Trump talked tough on NK, which Kim took advantage of to demonstrate that it was all just bluster and ultimately there was little the US could do because of China and the fact that NK had a viable nuclear deterrent.
Kim was able to use that as a bargaining chip when opening a dialogue with South Korea and the US. De-nuclearization suits NK, because it will include US nuclear weapons in the area too.
Kim quickly made concessions and progress with SK, which put pressure on Trump to not waste the opportunity. So Trump went to that meeting ready to make big concessions - end the wargames, de-nuclearize, and start opening up NK's economy with sanctions relief. Kim has boosted himself to the level of international statesman, praised by the POTUS and with a path towards a post-dictatorship life that isn't rotting in jail or hanging from a noose.
Even if it's for the wrong reasons, Trump's decision is the right one.
It's in the US's interests to do a good trade deal with China. There are many mutual benefits. However, destroying a huge company like ZTE and putting hundreds of thousands of Chinese out of work is unlikely to help secure such a deal.
The crime is breaking sanctions. Sanctions that Trump may have already decided to end (it's not clear ATM what was agreed at the Kim/Trump meeting). The punishment is arbitrary, it's not calculated to offset some loss or restore something that has been taken. And cancelling it probably nets the US a nice chunk of cash, because if ZTE simply died it wouldn't be paying that fine that it is now committed to clearing.
Plus it would have screwed all the US ZTE customers who would be left without support.
Hate to defend them but it seems they are using an audio fingerprint, similar to Shazam or Google's music ID system. The amount of data transmitted is very small, certainly not 100MB. Since it's just frequency/time histograms I'd estimate it to be in the tens of kilobytes range max, probably less.
Actually they already won, they just don't seem to have realized it.
I've been following these rules as they develop. The early draft did have a requirement for websites to actively filter new content for copyright infringement, based on submissions by copyright holders. That was removed ages ago, but people still seem to be panicking about it for some reason.
There is a requirement to remove infringing material after it has been reported, the same as the DMCA. The DMCA isn't great, but it's main flaw is people spamming automated DMCA notices with bots.
And in fact most web sites do pre-screen material as it is uploaded anyway, e.g. to block known illegal images because their customers don't like their feeds being filled with child pornography and ISIS beheading videos.
Also the "link tax" is long, long dead. There was a proposal to allow charging for "snippets" in search results, but it looks like it may well be DoA because where it was been tried it was a disaster. In Spain Google shut down it's news service and news sites saw a 6-30% loss of traffic. In Germany the news sites gave Google a free licence to use snippets anyway.
So all in all it's mostly just panicking about out of date, long rejected draft proposals.
I don't have that detail but I can tell you my experience as a user of machine translation for 15+ years.
Originally it only really worked on formal documents, and even then only produced something you could barely understand. The biggest issue seemed to be that it didn't understand context at all.
Google made some early improvements in making the translated text sound more natural. They also managed to fix a lot of common phrases that didn't quite fit the standard grammar model and thus didn't used to get translated properly. Apparently they did that by using the web as a resource for natural language and by allowing users to submit corrections.
Then AI started to be used. Baidu were the first I think and their Chinese/English translation was a huge improvement over everything else. It seemed to work slightly better going from English to Chinese though, and when Google released their AI updates not long after Chinese to English became nearly perfect.
It's actually incredible how good it is now. Often the resulting translation is not only accurate and seemingly context aware, it sounds like something a person might actually say. You don't have to think about what you are writing either. Before you had to be careful to phrase things so that the software could understand it, but not any more.
There are still some issues, like the way Japanese newspaper headlines often get translated as if it was a person speaking about their own experience (e.g. some houses were flooded, but the translation is "my house was flooded" because the software assumes that context), but for conversations between two people it's like Star Trek or something.
Fossil car manufacturers are already panicking, especially part manufacturers. For example, most Japanese manufacturers bet on hybrid tech and are now realizing that they don't have the experience and patents for EV drivetrains. Companies that have been making gearboxes for 80 years are suddenly finding that the Chinese have already cornered the market that is their future.
Same in Europe. European manufacturers are having to go to China for EV tech. Really only Nissan and Renault have anything significant of their own.
All the battery tech is Japanese, Korean and Chinese too.
I'm using them for data and basic 5V/900mA power, not the higher rated power delivery stuff.
For the high power stuff you need to be able to handle the higher voltage and to negotiate it with the host. I don't actually have anything that can deliver higher power anyway, just bog standard chargers up to 5V/3A.
Let's not let the trolls distract from how important this story is. Prostate cancer is one of the biggest causes of death for men, and like most cancer it's much easier to treat if caught early. Encouraging guys to self-check regularly is really important, but this test should be able to detect it even earlier.
Fortunately if caught early there are effective treatments. Check yourself once a month and don't hesitate to see your doctor if you find something.
Yeah. I was using micro USB because there were no USB-C connectors that I could hand solder. Now there are some that are mixed surface mount and through-hole with side tabs for mechanical strength, so I use those.
The most annoying part is that to actually test your device for full USB 2.0 compliance you need some expensive and hard to find equipment. The test suite only officially works with a bunch of ancient powered hubs, and you need several of them because the test suite needs to check compatibility with daisy chained controllers. Oh, and you need an Intel USB 2.0 chipset on your PC because they have a special compliance testing driver.
Things got a bit better with USB 3.0 in that you can just buy a compliance testing rig. It's very expensive but at least it's simple and available.
Well, the other problem is that to fully support the USB spec and pass the test suite your firmware needs to be quite complex and implement a bunch of rarely used stuff. Most manufacturers supply a stack that passes the suite, but it can be too large to fit into small bootloaders and the like, and of course you still need to get the electrical side right.
The use USB 1.1 speeds, aka low speed (1.5Mb) or full speed (12Mb). However, you are supposed to certify them against the latest version of the spec, so most peripherals around today are USB 2.0. They only use the lower speed but they use the USB 2.0 protocol and should pass the USBIF 2.0 test suite.
Maybe you forgot what a mess USB 1 was... Keyboards that didn't work in the BIOS, for example. It wasn't until USB 1.1 that it got half way good, and it only really took off with USB 2.0 which finally offered enough speed for things like external drives and scanners.
USB 3.0 is the protocol. It supports various types of cable. If you want higher speeds you need a SuperSpeed cable and port (coloured blue). USB-C is just a connector, it can work with USB 1.1 through to 3.1, by itself it doesn't give you any extra functionality.
It's like SCSI back in the day. You had SCSI, SCSI 2 and SCSI 3 which were the protocols. SCSI 2 introduced support for removable drives and peripherals like scanners, for example. But then you also had various different physical layers. SCSI, Wide SCSI, Ultra Wide SCSI, fibre channel, LVDS, LVDS single ended, 25 pin, 50 pin, 68 pin, 80 pin...
I agree they need to communicate this better. They have always been terrible at doing that.
Your Tesla has a huge battery that is actively cooled by the car during charging. The main limitation on phone charging speeds is heat generated during charging. If the phone charged any faster the battery would be damaged by heat, possibly failing catastrophically (expansion, fire).
There isn't really a good solution to this, unless you want a much larger phone that can dissipate more heat, maybe with a little fan or some water cooling built in.
No, it's due to the way USB 3 delivers up to 100W and cheap cable manufacturers not properly testing that.
With USB 3 the device can negotiate for high power delivery, which involves increasing the supplied voltage from 5V up to 20V. Due to physics increasing the voltage reduces the current needed to deliver 100W, which in turn reduces the amount of heat generated in the cable. Heat is wasted energy.
The problem is that the cheap cables don't implement the spec properly and are not rated for 20V/5A, so can end up supplying 20V on the wrong pins and damaging equipment. The equipment needed to properly test USB 3 power delivery costs thousands of Euro/USD, so some companies just skipped it.
The original USB spec has a lot of features that are rarely used these days. Alternative functions, alternative ways of sending data (e.g. HID reports over the control endpoint) that no-one uses. It was needlessly complex.
Then it got streamlined a bit, but also additional hacks were introduced such as the keyboard boot protocol that allowed USB keyboards to work in the BIOS. USB 2.0 increased the data rate, and battery charging was added but proved to be too slow so several proprietary protocols were introduced.
Then USB 3 came along and added extra pins to support higher speeds and higher power delivery. So by now it's the 4th generation at least (USB 1.0, 1.1, 2.0, 3.0, and now 3.1). And it actually works pretty well.
I think they could have done more for phone charging and for external batteries. Support for delivering power at say 3.7V (nominal LiPo voltage), to avoid the need for a step-up converter when using an external battery, would have been useful. At the moment a 5000mAh battery will only deliver about 3000mAh to the phone after conversion losses.
The final spec was actually the same as the preliminary one, the issue is that some companies just didn't bother to build their stuff properly. The test equipment and certification tools cost money, and they decided they could just bodge it for the â1 cables.
Having said that I buy decent but not expensive cables and have had no issues with USB 3. It works great for power delivery and data. I've started using it in my personal electronics projects (in USB 2.0 mode) as well.
This could be as little as adding the ability to read road signs, something that the old AP1 hardware could do but which the newer AP2 hardware has lacked for years.
It's probably just Tesla finally enabling all the cameras that they have installed in the car. At the moment only half of them are actually in use.
To get to fully autonomy they need to add the ability to read road signs, the ability to differentiate different kinds of vehicles (cars/bikes), and some kind of 3D vision (probably comparing consecutive video frames since they don't have stereo cameras), heavy redundancy so the car can act safely when cameras get splashed with rain or mud. They they need to take that data and build a 3D model of the world, and figure out how to navigate around it.
Keep in mind that they have been selling full self driving for a couple of years now, and promised that it would drive you to work and then go off and park itself, more than Google's system can do and using a vastly inferior sensor suite.
Implementing a proper RNG isn't expensive or time consuming at all. Windows has one built in, that continually gathers entropy from things like user input, system timers, I/O timing and the like. Additionally, most modern CPUs have a built in hardware RNG. In fact, even the CPU on a Raspberry Pi one.
The issue is that Excel uses a crappy old PRNG that is known to be quite bad. Presumably this is done for compatibility reasons. This RNG does not produce an even spread of numbers over the entire range, and with 15 minutes of testing someone could easily come up with a way to massively increase the probability that the candidate of their choice is chosen.
It seems like a good idea to make fraud a little bit harder than simply gaming the Excel RNG.
Amazon is not a person. Amazon is a huge corporation.
German law recognises that large corporations have an enormous amount of power, and are prone to acting like sociopaths if unchecked. It also expects them to contribute to the betterment of society, in exchange for all the benefits it recieves.
Not destroying perfectly good stuff at a cost to the environment, mostly externalised away from Amazon, is no infringement of personal liberties because Amazon is not a person.
I thought that's where we went, down town by the British embassy and north of that around some big shopping centres and the Tesla showroom. Then along by the river too. I'm not very familiar with the place.
Slashdot Mirror
Trump talked tough on NK, which Kim took advantage of to demonstrate that it was all just bluster and ultimately there was little the US could do because of China and the fact that NK had a viable nuclear deterrent.
Kim was able to use that as a bargaining chip when opening a dialogue with South Korea and the US. De-nuclearization suits NK, because it will include US nuclear weapons in the area too.
Kim quickly made concessions and progress with SK, which put pressure on Trump to not waste the opportunity. So Trump went to that meeting ready to make big concessions - end the wargames, de-nuclearize, and start opening up NK's economy with sanctions relief. Kim has boosted himself to the level of international statesman, praised by the POTUS and with a path towards a post-dictatorship life that isn't rotting in jail or hanging from a noose.
Even if it's for the wrong reasons, Trump's decision is the right one.
It's in the US's interests to do a good trade deal with China. There are many mutual benefits. However, destroying a huge company like ZTE and putting hundreds of thousands of Chinese out of work is unlikely to help secure such a deal.
The crime is breaking sanctions. Sanctions that Trump may have already decided to end (it's not clear ATM what was agreed at the Kim/Trump meeting). The punishment is arbitrary, it's not calculated to offset some loss or restore something that has been taken. And cancelling it probably nets the US a nice chunk of cash, because if ZTE simply died it wouldn't be paying that fine that it is now committed to clearing.
Plus it would have screwed all the US ZTE customers who would be left without support.
Hate to defend them but it seems they are using an audio fingerprint, similar to Shazam or Google's music ID system. The amount of data transmitted is very small, certainly not 100MB. Since it's just frequency/time histograms I'd estimate it to be in the tens of kilobytes range max, probably less.
Actually they already won, they just don't seem to have realized it.
I've been following these rules as they develop. The early draft did have a requirement for websites to actively filter new content for copyright infringement, based on submissions by copyright holders. That was removed ages ago, but people still seem to be panicking about it for some reason.
There is a requirement to remove infringing material after it has been reported, the same as the DMCA. The DMCA isn't great, but it's main flaw is people spamming automated DMCA notices with bots.
And in fact most web sites do pre-screen material as it is uploaded anyway, e.g. to block known illegal images because their customers don't like their feeds being filled with child pornography and ISIS beheading videos.
Also the "link tax" is long, long dead. There was a proposal to allow charging for "snippets" in search results, but it looks like it may well be DoA because where it was been tried it was a disaster. In Spain Google shut down it's news service and news sites saw a 6-30% loss of traffic. In Germany the news sites gave Google a free licence to use snippets anyway.
So all in all it's mostly just panicking about out of date, long rejected draft proposals.
I don't have that detail but I can tell you my experience as a user of machine translation for 15+ years.
Originally it only really worked on formal documents, and even then only produced something you could barely understand. The biggest issue seemed to be that it didn't understand context at all.
Google made some early improvements in making the translated text sound more natural. They also managed to fix a lot of common phrases that didn't quite fit the standard grammar model and thus didn't used to get translated properly. Apparently they did that by using the web as a resource for natural language and by allowing users to submit corrections.
Then AI started to be used. Baidu were the first I think and their Chinese/English translation was a huge improvement over everything else. It seemed to work slightly better going from English to Chinese though, and when Google released their AI updates not long after Chinese to English became nearly perfect.
It's actually incredible how good it is now. Often the resulting translation is not only accurate and seemingly context aware, it sounds like something a person might actually say. You don't have to think about what you are writing either. Before you had to be careful to phrase things so that the software could understand it, but not any more.
There are still some issues, like the way Japanese newspaper headlines often get translated as if it was a person speaking about their own experience (e.g. some houses were flooded, but the translation is "my house was flooded" because the software assumes that context), but for conversations between two people it's like Star Trek or something.
Fossil car manufacturers are already panicking, especially part manufacturers. For example, most Japanese manufacturers bet on hybrid tech and are now realizing that they don't have the experience and patents for EV drivetrains. Companies that have been making gearboxes for 80 years are suddenly finding that the Chinese have already cornered the market that is their future.
Same in Europe. European manufacturers are having to go to China for EV tech. Really only Nissan and Renault have anything significant of their own.
All the battery tech is Japanese, Korean and Chinese too.
I'm using them for data and basic 5V/900mA power, not the higher rated power delivery stuff.
For the high power stuff you need to be able to handle the higher voltage and to negotiate it with the host. I don't actually have anything that can deliver higher power anyway, just bog standard chargers up to 5V/3A.
I wonder if it's because of fraud. Rather than risk trying to buy stuff with a stolen credit card, buy crypto currency.
Let's not let the trolls distract from how important this story is. Prostate cancer is one of the biggest causes of death for men, and like most cancer it's much easier to treat if caught early. Encouraging guys to self-check regularly is really important, but this test should be able to detect it even earlier.
Fortunately if caught early there are effective treatments. Check yourself once a month and don't hesitate to see your doctor if you find something.
Yeah. I was using micro USB because there were no USB-C connectors that I could hand solder. Now there are some that are mixed surface mount and through-hole with side tabs for mechanical strength, so I use those.
The most annoying part is that to actually test your device for full USB 2.0 compliance you need some expensive and hard to find equipment. The test suite only officially works with a bunch of ancient powered hubs, and you need several of them because the test suite needs to check compatibility with daisy chained controllers. Oh, and you need an Intel USB 2.0 chipset on your PC because they have a special compliance testing driver.
Things got a bit better with USB 3.0 in that you can just buy a compliance testing rig. It's very expensive but at least it's simple and available.
Well, the other problem is that to fully support the USB spec and pass the test suite your firmware needs to be quite complex and implement a bunch of rarely used stuff. Most manufacturers supply a stack that passes the suite, but it can be too large to fit into small bootloaders and the like, and of course you still need to get the electrical side right.
The use USB 1.1 speeds, aka low speed (1.5Mb) or full speed (12Mb). However, you are supposed to certify them against the latest version of the spec, so most peripherals around today are USB 2.0. They only use the lower speed but they use the USB 2.0 protocol and should pass the USBIF 2.0 test suite.
Maybe you forgot what a mess USB 1 was... Keyboards that didn't work in the BIOS, for example. It wasn't until USB 1.1 that it got half way good, and it only really took off with USB 2.0 which finally offered enough speed for things like external drives and scanners.
USB 3.0 is the protocol. It supports various types of cable. If you want higher speeds you need a SuperSpeed cable and port (coloured blue). USB-C is just a connector, it can work with USB 1.1 through to 3.1, by itself it doesn't give you any extra functionality.
It's like SCSI back in the day. You had SCSI, SCSI 2 and SCSI 3 which were the protocols. SCSI 2 introduced support for removable drives and peripherals like scanners, for example. But then you also had various different physical layers. SCSI, Wide SCSI, Ultra Wide SCSI, fibre channel, LVDS, LVDS single ended, 25 pin, 50 pin, 68 pin, 80 pin...
I agree they need to communicate this better. They have always been terrible at doing that.
Your Tesla has a huge battery that is actively cooled by the car during charging. The main limitation on phone charging speeds is heat generated during charging. If the phone charged any faster the battery would be damaged by heat, possibly failing catastrophically (expansion, fire).
There isn't really a good solution to this, unless you want a much larger phone that can dissipate more heat, maybe with a little fan or some water cooling built in.
No, it's due to the way USB 3 delivers up to 100W and cheap cable manufacturers not properly testing that.
With USB 3 the device can negotiate for high power delivery, which involves increasing the supplied voltage from 5V up to 20V. Due to physics increasing the voltage reduces the current needed to deliver 100W, which in turn reduces the amount of heat generated in the cable. Heat is wasted energy.
The problem is that the cheap cables don't implement the spec properly and are not rated for 20V/5A, so can end up supplying 20V on the wrong pins and damaging equipment. The equipment needed to properly test USB 3 power delivery costs thousands of Euro/USD, so some companies just skipped it.
USB is well beyond second system.
The original USB spec has a lot of features that are rarely used these days. Alternative functions, alternative ways of sending data (e.g. HID reports over the control endpoint) that no-one uses. It was needlessly complex.
Then it got streamlined a bit, but also additional hacks were introduced such as the keyboard boot protocol that allowed USB keyboards to work in the BIOS. USB 2.0 increased the data rate, and battery charging was added but proved to be too slow so several proprietary protocols were introduced.
Then USB 3 came along and added extra pins to support higher speeds and higher power delivery. So by now it's the 4th generation at least (USB 1.0, 1.1, 2.0, 3.0, and now 3.1). And it actually works pretty well.
I think they could have done more for phone charging and for external batteries. Support for delivering power at say 3.7V (nominal LiPo voltage), to avoid the need for a step-up converter when using an external battery, would have been useful. At the moment a 5000mAh battery will only deliver about 3000mAh to the phone after conversion losses.
The final spec was actually the same as the preliminary one, the issue is that some companies just didn't bother to build their stuff properly. The test equipment and certification tools cost money, and they decided they could just bodge it for the â1 cables.
Having said that I buy decent but not expensive cables and have had no issues with USB 3. It works great for power delivery and data. I've started using it in my personal electronics projects (in USB 2.0 mode) as well.
The problem with all these little incremental improvements is that they just make the driver more and more reliant on features that are not reliable.
This could be as little as adding the ability to read road signs, something that the old AP1 hardware could do but which the newer AP2 hardware has lacked for years.
It's probably just Tesla finally enabling all the cameras that they have installed in the car. At the moment only half of them are actually in use.
To get to fully autonomy they need to add the ability to read road signs, the ability to differentiate different kinds of vehicles (cars/bikes), and some kind of 3D vision (probably comparing consecutive video frames since they don't have stereo cameras), heavy redundancy so the car can act safely when cameras get splashed with rain or mud. They they need to take that data and build a 3D model of the world, and figure out how to navigate around it.
Keep in mind that they have been selling full self driving for a couple of years now, and promised that it would drive you to work and then go off and park itself, more than Google's system can do and using a vastly inferior sensor suite.
Implementing a proper RNG isn't expensive or time consuming at all. Windows has one built in, that continually gathers entropy from things like user input, system timers, I/O timing and the like. Additionally, most modern CPUs have a built in hardware RNG. In fact, even the CPU on a Raspberry Pi one.
The issue is that Excel uses a crappy old PRNG that is known to be quite bad. Presumably this is done for compatibility reasons. This RNG does not produce an even spread of numbers over the entire range, and with 15 minutes of testing someone could easily come up with a way to massively increase the probability that the candidate of their choice is chosen.
It seems like a good idea to make fraud a little bit harder than simply gaming the Excel RNG.
Amazon is not a person. Amazon is a huge corporation.
German law recognises that large corporations have an enormous amount of power, and are prone to acting like sociopaths if unchecked. It also expects them to contribute to the betterment of society, in exchange for all the benefits it recieves.
Not destroying perfectly good stuff at a cost to the environment, mostly externalised away from Amazon, is no infringement of personal liberties because Amazon is not a person.
So in countries with worse consumer laws what happens to returned items? How is it better than the EU?
Damn it, you made me google that.
I thought that's where we went, down town by the British embassy and north of that around some big shopping centres and the Tesla showroom. Then along by the river too. I'm not very familiar with the place.