As recently as the 1950s, it was accepted that half the population (the female half, of course) "should" be unemployed. Except they didn't call it being unemployed, they called it being a homemaker. We'll need big social changes, but changes like that have happened before. Hopefully we'll pull it off without causing too much pain to too many people.
A UBI would be a good approach. Not enough to replace working, just to reduce the demand for jobs. So more high school and college students decide they don't need jobs in addition to going to school. So more married couples decide they can afford to have one of them stay home and care for the kids. Then as the supply of jobs gradually decreases, you can gradually increase the UBI to try to keep the demand matched. I'm not saying it will be easy, but it's possible to do it without huge social disruptions.
After all his efforts to help Trump get elected! He's been claiming for years the U.S. was out to get him, and he couldn't return to Sweden to face his rape charges because if he did they'd extradite him. But through all that time, the Obama administration never made any move to charge him with anything.
Then Trump comes into office with help from Assange. And hardly three months later, they're preparing to charge him. I don't think you got what you were hoping for!
I see the occasional discussion of carbon sequestration and that sort of thing, but far more often the "solution" is just a cloak hiding the proposer's socialist SJW motives.
Why are you making things up? Most of the solutions being proposed have nothing to do with socialism. By far the most popular proposal among economists is a carbon tax, which is about as non-political and pro-market as you could ask for. Make people pay for the damage they do to the environment, then let the market figure out the best way to deal with it. Other popular proposals include things like raising the fuel efficiency standards for cars, subsidizing renewable energy, increased funding for energy research, etc. If you think those are socialism, you have a strange idea of what the word means.
But instead you just give quotes from a bunch of people I've never heard of with titles like "former leader of the Communist Party USA" and "climate justice campaigner". Couldn't you have quoted present day, mainstream political figures instead? Of course not, because they don't believe those things. But since mainstream politicians aren't socialists, instead you quote a bunch of socialists, pretend they reflect the views of mainstream politicians, and then claim this discredits anyone who actually wants to do something about climate change.
Almost every comment posted so far about this story is totally wrong. Adversarial examples are a hot topic in deep learning right now. We've learned a lot about how they work and how to protect against them. They have nothing to do with "weak" versus "strong" AI. Humans are also susceptible to optical illusions, just different ones from neural nets. They don't mean that computers can never be trusted. Computers can be made much more reliable than humans. And they also aren't random failures, or something that's hard to create. In fact, they're trivial to create in a simple, systematic way.
They're actually a consequence of excessive linearity in our models. If you don't know what that means, don't worry about it. It's just a quirk of how models have traditionally been trained. And if you make a small change to encourage them to work in a nonlinear regime, they become much more resistant to adversarial examples. By the time fully autonomous cars hit the roads in a few years, this should be a totally solved problem.
If you build deep learning systems, you need to care about this. If you don't, you can ignore it. It's not a problem you need to care about, any more than you care what activation function or regularization method your car is using.
Because it's not like EVERY SINGLE AUTONOMOUS CAR currently being tested already does exactly this, forcing the driver to take over when it encounters a situation it can't handle. Or like the driver assistance features already available in cars do this the other way, such as hitting the breaks for you if you're about to run into something.
But it's ok, they added "with machine learning", so that makes it new. I guess that's the new version of "on the internet". Take anything that people are already doing, add the words "with machine learning", and now you can patent it.
This is just absurd. Have you ever heard of any installer for any product first doing a check to see if your drive can "withstand the stress" of the installation? What does that even mean? What would you look for?
Now let's do some math. The average life of a computer is... maybe five years? So 60 months. And "data loss or damage to software or hardware" is a really vague category. People do lose data sometimes. Software suffers corruption. Hardware wears out. Often that's what leads someone to get a new computer. So probably about 2% of all computers will experience "data loss or damage to software or hardware" in any month, even without any OS upgrade.
Now how many millions of computers in the US have been upgraded to Windows 10? Multiply that by 2% and you get tens of thousands, maybe hundreds of thousands, that just by chance happened to develop problems in the month after the upgrade. For reasons unrelated to the upgrade. The lawyers are looking at that number and seeing dollar signs.
That isn't how patents work. Every claim is independently asserted to be a novel invention. If someone violates even one claim, they're violating the patent.
By the way, "an accessory display configured to present visual content" is also known as an external monitor. An "input device configured to detect a touch event" is also known as a trackpad. The Atrix dock had both. So have countless other docking stations for various computers over the years.
Or more simply, just put the money into the state's general fund. You're trying to make sure it gets spent "fairly", but I'm not sure your system really achieves that. It would leave out anyone who doesn't file a tax return (children, a lot of poor people).
But yeah, the main point is that cities shouldn't get to keep the money from fines. We need to make sure they're really just doing it for public safety, not to get more money.
1. An electronic accessory device, comprising: an operational component that provides an output to a user; a housing carrying the operational component, the housing having a recess; and a control interface coupled to the operational component and configured to receive a control signal from an electronic host device when the electronic host device is positioned within the recess and coupled to the control interface, wherein the electronic accessory device is inoperable without the electronic host device being coupled to the control interface.
Can you point out to me what are the "new things" in that? I'd say the Motorola Atrix fit every last piece of that. But Apple is claiming it as novel and asking for patent protection on it.
Maybe. Or maybe not. If the conditions are such reasonable, industry standard ones, why isn't wikileaks disclosing what they are? Given Assange's history, he has zero credibility in my book. For an organization that's supposedly dedicated to public disclosure, they're awfully fond of keeping things secret. I mean really, they won't tell us what conditions they're asking the companies to agree to? Then I certainly won't assume they're as reasonable as he wants us to think they are.
It's important to see both sides of the issue. Yes, this law would almost certainly violate the first amendment. Yes, it could easily be abused. But it's also a sincere attempt at fixing a real problem.
Someone falsely accuses you of some terrible crime. Maybe you have a bad breakup and your ex decides to get revenge by accusing you of child abuse or theft or something like that. It gets reported in the local news. The accusations are totally false, the police figure that out really quickly, and all charges get dropped. But still, if anyone googles your name, the top hits are all news stories about you being accused of something terrible. It's ruining your life. When you apply for jobs they first seem really interested, and then suddenly tell you to go away, and you know exactly why. There's nothing you can do about it.
It's easy to criticize this bill as a badly thought out idea—which it is. But it's still a real problem. So anyone who criticizes this needs to be able to answer, what should we do instead? "Ignore the problem and pretend it doesn't exist" isn't an answer. It's a real problem and it's hurting real people. So what should we do about it?
Ordinary old-style meters do an adequate job, and give employment to a lot of meter-readers. (That's a good thing, by the way).
Why do you think it's a good thing? I usually consider inefficiency a bad thing. There's no benefit to society from making people do busywork that a computer can do far better. Why waste the time of all those meter readers when they could instead spend it doing a useful job that creates value for someone? Or if there really is nothing better for them to do, let them go home and spend their time however they want. Making them do useless busywork to get their salary doesn't help you or them or anyone else.
They left out a critical detail: is this by the employee's choice or the company's? Are they less likely to choose to leave, or less likely to get laid off? If it's the employee's choice, is it because people like their jobs better, or because they have fewer other options?
Without knowing that, I can't tell if this makes Seattle a better or worse place to work. Not getting laid off is good. Liking your job is good. Having few options is bad. In any case, I doubt it has much to do with loyalty.
It happened about ten years ago with the rise of GPUs for general purpose computing. Suddenly we could do a lot of things 10-100 times faster than before. You program GPUs really differently than CPUs, so we had to rewrite a lot of code and design new algorithms. But the benefit was huge.
It may be happening again with specialized chips for deep learning, like Google's TPU. These chips are designed for just one class of applications, but it's a really important class, and they can be 10x faster or more efficient for those applications.
There've been other times when a new generation brought a sudden major improvement in speed, like with vector units or multicore CPUs. But always at the cost of having to rewrite how your code works.
Now if you want new chips that work just like the old ones and run the same programs as before, just 10x faster, sorry. That isn't likely to happen. Huge jumps like that require major changes of approach.
You just demonstrated my point. There's no such thing as "the fastest possible way". Is your data uniformly distributed over a fixed range? If so, you can use a bucket sort to process it in O(n) time. But if the distribution is non-uniform, a bucket sort can be very slow. Maybe your input data is partially sorted. If so, some algorithms will take advantage of that to sort it faster and others won't. Did you ever think to wonder whether the sort function is multithreaded? Probably not, because if you're already using multiple threads at a higher level, that would just slow things down. But if you aren't, that could speed things up a lot.
This is a great example of the law of leaky abstractions. In most cases, just calling the framework's sort function is the best thing to do. Except that every now and then, it isn't. A good software engineer understands that. They know the framework provides an abstraction, and that abstraction sometimes breaks down, and they can deal with it when it does. Even if they don't know all the low level details, they know enough to know what questions to ask.
Even worse, a lot of candidates don't even know what a bubble sort is, or that there's such a thing as a bubble sort. They don't understand that there's more than one way of sorting a list, or that some ways are more efficient than others. They don't realize that the best method in one case could be different from the best method in another. They don't realize that some methods scale better than others with list size, or that some are faster if the input list is partially sorted, or that some require extra memory and others don't, or that some can be parallelized better than others.
And these are people who claim to be experienced software engineers.
I often ask about sorting on interviews, and it's a great way to quickly tell how competent someone is. I don't care if they understand the details of particular algorithms, or if they can write them from memory. I care whether they get the concepts and know what questions to ask.
At the other extreme, I used to work with someone who started every interview by asking, "Write a complete, working program in any language of your choice that prints 'Hello World' ten times." That was also illuminating, in a horrifying sort of way. You wouldn't believe how many people struggled with that. We usually ended those interviews pretty quickly.
Here's a recent article that seems relevant. It's talking about suicide rather than homicide, but the point is the same. People who try to kill themselves with drugs or knives only have a 5% chance of dying. People who use guns have a 90% chance. Guns really are different. They're amazingly effective tools for killing. That's what they're designed for. Now add the fact that you can carry one with you everywhere you go, and it only takes a moment to pull it out and start shooting. The time from when you think, "I want to kill this guy," to when he's lying on the floor with a fatal injury might only be seconds. There's nothing else like them.
I dare you to even try to read your ISP's terms of service. Or the terms of service of lots of websites you access all the time, that you supposedly agree to by accessing the site. Or the contracts you agree to by opening a bank account or renting a car or buying a plane ticket. You'll find that most of those contracts incorporate other documents by reference, which may add up to hundreds of pages, all of which you're supposed to have read.
This action isn't about what businesses have to read. It's about what information they have to disclose to their customers.
Tech reporting has gotten more negative than it used to be. I mean really, could you ask for a clearer example? We just elected a totally insane president who wants to build more nukes and denounces the press as the enemy of the people. So obviously we should blame social media for people being more stressed???
The main way people get toxoplasma is from food, especially undercooked meat and unwashed vegetables. But the media makes you think it's from cats because the headline "Cats Make You Crazy" is much better clickbait than "Eating Undercooked Meat Makes You Crazy". Transmission from cats is much rarer. Also, cats get infected by eating infected mice. If you have indoor cats who eat Friskies instead of mice, they aren't at risk and neither are you. Unless you eat undercooked meat, of course, but the media would rather warn you about fake risks than real ones.
Maybe you misread the summary? A similar measure passed the house unanimously last year, but was blocked in the senate "amid opposition by a handful of Republican lawmakers." They could easily have overridden a veto. But they never let it get that far, and it was Republicans who blocked it.
I won't draw any partisan conclusions from that. It clearly had strong support in both parties, and most Republicans supported it. But trying to blame Obama doesn't fit the facts.
Let's face it, now one is going to give out trade secrets in those things.
Not trade secrets, but non-public information is possible. Apple's WWDC is famous for working like that. Most of the sessions are under NDA, and it's when they present all the new technologies that will be coming in future OS updates. If you want to find out about those technologies in advance, going to the conference is really useful.
Maybe it will take 30-50 years to make most human labor obsolete, or maybe even longer. But that doesn't mean everything will go along as usual until one day all workers suddenly become unneeded. Workers will steadily get replaced, and we need to deal with that as it happens. Automation is progressing really quickly, and it's doing it right now. We can't wait until the process is all done and then say, "Ok, now it's time to deal with it."
Slashdot Mirror
As recently as the 1950s, it was accepted that half the population (the female half, of course) "should" be unemployed. Except they didn't call it being unemployed, they called it being a homemaker. We'll need big social changes, but changes like that have happened before. Hopefully we'll pull it off without causing too much pain to too many people.
A UBI would be a good approach. Not enough to replace working, just to reduce the demand for jobs. So more high school and college students decide they don't need jobs in addition to going to school. So more married couples decide they can afford to have one of them stay home and care for the kids. Then as the supply of jobs gradually decreases, you can gradually increase the UBI to try to keep the demand matched. I'm not saying it will be easy, but it's possible to do it without huge social disruptions.
After all his efforts to help Trump get elected! He's been claiming for years the U.S. was out to get him, and he couldn't return to Sweden to face his rape charges because if he did they'd extradite him. But through all that time, the Obama administration never made any move to charge him with anything.
Then Trump comes into office with help from Assange. And hardly three months later, they're preparing to charge him. I don't think you got what you were hoping for!
I see the occasional discussion of carbon sequestration and that sort of thing, but far more often the "solution" is just a cloak hiding the proposer's socialist SJW motives.
Why are you making things up? Most of the solutions being proposed have nothing to do with socialism. By far the most popular proposal among economists is a carbon tax, which is about as non-political and pro-market as you could ask for. Make people pay for the damage they do to the environment, then let the market figure out the best way to deal with it. Other popular proposals include things like raising the fuel efficiency standards for cars, subsidizing renewable energy, increased funding for energy research, etc. If you think those are socialism, you have a strange idea of what the word means.
But instead you just give quotes from a bunch of people I've never heard of with titles like "former leader of the Communist Party USA" and "climate justice campaigner". Couldn't you have quoted present day, mainstream political figures instead? Of course not, because they don't believe those things. But since mainstream politicians aren't socialists, instead you quote a bunch of socialists, pretend they reflect the views of mainstream politicians, and then claim this discredits anyone who actually wants to do something about climate change.
your answer, and the certainty with which you issue it depends on whether you're a member of the AGW secular religion.
lol
Almost every comment posted so far about this story is totally wrong. Adversarial examples are a hot topic in deep learning right now. We've learned a lot about how they work and how to protect against them. They have nothing to do with "weak" versus "strong" AI. Humans are also susceptible to optical illusions, just different ones from neural nets. They don't mean that computers can never be trusted. Computers can be made much more reliable than humans. And they also aren't random failures, or something that's hard to create. In fact, they're trivial to create in a simple, systematic way.
They're actually a consequence of excessive linearity in our models. If you don't know what that means, don't worry about it. It's just a quirk of how models have traditionally been trained. And if you make a small change to encourage them to work in a nonlinear regime, they become much more resistant to adversarial examples. By the time fully autonomous cars hit the roads in a few years, this should be a totally solved problem.
If you build deep learning systems, you need to care about this. If you don't, you can ignore it. It's not a problem you need to care about, any more than you care what activation function or regularization method your car is using.
Because it's not like EVERY SINGLE AUTONOMOUS CAR currently being tested already does exactly this, forcing the driver to take over when it encounters a situation it can't handle. Or like the driver assistance features already available in cars do this the other way, such as hitting the breaks for you if you're about to run into something.
But it's ok, they added "with machine learning", so that makes it new. I guess that's the new version of "on the internet". Take anything that people are already doing, add the words "with machine learning", and now you can patent it.
This is just absurd. Have you ever heard of any installer for any product first doing a check to see if your drive can "withstand the stress" of the installation? What does that even mean? What would you look for?
Now let's do some math. The average life of a computer is... maybe five years? So 60 months. And "data loss or damage to software or hardware" is a really vague category. People do lose data sometimes. Software suffers corruption. Hardware wears out. Often that's what leads someone to get a new computer. So probably about 2% of all computers will experience "data loss or damage to software or hardware" in any month, even without any OS upgrade.
Now how many millions of computers in the US have been upgraded to Windows 10? Multiply that by 2% and you get tens of thousands, maybe hundreds of thousands, that just by chance happened to develop problems in the month after the upgrade. For reasons unrelated to the upgrade. The lawyers are looking at that number and seeing dollar signs.
That isn't how patents work. Every claim is independently asserted to be a novel invention. If someone violates even one claim, they're violating the patent.
By the way, "an accessory display configured to present visual content" is also known as an external monitor. An "input device configured to detect a touch event" is also known as a trackpad. The Atrix dock had both. So have countless other docking stations for various computers over the years.
Or more simply, just put the money into the state's general fund. You're trying to make sure it gets spent "fairly", but I'm not sure your system really achieves that. It would leave out anyone who doesn't file a tax return (children, a lot of poor people).
But yeah, the main point is that cities shouldn't get to keep the money from fines. We need to make sure they're really just doing it for public safety, not to get more money.
Here is claim 1 from Apple's patent application.
1. An electronic accessory device, comprising: an operational component that provides an output to a user; a housing carrying the operational component, the housing having a recess; and a control interface coupled to the operational component and configured to receive a control signal from an electronic host device when the electronic host device is positioned within the recess and coupled to the control interface, wherein the electronic accessory device is inoperable without the electronic host device being coupled to the control interface.
Can you point out to me what are the "new things" in that? I'd say the Motorola Atrix fit every last piece of that. But Apple is claiming it as novel and asking for patent protection on it.
Maybe. Or maybe not. If the conditions are such reasonable, industry standard ones, why isn't wikileaks disclosing what they are? Given Assange's history, he has zero credibility in my book. For an organization that's supposedly dedicated to public disclosure, they're awfully fond of keeping things secret. I mean really, they won't tell us what conditions they're asking the companies to agree to? Then I certainly won't assume they're as reasonable as he wants us to think they are.
It's important to see both sides of the issue. Yes, this law would almost certainly violate the first amendment. Yes, it could easily be abused. But it's also a sincere attempt at fixing a real problem.
Someone falsely accuses you of some terrible crime. Maybe you have a bad breakup and your ex decides to get revenge by accusing you of child abuse or theft or something like that. It gets reported in the local news. The accusations are totally false, the police figure that out really quickly, and all charges get dropped. But still, if anyone googles your name, the top hits are all news stories about you being accused of something terrible. It's ruining your life. When you apply for jobs they first seem really interested, and then suddenly tell you to go away, and you know exactly why. There's nothing you can do about it.
It's easy to criticize this bill as a badly thought out idea—which it is. But it's still a real problem. So anyone who criticizes this needs to be able to answer, what should we do instead? "Ignore the problem and pretend it doesn't exist" isn't an answer. It's a real problem and it's hurting real people. So what should we do about it?
Ordinary old-style meters do an adequate job, and give employment to a lot of meter-readers. (That's a good thing, by the way).
Why do you think it's a good thing? I usually consider inefficiency a bad thing. There's no benefit to society from making people do busywork that a computer can do far better. Why waste the time of all those meter readers when they could instead spend it doing a useful job that creates value for someone? Or if there really is nothing better for them to do, let them go home and spend their time however they want. Making them do useless busywork to get their salary doesn't help you or them or anyone else.
They left out a critical detail: is this by the employee's choice or the company's? Are they less likely to choose to leave, or less likely to get laid off? If it's the employee's choice, is it because people like their jobs better, or because they have fewer other options?
Without knowing that, I can't tell if this makes Seattle a better or worse place to work. Not getting laid off is good. Liking your job is good. Having few options is bad. In any case, I doubt it has much to do with loyalty.
It happened about ten years ago with the rise of GPUs for general purpose computing. Suddenly we could do a lot of things 10-100 times faster than before. You program GPUs really differently than CPUs, so we had to rewrite a lot of code and design new algorithms. But the benefit was huge.
It may be happening again with specialized chips for deep learning, like Google's TPU. These chips are designed for just one class of applications, but it's a really important class, and they can be 10x faster or more efficient for those applications.
There've been other times when a new generation brought a sudden major improvement in speed, like with vector units or multicore CPUs. But always at the cost of having to rewrite how your code works.
Now if you want new chips that work just like the old ones and run the same programs as before, just 10x faster, sorry. That isn't likely to happen. Huge jumps like that require major changes of approach.
They varied, but a lot were recent CS graduates.
You just demonstrated my point. There's no such thing as "the fastest possible way". Is your data uniformly distributed over a fixed range? If so, you can use a bucket sort to process it in O(n) time. But if the distribution is non-uniform, a bucket sort can be very slow. Maybe your input data is partially sorted. If so, some algorithms will take advantage of that to sort it faster and others won't. Did you ever think to wonder whether the sort function is multithreaded? Probably not, because if you're already using multiple threads at a higher level, that would just slow things down. But if you aren't, that could speed things up a lot.
This is a great example of the law of leaky abstractions. In most cases, just calling the framework's sort function is the best thing to do. Except that every now and then, it isn't. A good software engineer understands that. They know the framework provides an abstraction, and that abstraction sometimes breaks down, and they can deal with it when it does. Even if they don't know all the low level details, they know enough to know what questions to ask.
Even worse, a lot of candidates don't even know what a bubble sort is, or that there's such a thing as a bubble sort. They don't understand that there's more than one way of sorting a list, or that some ways are more efficient than others. They don't realize that the best method in one case could be different from the best method in another. They don't realize that some methods scale better than others with list size, or that some are faster if the input list is partially sorted, or that some require extra memory and others don't, or that some can be parallelized better than others.
And these are people who claim to be experienced software engineers.
I often ask about sorting on interviews, and it's a great way to quickly tell how competent someone is. I don't care if they understand the details of particular algorithms, or if they can write them from memory. I care whether they get the concepts and know what questions to ask.
At the other extreme, I used to work with someone who started every interview by asking, "Write a complete, working program in any language of your choice that prints 'Hello World' ten times." That was also illuminating, in a horrifying sort of way. You wouldn't believe how many people struggled with that. We usually ended those interviews pretty quickly.
Here's a recent article that seems relevant. It's talking about suicide rather than homicide, but the point is the same. People who try to kill themselves with drugs or knives only have a 5% chance of dying. People who use guns have a 90% chance. Guns really are different. They're amazingly effective tools for killing. That's what they're designed for. Now add the fact that you can carry one with you everywhere you go, and it only takes a moment to pull it out and start shooting. The time from when you think, "I want to kill this guy," to when he's lying on the floor with a fatal injury might only be seconds. There's nothing else like them.
I dare you to even try to READ the order.
I dare you to even try to read your ISP's terms of service. Or the terms of service of lots of websites you access all the time, that you supposedly agree to by accessing the site. Or the contracts you agree to by opening a bank account or renting a car or buying a plane ticket. You'll find that most of those contracts incorporate other documents by reference, which may add up to hundreds of pages, all of which you're supposed to have read.
This action isn't about what businesses have to read. It's about what information they have to disclose to their customers.
Tech reporting has gotten more negative than it used to be. I mean really, could you ask for a clearer example? We just elected a totally insane president who wants to build more nukes and denounces the press as the enemy of the people. So obviously we should blame social media for people being more stressed???
The main way people get toxoplasma is from food, especially undercooked meat and unwashed vegetables. But the media makes you think it's from cats because the headline "Cats Make You Crazy" is much better clickbait than "Eating Undercooked Meat Makes You Crazy". Transmission from cats is much rarer. Also, cats get infected by eating infected mice. If you have indoor cats who eat Friskies instead of mice, they aren't at risk and neither are you. Unless you eat undercooked meat, of course, but the media would rather warn you about fake risks than real ones.
Maybe you misread the summary? A similar measure passed the house unanimously last year, but was blocked in the senate "amid opposition by a handful of Republican lawmakers." They could easily have overridden a veto. But they never let it get that far, and it was Republicans who blocked it.
I won't draw any partisan conclusions from that. It clearly had strong support in both parties, and most Republicans supported it. But trying to blame Obama doesn't fit the facts.
Let's face it, now one is going to give out trade secrets in those things.
Not trade secrets, but non-public information is possible. Apple's WWDC is famous for working like that. Most of the sessions are under NDA, and it's when they present all the new technologies that will be coming in future OS updates. If you want to find out about those technologies in advance, going to the conference is really useful.
Maybe it will take 30-50 years to make most human labor obsolete, or maybe even longer. But that doesn't mean everything will go along as usual until one day all workers suddenly become unneeded. Workers will steadily get replaced, and we need to deal with that as it happens. Automation is progressing really quickly, and it's doing it right now. We can't wait until the process is all done and then say, "Ok, now it's time to deal with it."