretty likely that alcohol caused some percentage of those accidents, though determining precisely what percentage is challenging.
By contrast, only about 7% of fatal accidents in people under 35 test positive for THC. Yet about 13% of U.S. adults used marijuana in 2017. This suggests that perhaps pot smoking makes you a safer driver — probably because it's hard to have a fatal accident when you're only driving 5 MPH.:-D
I'd hardly call more than 100% year-over-year growth "tanking". Most companies would LOVE to fail that hard. Yes, they missed Street estimates because of logistics problems overseas, but it happens. It definitely is not an indication of soft demand, as you seem to be implying.
So who pays for that? Hopefully the cost of repairing the roads doesn't go up because many cities are struggling as it is.
It shouldn't cost more, because they should already be doing those things, because they also reduce accidents caused by humans getting confused. But if they aren't, then yes, it will cost slightly more. It always costs more money to do things right than to do them half-a**ed.:-)
Yes, in theory, construction zones can be a problem, but in practice, the problem is mostly political, not technological. Most of the problems you describe are actually caused by the various transit agencies not doing the necessary preparation for construction. In an era of self-driving cars, proper construction signage will likely be mandatory to avoid creating situations that require human intervention.
For short-term lane shifts, safety cones work are unambiguous and easily followed. Transit agencies also need to grind off lane lines and bag any invalid WRONG WAY signs, STOP signs, speed limit signs, traffic lights, etc. so that self-driving cars won't be confused. That said, in most places, they already do those things to avoid confusing human drivers.
So really, the hard part is interpreting traffic signals from a person. Everything else *can* be made unambiguous, given a willingness by the relevant government agencies to do so.
Not at all. I just figured most people who don't drive Teslas have no real concept of just how far away from full self driving Tesla's implementation currently is, and therefore don't realize that it can barely even stay between the lane lines consistently, much less pull off the road.:-)
If pulling over is so easy, why does a Tesla keep driving even when it knows the driver is not responding as required for being in the car?
If Tesla AP were anywhere approaching full autonomy, pulling over and stopping would be relatively easy to add — far easier than driving on residential streets with no lane markings and pedestrian dangers. However, the Tesla AutoPilot feature isn't anywhere near self-driving yet, and their AP computers are running at the limits of their hardware capabilities even without all the features that we'd like for them to add. They probably don't even have the spare cycles for speed limit sign interpretation (brain-dead simple by comparison), much less pulling over.
To support pulling over, Tesla would have to add AP support for:
Deciding whether a shoulder is safe to stop on, including:
Verifying that it is wide enough.
Verifying that it is clear for at least the required stopping distance.
Dealing with the possibility that the car in front of you might make the same decision).
Identifying (on city streets) whether street parking is available and:
Determining if that street parking is available to the general public, reserved, temporarily a no parking area, etc.
Determining if that street parking costs money.
Identifying (on city streets) whether a parking lot is nearby and:
Determining if that parking lot allows public parking.
Determining if that parking lot costs money.
On highways:
Determining whether is safe to stop based on prevailing traffic conditions.
Determining whether is legal to pull over at a given location.
And so on. None of those are hopelessly hard problems, but each one is something that would have to be created and tested/validated before they could roll out such a feature. Also, bear in mind that until a couple of weeks ago, AutoPilot didn't even have support for changing lanes without human confirmation, which would also be an absolute requirement for any sort of pulling over, obviously.
So again, if Tesla AP were anywhere approaching full autonomy, pulling over and stopping would be relatively easy to add, but it does require explicit code support, and right now, they almost certainly don't have the GPU power to do it until they roll out HW3. And even if they did, there are far more critical things for them to add first (e.g. stop signs and traffic lights), along with far more low-hanging fruit that they'll probably add just because it is easy to do so (e.g. speed limit sign reading).
The hard ones are the ones where a guy with a stop/slow sign directs you to drive on the wrong side of the road. Or sometimes he's just using his hands instead of a sign.
That's really not a detour, though. The word "detour" (from French détour: n. a turn or other change of direction) is typically defined as taking a different route that is significantly longer than the normal route (whether because of a road closure, heavy traffic, or just rerouting through town because of a Big Mac attack). If you stay on the same road, you aren't turning, so it isn't a detour.
In the interest of terminology precision, if you're just using a different lane on the same road, that's a lane shift. If a lane shift requires a flagman or a traffic light, that's one-way traffic control. Either also could be called a construction zone, ostensibly (though around here, some one-way traffic control seems to be semi-permanent, so maybe not).
BTW, you'll note that I mentioned the flagman case as a hard case. One-way traffic control by traffic light could, I suppose, also be slightly problematic, but it shouldn't be that much harder than any other properly marked lane shift. Mainly, the software just has to know how to go to the correct side of the center stripe on the other side as soon as it is possible to do so, with the details depending on the country.
As I keep pointing out, think of all the situations you've encountered on the road in the last year that no computer could handle without a vast increase in the capability of embedded AI. Weather, emergency vehicles, detours, mismarked roads, etc. all make fully autonomous vehicle a fantasy, at least in 2019 and the near future.
Weather? Maybe. It kind of depends on the nature of the weather, the type of self-driving tech, what they're doing to de-noise the data, etc. But it causes problems for human drivers, too, so that isn't *necessarily* a show-stopper, so long as it can at least detect when things are getting bad.
Emergency vehicles? If you mean avoiding them, that's a mostly solved problem. If you mean pulling over when chased by a police car, that's still probably pretty easy, at least in relative terms.
Detours? On major roads in most states, this has been a solved problem for many years. The relevant road agency posts a closure notice on their website, some bot scrapes that website and converts the road closure data into a form suitable for algorithmic routing, and the routing algorithm in the navigation system or app guides you around it.
Mismarked roads? Assuming somebody is keeping the maps up-to-date, this usually isn't a problem, either. Of course, if the map data is wrong, that's a different problem. Either way, the right answer is to take the more conservative choice of the two; if the map says you can go, but the signs say no, you find a different route, and vice versa. This should not be a hard problem except to the extent that the underlying sign reading is (and AFAIK, that really isn't).
The hard problems, IMO, are things like:
Noticing that a driver ahead of you is driving erratically and slowing down just in case he or she does something stupid.
Noticing kids playing near the road and slowing down just in case they do something stupid.
Judging the intention of pedestrians.
Interpreting the hand gestures of cyclists.
Interpreting the hand gestures of traffic cops, flagmen, and other similar personnel.
Compared with those things, everything on your list except possibly WX is easy. Then again, Tesla's Nav-on-AutoPilot still hasn't figured out that on a road with six or more lanes, the second lane from the right is the "don't merge into me" lane, not the passing lane, so I guess we shouldn't assume that any given car company will solve even the easy problems.:-/
No, but if you shot at someone and hit a cell phone in that person's backpack, you would be guilty of murder. By placing the backpack on that person, someone has extended that person to include the backpack and its contents. So by killing the cell phone, you have killed the person.
Also, by extension, had he held the cell phone in his hand, it would have become an extension of himself, and thus would no longer be a cell phone, and he would have been found not guilty.
I'm pretty sure that doesn't answer the question at all. The question, as I understand it, is this: If a black hole is a sphere, why would we not expect objects to spiral into it in every possible orientation, similar to the way comets orbit our sun on random planes? And if they do, why don't we see those objects in front of the black hole, rather than just seeing the black disc?
You could require that in the event of bankruptcy whoever buys the assets has to maintain the service. If the assets can't be sold then the liquidator should use the money available to maintain the service, or release as much information as possible so that people can set up their own servers or install open source firmware.
Simpler laws tend to work better, so I'd suggest the simpler approach.
For any product with a cost of $100 or more, if that product requires the use of servers maintained by the company or any third party for any significant portion of its functionality, the company shall either:
Design the product to allow the use of alternative servers and make the server software available to the end user under an OSI-approved open source license OR
Guarantee the continued operation of that service for a minimum period of ten years AND provide a surety bond to cover the cost of that continued operation in the event of the company becoming unable to do so. This guarantee shall not be discharged by bankruptcy, lien, or tort.
Slashdot Mirror
... then bring it back to life.
Inconceivable!
I'm not sure what DOJ's problem is, but I hope that the two companies are willing to address it. As you say, Sprint is effectively dead at this point (I say this as a 17-year customer of Sprint and Nextel), and will declare bankruptcy. They recently announced the sale of their HQ campus in Overland Park, KS, with a deal to lease it back. That's never a good sign. It's unclear how much more money Masayoshi Son is willing to pour into it, but it has to stop at some point. Then the remaining three companies will scramble to buy up parts of it at pennies on the dollar, and I doubt it will be as clean or have as even a result as the current merger could.
There's another possibility: A tech company that would benefit from owning a carrier (Apple, LG, Samsung, etc.) could buy them. Apple was reported to have been considering starting an MVNO. If Apple really wanted to get into services and reduce their ties to hardware, I can't think of a better way to make money off the Android ecosystem than to own one of the carriers that provides service to Google Fi (which currently roams on Sprint, T-Mobile, and U.S. Cellular).
And the only reason Sprint is losing money is that it doesn't have enough customers to cover the cost of upgrading their service hardware. But imagine if every iPad came with some basic level of free service, and that were used as a selling point for the devices themselves. Suddenly, that concern essentially goes away.
And the same argument could be made for any other hardware manufacturer (except perhaps Google, because it already has Fi).
Now ask yourself why it costs so much. In 2002, the cost of building a third-generation plant was only $2 billion. I can pretty much guarantee the actual cost of construction hasn't gone up by almost an order of magnitude in 17 years. And the design hasn't changed significantly, either. So where did that extra $14B go? Mostly defending against frivolous lawsuits by NIMBY groups and other bureaucratic red tape that has no basis in safety.
Don't get me wrong, asking questions is good, and demanding improvements to safety standards (and even upgrades to existing plants) as new threats are discovered is also good. But there's a right way to go about it, and the right way involves pushing the NRC (or the equivalent group in whatever country you live in) to raise standards and require compliance upgrades. Tying up construction in court for years just results in skyrocketing costs and results in bringing power plants online that are based on outdated technology, and doing so decades after they should have been decommissioned.
When are we going to find out that all the older reactors similar to Fukushima are shut down safely... Seriously, I have no great problem with the newer designs, but the older ones need phased out, not renewed.
When the NIMBYs stop fighting the deployment of newer designs. As long as pseudo-regulatory barriers erected by the general public make development of new plants based on new designs financially infeasible, companies will stretch the operation of their existing plants far beyond their original design lifespan. It is rather amazing just how much the anti-nuclear movement has made nuclear safety worse.
Cool story bro! You just admitted that the older reactors were unsafe and should not have been built. Even if you had to blame the so called NIMBY's Why would someone want a reactor that you admit is unsafe to be built nearby?
Nothing is perfectly safe. You can either hide in a cave and hope that it doesn't collapse on you or you can embrace technology and the benefits it provides. If you do the latter, each subsequent generation is more reliable and safer than the last, barring serious mistakes. Therefore, it is almost always better to replace existing technology with never versions. This is true whether you're talking about nuclear powerplants, cars, airplanes, etc.
A couple problems with that. First, and probably the biggest problem is that the public was told that these earlier reactors were safe. In some cases, reality proved otherwise. So you have no credibility whatsoever. Why should they continue to believe your dismissive assertions over their lying eyes?
Next up, I'll shock you by saying that a reactor can be built that will be very safe.
Nobody could have predicted a record tsunami. And yet in spite of massively exceeding the design specs, the reactors did not kill people en masse. So I would argue that it was safe. It just wasn't as safe as it would have been if the technology had been kept up-to-date and made to comply with newer safety standards. As a result, it will take years to fully clean up the site.
It won't though. Humans are in the mix.
So you get shady siting decisions, you get unfathomable decisions to build seawalls that are well known both from history and geological records that will simply be breached. It is not possible for them not to be breached eventually unless plate tectonics suddenly stops. Earthquakes will happen, and Tsunami will occur.
You get people making unauthorized experiments that fail and destroy reactors. And on and on.
You get bean counters making/overriding engineering and safety decisions. You get managers demanding schedules be met.
People are the problem. The reactor itself can be made safe. It is an exercise i knowing how much energy is in the reactor, and devising the physical plant to control it. It isn't super easy, and it is very expensive to do, but it can be done if you control both the internal effects and the externalities like siting.
But as long as present day humans are in the mix, ain't happenin' bro!
You're right. And that's why we have laws that force various types of maintenance and inspections — to ensure that even if humans do stupid things, the plants still remain mostly safe, just as Fukushima did.
Is there a worldwide list of reactors that are at risk, and that need to be replaced by newer designs ? And was the Fukushima reactor on that list prior to the incident ?
Yes. Every Generation I or II reactor (everything built before 1996) should be replaced by or upgraded in place to being a Generation III reactor as soon as it is practical to do so. The last Generation I reactor was still online for about four years after Fukushima happened. Fukushima was generation II.
Yes, but uranium mining is an automated process (by necessity), and AFAIK none of that process involves leaving uranium ore lying around to contaminate runoff.
Every single one of the reactors that got shut down was more than 30 years old at the time it failed. And prior to the NIMBY movement, that was generally considered to be the design life of reactors, IIRC. So I would argue that the Fukushima failure would probably not have occurred at all had there been less NIMBY resistance to nuclear power, because it would have been replaced by a newer plant and shut down entirely by 2011.
Now I realize that the timing was entirely random, so it *could* have happened earlier, but the fact remains that newer designs are significantly safer than older designs, and the longer they keep older reactors running, the less safe everyone is.
Also, the NIMBY movement, by making the construction of replacement nuclear plants less attractive, also makes research into nuclear plant design less attractive, which slows progress towards improved safety. So that mindset literally makes the world less safe all around. But that effect is a lot harder to quantify.
Why are most people on the internet so fucking dumb now? Nobody has enough knowledge on literally any subject to make an intelligent post. It's all pseudoscience and conspiracy tards everywhere there's a public interface.
This can be explained pretty easily with statistics. When a new technology comes into existence, assuming it is sufficiently unpolished, only smart people can figure out how to use it. So the median intelligence of users is high. As they make it easier for the masses to use, the median intelligence of users falls. Eventually, most of the smart people move on to something else, and what is left are the average people plus the folks who are just too stubborn to bother changing. :-)
That's the same reason I'm always horrified when someone releases a new app framework or language in an attempt to make it easier to write software. The end result is invariably more software, but written by people who genuinely shouldn't be doing so, with predictably bad results. (Case in point, PHP.)
When are we going to find out that all the older reactors similar to Fukushima are shut down safely... Seriously, I have no great problem with the newer designs, but the older ones need phased out, not renewed.
When the NIMBYs stop fighting the deployment of newer designs. As long as pseudo-regulatory barriers erected by the general public make development of new plants based on new designs financially infeasible, companies will stretch the operation of their existing plants far beyond their original design lifespan. It is rather amazing just how much the anti-nuclear movement has made nuclear safety worse.
I doubt a year goes by in which Louisville doesn't experience at least two or three freezes. Mind you, it isn't a deep freeze, but it doesn't have to freeze below the ground's surface to cause damage to roads that are sitting on top of the surface. All it takes is water getting into the cracks. :-)
Really? All you have to do to prove that theory wrong is measure the amount of time it takes for a boat or aircraft to travel from various locations to various other locations. The fact that it does not take a hundred times longer to sail between the tip of Africa and the tip of South America (~6743 km) than between Maine and Sicily (6691 km) proves beyond any reasonable doubt that their theory is absurd....
Then again, the fact that you can fly either direction between those locations, over entirely different parts of the world, proves that their theory is absurd, too, so I guess I shouldn't be surprised....
By that, I'm assuming you meant to say that if a third of drivers on the road have recently consumed alcohol, then it is having no effect. It seems unlikely that the percentage of drivers who have recently consumed alcohol is anywhere near that high, so it is pretty likely that alcohol caused some percentage of those accidents, though determining precisely what percentage is challenging.
By contrast, only about 7% of fatal accidents in people under 35 test positive for THC. Yet about 13% of U.S. adults used marijuana in 2017. This suggests that perhaps pot smoking makes you a safer driver — probably because it's hard to have a fatal accident when you're only driving 5 MPH. :-D
Often in the same legal proceeding.
I'd hardly call more than 100% year-over-year growth "tanking". Most companies would LOVE to fail that hard. Yes, they missed Street estimates because of logistics problems overseas, but it happens. It definitely is not an indication of soft demand, as you seem to be implying.
It shouldn't cost more, because they should already be doing those things, because they also reduce accidents caused by humans getting confused. But if they aren't, then yes, it will cost slightly more. It always costs more money to do things right than to do them half-a**ed. :-)
Yes, in theory, construction zones can be a problem, but in practice, the problem is mostly political, not technological. Most of the problems you describe are actually caused by the various transit agencies not doing the necessary preparation for construction. In an era of self-driving cars, proper construction signage will likely be mandatory to avoid creating situations that require human intervention.
For short-term lane shifts, safety cones work are unambiguous and easily followed. Transit agencies also need to grind off lane lines and bag any invalid WRONG WAY signs, STOP signs, speed limit signs, traffic lights, etc. so that self-driving cars won't be confused. That said, in most places, they already do those things to avoid confusing human drivers.
So really, the hard part is interpreting traffic signals from a person. Everything else *can* be made unambiguous, given a willingness by the relevant government agencies to do so.
Not at all. I just figured most people who don't drive Teslas have no real concept of just how far away from full self driving Tesla's implementation currently is, and therefore don't realize that it can barely even stay between the lane lines consistently, much less pull off the road. :-)
If Tesla AP were anywhere approaching full autonomy, pulling over and stopping would be relatively easy to add — far easier than driving on residential streets with no lane markings and pedestrian dangers. However, the Tesla AutoPilot feature isn't anywhere near self-driving yet, and their AP computers are running at the limits of their hardware capabilities even without all the features that we'd like for them to add. They probably don't even have the spare cycles for speed limit sign interpretation (brain-dead simple by comparison), much less pulling over.
To support pulling over, Tesla would have to add AP support for:
And so on. None of those are hopelessly hard problems, but each one is something that would have to be created and tested/validated before they could roll out such a feature. Also, bear in mind that until a couple of weeks ago, AutoPilot didn't even have support for changing lanes without human confirmation, which would also be an absolute requirement for any sort of pulling over, obviously.
So again, if Tesla AP were anywhere approaching full autonomy, pulling over and stopping would be relatively easy to add, but it does require explicit code support, and right now, they almost certainly don't have the GPU power to do it until they roll out HW3. And even if they did, there are far more critical things for them to add first (e.g. stop signs and traffic lights), along with far more low-hanging fruit that they'll probably add just because it is easy to do so (e.g. speed limit sign reading).
That's really not a detour, though. The word "detour" (from French détour: n. a turn or other change of direction) is typically defined as taking a different route that is significantly longer than the normal route (whether because of a road closure, heavy traffic, or just rerouting through town because of a Big Mac attack). If you stay on the same road, you aren't turning, so it isn't a detour.
In the interest of terminology precision, if you're just using a different lane on the same road, that's a lane shift. If a lane shift requires a flagman or a traffic light, that's one-way traffic control. Either also could be called a construction zone, ostensibly (though around here, some one-way traffic control seems to be semi-permanent, so maybe not).
BTW, you'll note that I mentioned the flagman case as a hard case. One-way traffic control by traffic light could, I suppose, also be slightly problematic, but it shouldn't be that much harder than any other properly marked lane shift. Mainly, the software just has to know how to go to the correct side of the center stripe on the other side as soon as it is possible to do so, with the details depending on the country.
Weather? Maybe. It kind of depends on the nature of the weather, the type of self-driving tech, what they're doing to de-noise the data, etc. But it causes problems for human drivers, too, so that isn't *necessarily* a show-stopper, so long as it can at least detect when things are getting bad.
Emergency vehicles? If you mean avoiding them, that's a mostly solved problem. If you mean pulling over when chased by a police car, that's still probably pretty easy, at least in relative terms.
Detours? On major roads in most states, this has been a solved problem for many years. The relevant road agency posts a closure notice on their website, some bot scrapes that website and converts the road closure data into a form suitable for algorithmic routing, and the routing algorithm in the navigation system or app guides you around it.
Mismarked roads? Assuming somebody is keeping the maps up-to-date, this usually isn't a problem, either. Of course, if the map data is wrong, that's a different problem. Either way, the right answer is to take the more conservative choice of the two; if the map says you can go, but the signs say no, you find a different route, and vice versa. This should not be a hard problem except to the extent that the underlying sign reading is (and AFAIK, that really isn't).
The hard problems, IMO, are things like:
Compared with those things, everything on your list except possibly WX is easy. Then again, Tesla's Nav-on-AutoPilot still hasn't figured out that on a road with six or more lanes, the second lane from the right is the "don't merge into me" lane, not the passing lane, so I guess we shouldn't assume that any given car company will solve even the easy problems. :-/
No, but if you shot at someone and hit a cell phone in that person's backpack, you would be guilty of murder. By placing the backpack on that person, someone has extended that person to include the backpack and its contents. So by killing the cell phone, you have killed the person.
Also, by extension, had he held the cell phone in his hand, it would have become an extension of himself, and thus would no longer be a cell phone, and he would have been found not guilty.
Q.E.D.
I'm pretty sure that doesn't answer the question at all. The question, as I understand it, is this: If a black hole is a sphere, why would we not expect objects to spiral into it in every possible orientation, similar to the way comets orbit our sun on random planes? And if they do, why don't we see those objects in front of the black hole, rather than just seeing the black disc?
The answer, I think, is that solar systems tend to flatten out over time because of the influence of gravity, and black holes are presumably very, very old suns.
But I am not an astrophysicist, so take that with a grain of salt.
640chan should be enough for anybody.
Simpler laws tend to work better, so I'd suggest the simpler approach.
For any product with a cost of $100 or more, if that product requires the use of servers maintained by the company or any third party for any significant portion of its functionality, the company shall either:
And you're done.
Goatse-like image on a billboard.... Yikes.