No, the 85th percentile is a good limit as a baseline. You should aggressively go after people who are driving above... say the 99th percentile. Enforcing any law that 15% of people break is generally a bad idea.
Highway 17's biggest problem is that they allow trucks and buses on it.
The car speed limit is 50. Most cars drive 50-65, with the most common speed being about 55, but with 65+ being not uncommon.
The truck speed limit is 35.
So, there's a 30+ MPH difference between the fastest and slowest vehicles. Anything more than about a 15 MPH speed difference translates to a very, very unsafe road.
When you get a truck going 35 in the right lane, all the cars who want to go 50 have to pass in the same fast lane as the cars who want to go 65. Half of them are scared to pass, and end up passing a 35 MPH truck at 40 MPH, creating a cascading backup that can stretch for miles. The other half are angry that they can't go 60+. It is in this sort of backup that accidents are likely to occur, particularly when changing lanes to pass the slow vehicle.
By comparison, 175 might be curvy, but AFAIK (I've never driven that road) it has only one lane in each direction, and where it is unsafe, it is obviously unsafe, so people don't drive like morons. In much the same way, highway 9 is a curvy nightmare, but the problem spots are clearly marked, and people generally respect the road. What makes 17 so bad is that what looks like a halfway decent road really isn't.:-)
Also, it's not very windy at all. The 101, that's windy. Or did you mean winding?
Cameras on poles aren't able to accurately judge a vehicle's speed, and aren't as likely to spot unsafe driving as something that can see large stretches of road at a time. Once you've identified a problem vehicle, then you can either reduce altitude to shoot the vehicle's plate or follow it until it passes a stationary traffic camera that can snap its plate.
You need trained pilots/operators and still are limited by range.
AFAIK, most newer UAVs are autonomous, including takeoff and landing. They require an operator only when they discover something that warrants a person's attention. You could quite literally run these things up and down the road all day, and your only personnel costs would be the person examining the suspicious footage to write the tickets and the person filling them up with fuel.
This is why you post a speed. That speed dictates what is typically safe. If your car is well maintained and the road is wide open and dry, you can go a bit over that. If the road is wet and you have bald tires, you'd better go more slowly.
Who decides that the speed was unsafe?
Whoever is patrolling that stretch of highway, and has footage showing you swerving between lanes at 85 while the rest of the cars are going 50. For the most part, roads self-regulate, and cars tend to run along at about the same speed. Having a few cars going 10 MPH slower to obey an arbitrary posted speed limit actually results in a significantly higher risk of accidents than having everyone going 10 MPH over the limit, on the average.
Unless they're on the approach path near an airport. When planes fly over I-880 in San Jose, for example, I'm pretty sure they're way below 500 feet in HAAT. There is basically no safe altitude for a UAV over that stretch of road, unless the UAV is piloted by a real person, coordinating with ATC.
How are these going to generate revenue from a thousand feet up without the help of another officer on the ground to actually issue the ticket? Bar codes on every car roof?
Umm... we're not talking about commercial airlines here. That's an entirely different universe. The comparison was between drones and traffic aircraft, which almost certainly are not fly-by-wire. They're usually either small airplanes (e.g. Cessna) or small helicopters.
Manned planes have a person who can actively steer them to minimize collateral damage if they're going down.
Manned planes are required by law to comply with maintenance schedules designed to ensure safety.
Manned planes are not (usually) controlled by a computer that can be vulnerable to software bugs.
Manned planes are not (usually) vulnerable to EMPs or random stray cosmic rays.
Manned planes are not made en masse by the lowest bidder.
Manned planes are inherently limited in number by the number of available pilots and the salary of those pilots.
Need I continue?
This is not to say that drones won't have fewer accidents per vehicle (assuming a single pilot who could have a heart attack), but if you can buy the drones for a few hundred bucks apiece, you can potentially deploy tens of thousands of them for less money than a single manned aircraft (over the long term). So even tiny failure rates can pose a very serious problem.
Because if they raise the limit to 75, people will drive 85.
And it would probably still be safe....
Americans have been conditioned to believe that the "real" speed limit is at least 10 mph over the posted limit.
Because it usually is. As far as I'm concerned, the law should simply say, "You may not travel at a speed that is unsafe for the current road conditions." Anything demanding strict conformance to a posted number (rather than driving at a speed that feels safe) is just asking for people to ignore the law... or worse.
The worst example of a highway safety law is California's 65 MPH law. Except for a few roads where it is specifically posted at 70 MPH, it is illegal to drive faster than 65 MPH in California, period. All other speed limits are flexible, depending on driving conditions. What this means is:
If I speed in a 50 zone, if everyone is going 64 (even if that is verging on unsafe), you can potentially argue your way out of the ticket.
If I go 66 in a 65 zone, even if everyone else is going 66, you can't argue your way out of the ticket.
So if you're running behind and trying to decide where to exceed the speed limit, you're better off speeding on the city street portions of your trip (where there are pedestrians) or the windy highways from hell (CA SR-17 with its constant switchbacks) than on the relatively safe 65 MPH stretches. In short, by any rational interpretation of California traffic laws, the 65 MPH maximum speed law is actively making the roads less safe, because on the roads where speeding would provably pose the least additional risk, the law restricts your speed in the strictest way possible, and on the roads where speeding would provably pose the most additional risk, the law restricts your speed in the most lax way possible.
And people wonder why I think traffic laws are almost entirely written by idiots.
How often do these things fall out of the sky, and does the added revenue offset the lives lost when they do?
Just saying.
I tend to think that drones should be used only in unusual circumstances, where unusual is translated as "high reward and low risk." Locating a lost hiker in a national park qualifies. Raising traffic fine revenue does not.
Without statistics on what percentage of people are driving buzzed, you can't make the assertion that they are a greater risk than the general population. To get usable statistics, the police need to randomly pull over people who are doing nothing wrong and subject them a breathalyzer test with the promise that this will not be used against them no matter what they blow (so that your statistics aren't distorted by having to wait to get a blood test) and find out how many people are actually driving buzzed.
Once you have that information, the analysis is fairly trivial. If the percentage of people who blow over a particular limit in a random sampling is significantly less (statistical significance here) than the percentage of at-fault drivers who tested over that limit, then the limit is too high, because those people are causing a disproportionate number of accidents. If the percentage of randomly selected people who blow over that limit is similar to or more than the percentage of at-fault drivers, then the limit is either about right or unreasonably low, and might even be doing more harm than good.
To put it less abstractly, if 60% of accidents are caused by someone who is driving at or above some particular level of buzz (say.02), but 80% of drivers are buzzed by that standard, then the buzzed drivers are being so much more attentive than average drivers that they are actually causing fewer accidents than other people. Thus, penalizing them further does not increase safety. If 60% of accidents are caused by someone who is driving while buzzed (by that same standard), and only 5% of drivers are buzzed (by that standard), then the buzzed drivers are causing many more accidents than your average person, and penalizing them would increase safety.
Without a complete picture, any assertion that lowering or raising the limit will have any particular effect on the number of accidents is basically pure speculation, with no scientific basis. It's like reading that 73% of males develop male pattern baldness and then saying that MPB is more common in males than the general population, without knowing what percentage of non-males exhibit MPB. It happens to be correct, but without that additional piece of information, the conclusion is a non sequitur. This is really no different than the cell phone bans, and their logic is faulty for precisely the same reason.
Stop saddling me with your damn phone interface and we'll see.
Hear, hear. I put up with those sorts of interfaces on my phone because of what it is. The interface inherently must be limited, or else it would not be usable on a tiny screen when operated by big, clumsy fingers.
When I'm on a computer, I have a nice, big screen, a mouse, and a keyboard. There's plenty of screen real estate to use for things like multiple windows with scroll bars and title bars, tabs, navigation controls galore, etc. There is no good reason to be stingy in terms of your user interface. If I wanted a limited UI, I would have bought a tablet in the first place.
Right about now, I'm thinking. Up until now, everything they did was at least ostensibly legal, albeit unethical. This, however, is pretty much open and shut blackmail and extortion, both of which are federal crimes in the United States and nearly every other country in the world. In the U.S., a single letter like this is enough to qualify the sender for two separate federal charges with a maximum combined sentence of three years in federal prison, on top of various state and local charges that may also apply.
Intel might be the first to do it on a CPU die, but they're not the first to do on-silicon inductors by any stretch. Switching regulators with inductors on silicon have been commercially available for several years now. The R-78 and MIC33030, for example, are drop-in replacements for linear regulators, with all components on die.
The real question in my mind is why anyone still uses linear regulators for anything, but I digress.
Not quite. You don't have to make a copy of a seed to eat it; you only do so if you choose to plant it. By contrast, you do have to make an ephemeral copy of digital content to do pretty much anything with it (other than use it as a frisbee if it happens to be on physical media).
Actually, digital music and software do create copies of themselves when used. That copy is specifically exempted in copyright law because it is a necessary part of the use of the product.
In fact it is stunningly obvious that such an abstract algorithm be processed on a computer. This obviousness itself is a major problem with software patents.
That's not really the problem. It was stunningly obvious that the assembly line (Olds patent) would involve people doing the assembly. That, in itself, does not make the patent obvious.
What makes software algorithms problematic is that there is usually exactly one way to do something, or very nearly so. Therefore, a patent on "a means to do X" usually translates to a patent on "doing X". Given a mathematical data compression algorithm, for example, apart from performance optimizations by vectorization and other tricks, there's usually exactly one way to implement it that doesn't result in precision loss.
The intent of patent law has always been that the obviousness test is not on the problem to be solved or the implementation of the steps to solve it, but rather on whether the steps would be obvious to someone who had not seen them, given the task it is supposed to concern. If there is exactly one way to do it, or exactly one way to do it without performing extra steps, or exactly one way to do it that isn't utterly idiotic, then the steps are obvious, and do not deserve a process patent.
For this reason, any patent in which the supposedly inventive step is "on a computer" is bogus, not because the concept of doing it on a computer is obvious, but because it is begging the question. The problem can be stated as "How would I do this on a computer?" and the answer is "On a computer." The obviousness test fails because the obvious way to do things on a computer is to approximate the way it is done in the real world. Therefore, the steps are obvious. If the steps are substantially unusual in some way that does not occur in the real world, then there's something non-obvious, and it might be worthy of a patent.
Each application should be accompanied by an xkcd style cartoon (probably longer than 4 panels, but typically less than 30) describing the invention and why it's useful and novel. The typical format would be: this is the current situation; here's what we're adding that's new; here's some detail about our method. If the xkcd is terrible or opaque, that should be a red flag.
One of those rules requires that it not be obvious to a person having ordinary skill in the art. Therefore, the opinion of engineers as to whether it is or is not inventive is, or at least should be, relevant.
Slashdot Mirror
No, the 85th percentile is a good limit as a baseline. You should aggressively go after people who are driving above... say the 99th percentile. Enforcing any law that 15% of people break is generally a bad idea.
Highway 17's biggest problem is that they allow trucks and buses on it.
So, there's a 30+ MPH difference between the fastest and slowest vehicles. Anything more than about a 15 MPH speed difference translates to a very, very unsafe road.
When you get a truck going 35 in the right lane, all the cars who want to go 50 have to pass in the same fast lane as the cars who want to go 65. Half of them are scared to pass, and end up passing a 35 MPH truck at 40 MPH, creating a cascading backup that can stretch for miles. The other half are angry that they can't go 60+. It is in this sort of backup that accidents are likely to occur, particularly when changing lanes to pass the slow vehicle.
By comparison, 175 might be curvy, but AFAIK (I've never driven that road) it has only one lane in each direction, and where it is unsafe, it is obviously unsafe, so people don't drive like morons. In much the same way, highway 9 is a curvy nightmare, but the problem spots are clearly marked, and people generally respect the road. What makes 17 so bad is that what looks like a halfway decent road really isn't. :-)
Windy (long "I") is a synonym for winding. :-)
Cameras on poles aren't able to accurately judge a vehicle's speed, and aren't as likely to spot unsafe driving as something that can see large stretches of road at a time. Once you've identified a problem vehicle, then you can either reduce altitude to shoot the vehicle's plate or follow it until it passes a stationary traffic camera that can snap its plate.
AFAIK, most newer UAVs are autonomous, including takeoff and landing. They require an operator only when they discover something that warrants a person's attention. You could quite literally run these things up and down the road all day, and your only personnel costs would be the person examining the suspicious footage to write the tickets and the person filling them up with fuel.
This is why you post a speed. That speed dictates what is typically safe. If your car is well maintained and the road is wide open and dry, you can go a bit over that. If the road is wet and you have bald tires, you'd better go more slowly.
Whoever is patrolling that stretch of highway, and has footage showing you swerving between lanes at 85 while the rest of the cars are going 50. For the most part, roads self-regulate, and cars tend to run along at about the same speed. Having a few cars going 10 MPH slower to obey an arbitrary posted speed limit actually results in a significantly higher risk of accidents than having everyone going 10 MPH over the limit, on the average.
Unless they're on the approach path near an airport. When planes fly over I-880 in San Jose, for example, I'm pretty sure they're way below 500 feet in HAAT. There is basically no safe altitude for a UAV over that stretch of road, unless the UAV is piloted by a real person, coordinating with ATC.
No, and I don't guess that 5% have, either. Those numbers were entirely arbitrary.
License plate readers. That part isn't even hard.
Umm... we're not talking about commercial airlines here. That's an entirely different universe. The comparison was between drones and traffic aircraft, which almost certainly are not fly-by-wire. They're usually either small airplanes (e.g. Cessna) or small helicopters.
There are many differences:
Need I continue?
This is not to say that drones won't have fewer accidents per vehicle (assuming a single pilot who could have a heart attack), but if you can buy the drones for a few hundred bucks apiece, you can potentially deploy tens of thousands of them for less money than a single manned aircraft (over the long term). So even tiny failure rates can pose a very serious problem.
Which, on the New Jersey Turnpike, is about as likely as a polar bear on the New Jersey Turnpike.
And it would probably still be safe....
Because it usually is. As far as I'm concerned, the law should simply say, "You may not travel at a speed that is unsafe for the current road conditions." Anything demanding strict conformance to a posted number (rather than driving at a speed that feels safe) is just asking for people to ignore the law... or worse.
The worst example of a highway safety law is California's 65 MPH law. Except for a few roads where it is specifically posted at 70 MPH, it is illegal to drive faster than 65 MPH in California, period. All other speed limits are flexible, depending on driving conditions. What this means is:
So if you're running behind and trying to decide where to exceed the speed limit, you're better off speeding on the city street portions of your trip (where there are pedestrians) or the windy highways from hell (CA SR-17 with its constant switchbacks) than on the relatively safe 65 MPH stretches. In short, by any rational interpretation of California traffic laws, the 65 MPH maximum speed law is actively making the roads less safe, because on the roads where speeding would provably pose the least additional risk, the law restricts your speed in the strictest way possible, and on the roads where speeding would provably pose the most additional risk, the law restricts your speed in the most lax way possible.
And people wonder why I think traffic laws are almost entirely written by idiots.
How often do these things fall out of the sky, and does the added revenue offset the lives lost when they do?
Just saying.
I tend to think that drones should be used only in unusual circumstances, where unusual is translated as "high reward and low risk." Locating a lost hiker in a national park qualifies. Raising traffic fine revenue does not.
Without statistics on what percentage of people are driving buzzed, you can't make the assertion that they are a greater risk than the general population. To get usable statistics, the police need to randomly pull over people who are doing nothing wrong and subject them a breathalyzer test with the promise that this will not be used against them no matter what they blow (so that your statistics aren't distorted by having to wait to get a blood test) and find out how many people are actually driving buzzed.
Once you have that information, the analysis is fairly trivial. If the percentage of people who blow over a particular limit in a random sampling is significantly less (statistical significance here) than the percentage of at-fault drivers who tested over that limit, then the limit is too high, because those people are causing a disproportionate number of accidents. If the percentage of randomly selected people who blow over that limit is similar to or more than the percentage of at-fault drivers, then the limit is either about right or unreasonably low, and might even be doing more harm than good.
To put it less abstractly, if 60% of accidents are caused by someone who is driving at or above some particular level of buzz (say .02), but 80% of drivers are buzzed by that standard, then the buzzed drivers are being so much more attentive than average drivers that they are actually causing fewer accidents than other people. Thus, penalizing them further does not increase safety. If 60% of accidents are caused by someone who is driving while buzzed (by that same standard), and only 5% of drivers are buzzed (by that standard), then the buzzed drivers are causing many more accidents than your average person, and penalizing them would increase safety.
Without a complete picture, any assertion that lowering or raising the limit will have any particular effect on the number of accidents is basically pure speculation, with no scientific basis. It's like reading that 73% of males develop male pattern baldness and then saying that MPB is more common in males than the general population, without knowing what percentage of non-males exhibit MPB. It happens to be correct, but without that additional piece of information, the conclusion is a non sequitur. This is really no different than the cell phone bans, and their logic is faulty for precisely the same reason.
Hear, hear. I put up with those sorts of interfaces on my phone because of what it is. The interface inherently must be limited, or else it would not be usable on a tiny screen when operated by big, clumsy fingers.
When I'm on a computer, I have a nice, big screen, a mouse, and a keyboard. There's plenty of screen real estate to use for things like multiple windows with scroll bars and title bars, tabs, navigation controls galore, etc. There is no good reason to be stingy in terms of your user interface. If I wanted a limited UI, I would have bought a tablet in the first place.
Right about now, I'm thinking. Up until now, everything they did was at least ostensibly legal, albeit unethical. This, however, is pretty much open and shut blackmail and extortion, both of which are federal crimes in the United States and nearly every other country in the world. In the U.S., a single letter like this is enough to qualify the sender for two separate federal charges with a maximum combined sentence of three years in federal prison, on top of various state and local charges that may also apply.
Intel might be the first to do it on a CPU die, but they're not the first to do on-silicon inductors by any stretch. Switching regulators with inductors on silicon have been commercially available for several years now. The R-78 and MIC33030, for example, are drop-in replacements for linear regulators, with all components on die.
The real question in my mind is why anyone still uses linear regulators for anything, but I digress.
Not quite. You don't have to make a copy of a seed to eat it; you only do so if you choose to plant it. By contrast, you do have to make an ephemeral copy of digital content to do pretty much anything with it (other than use it as a frisbee if it happens to be on physical media).
Actually, digital music and software do create copies of themselves when used. That copy is specifically exempted in copyright law because it is a necessary part of the use of the product.
None of the above. I was talking about the risks involved in making meth.
Doing it repeatedly without blowing the doors off your house is the hard part.
The opinion of the inventor?
That's not really the problem. It was stunningly obvious that the assembly line (Olds patent) would involve people doing the assembly. That, in itself, does not make the patent obvious.
What makes software algorithms problematic is that there is usually exactly one way to do something, or very nearly so. Therefore, a patent on "a means to do X" usually translates to a patent on "doing X". Given a mathematical data compression algorithm, for example, apart from performance optimizations by vectorization and other tricks, there's usually exactly one way to implement it that doesn't result in precision loss.
The intent of patent law has always been that the obviousness test is not on the problem to be solved or the implementation of the steps to solve it, but rather on whether the steps would be obvious to someone who had not seen them, given the task it is supposed to concern. If there is exactly one way to do it, or exactly one way to do it without performing extra steps, or exactly one way to do it that isn't utterly idiotic, then the steps are obvious, and do not deserve a process patent.
For this reason, any patent in which the supposedly inventive step is "on a computer" is bogus, not because the concept of doing it on a computer is obvious, but because it is begging the question. The problem can be stated as "How would I do this on a computer?" and the answer is "On a computer." The obviousness test fails because the obvious way to do things on a computer is to approximate the way it is done in the real world. Therefore, the steps are obvious. If the steps are substantially unusual in some way that does not occur in the real world, then there's something non-obvious, and it might be worthy of a patent.
They would always be #927.
One of those rules requires that it not be obvious to a person having ordinary skill in the art. Therefore, the opinion of engineers as to whether it is or is not inventive is, or at least should be, relevant.