We need to encourage a future in which the vast amount of charging occurs at home. Slow charges are much better for the batteries and the impact on the grid is less if charging is spread out, coordinated with home systems, and occurs largely overnight. The positive impact to the grid is also maximized in a future where these vehicles are connected to the grid when parked for long periods of time and can be utilized to stabilize it.
Of these three initiatives, New Jersey's sounds best due to more prominent support for large numbers of home installations.
Governments do have a place in this. The place is to clear the regulatory hurdles to installations.
For example, most states will need a law clearing HOA hurdles, especially for those living in condominium and townhome situations who need to be given the right to run electric to their parking spaces without getting HOA permission which is almost always impossible, even if the HOA boards wish to do so. Their hands are usually tied by bylaws that can't be altered without impossible to attain vote percentages. Our HOA has had trouble even electing board members to abandoned spots because they can't gather a quorum. Homeowners are just completely apathetic to the process.
Another is to ensure that new major household systems such as ovens, water heaters, and air conditioners include some communication protocols to help manage household load. This can allow vehicle charging to work around A/C and water heater usage and keep peak draws no higher than they currently are. Since the A/C and water heaters could work around each other as well, it could even lower peak draws versus today. Regulating production of household systems to ensure a healthy grid is nothing new. Without existing regulations, the grid would have collapsed long ago.
The leading edge is moving away from this. Training is perhaps the hottest area of AI research. I just read an article about one that could learn to play a game by watching videos of the games being played. There are robotics systems that can train from having an action demonstrated. And there is research into methods of training that reduce the dataset sizes that has produced better networks.
I don't think you're entirely off on "innate knowledge". I think the training will become easier as we move away from the idea of training stand-alone networks to perform well-defined tasks, start connecting many trained networks together, and then train the whole in new skills. The system must build on knowledge, not relearn everything every time. Networks need to be highly interconnected "subroutines" though that is a limiting analogy because they must self-activate when needed as opposed to being invoked.
Absolutely agree that your scenario is most likely. I just haven't given up hope that tech can be a force that enables and frees us instead of one that enslaves us.
The reason the cloud has been able to take over is because the PC stopped developing technologies that serve the user and require more real-time bandwidth than is available on a home internet connection.
I want two big things to happen in PC development.
First, bring the peripherals into this century. Free it from fixed displays and support mixed voice, gesture and keyboard input in everything. Most importantly, I want wireless A/R based displays that allow me to see many virtual displays, sheets of paper, talking avatars, or whatever I want to use to represent data naturally appearing in the environment around me. Blow up the limited "desktop" metaphor and expand it to my whole world. I want to be able to create display walls like those that would be in an advanced military command center and have everything updating in real time.
Second, give it native AI capabilities that can operate at the level of Deepmind's best and be trained at home. This undoubtedly means giving it a separate AI processor that is not just a tweaked GPU and probably operates in the analog domain to pack the power of hundreds of Google TPUs into a single chip. It should be able to play games, screen callers not on my contact list, screen people at the door (announcing them and letting them in if necessary), continuously monitor news for my interests, help me to automate my stock trading activities, help me to write my software, monitor the baby room, monitor my vitals and adjust the environment for my comfort at all times, order my groceries, suggest meals based on its knowledge of what I like and my mood, and many other things in a 24x7 simultaneous fashion.
That being said, is there an indie/homegrown market for home automation?
Most smart devices can function without phoning home though updates would still require it.
There are several open source central servers in development that can run at home and control these. Here is an article that reviews some.
However, all of these seem to be requiring the users to memorize special control phrases. There doesn't seem to be any effort to create a deep neural network based open source assistant that recognizes users and understands context. Without this, it is really not a smart device. It is based on keyword recognition, not natural language processing.
The whole point of this is to keep the data at home and cut out the vendor. It doesn't work without open source AI including automated machine learning.
The open source community needs to come together and create an AI system that competes with the best and learns from its everyday users and the data they feed it without the help of an expert. That is the only way we can both enjoy the advances that AI assistants can bring to our lives and free our data from the vendors.
Of course, even if that happens, it is an open question as to whether we will find a hardware vendor willing to support home-based AI. It is in direct conflict with their lucrative data center business.
That probably means that we need a disruptive development either in AI software or hardware to achieve the coup - something from a startup (that chooses not to be bought) that enables AIs in the same class as Deepmind's best to be trained (not just executed) at home in a continuous real-time fashion.
Most of the problems with this law would be corrected if you ban or severely limit profit off of it at all levels from clinic to manufacturer. Require that it be provided at cost of production, shipping, and administration of the therapy. Do not allow any fraction of R&D costs to be billed, and do not allow clinical profits that would cause clinics to push it.
Drug companies would still support it for the sake of their development process.
weed killer != pesticide. Weed killer is designed to kill plants with a preference hopefully for weeds. Pesticide is designed to kill bugs. The dominant weed killers in the city are concrete and asphalt.
At one point we had the delusion that weed killers don't have an effect on humans. We've since found that our bodies do often have responses. For example, those based on plant estrogens have caused early puberty in some females - as early as toddler ages in cases where they got really stupid and put plant estrogens in a shampoo.
I worked several years at a software company whose 180 or so engineers mostly worked from home. The main office contained a number of high tech conference rooms of varying sizes with the largest one able to hold around 50 people at a time. There were also a few small offices available for those that just needed to get away from home for a few days to concentrate on something. Most teams would meet about once a week in one of the conference rooms.
The conference rooms all had WiFi projectors that we could share from our laptops, high-end teleconference systems, smartboards, and a few computers.
This setup allowed us to have the best of both worlds. We worked about 80-90% at home and the rest at the office. We only drove about one day a week so we saved hugely on vehicle wear and tear also. The office was much smaller than would have been necessary if everyone worked there, so the company saved a lot of overhead which was generously reflected in our salaries and in the overhead rates we charged to the customer.
Co-working spaces just take this concept and make a separate business out of it.
The idea that these are "papers" is exactly what I think we should rethink. Initially, yes, they followed that model. But, the model of what they store and how they connect into our lives is changing. We are already connecting devices into our nervous system. We could never do that with papers and gain any usefulness from it. They are becoming personal augmentations. We need to realize that even the human brain is just a device and expand to allow both device upgrades within the body and some device peripherals outside to fall under the 5th's protections as part of "self".
Tech reporting has really, really gone downhill. And Slashdot editors chose to quote it?
57% better than [615ppi]
Let's see. That would be 615 * (1 + 0.57) = 966 ppi, not 1443 ppi.
1443 ppi is (1443 / 615) - 1 = 135% better than 615 ppi. Apparently this author thinks that because 615 is about 43% of 1443 that the 1443 ppi display is 57% better. Wow. Percentages are taught in 5th grade math.
there's already one display that's better than anything on offer, and that's your own vision
WTF? The human eye is a display? Does the author maybe think that some people really can read other people's eyes? Like you look in there and there is a tiny little screen?
The approach used isn't to have the person think about it and read what they are thinking about. You tell them that you're seeking the password but you don't ask them to think about it. Then you show letters and look for involuntary recognition responses (or even looking away responses) when you show them symbols.
An fMRI is actually overkill. It's better to start with precision eye tracking, one of the new thought "microphones" that read the weak involuntary nerve signals going to your speech muscles as you think (this is not subvocal and works without any attempt to actually say a word), and perhaps similar sensors on the nerves controlling the fingers. Then show fields of symbols with a blank up above them and see what happens. If you want to scare the heck out of them first, you do a "tune up" by asking them to think about their name and gradually fill in the letters making sure to miss a couple just for show.
Most have a target mileage. My target is about 180K miles these days since reliability has increased. After that point, I'm going to consider mine worn out and get a newer one. Someone else will buy mine and drive it till it is worn out by their standard. My mom's target is about 100K miles.
All I was saying was that people who were initially 100K away from their target mileage and thought it silly to make a reservation are getting there by now.
The question is whether there is evidence needed to prosecute the crime whose investigation justified the taking of the phone. If the crime was successfully prosecuted without it, then there was no evidence on the phone needed to prosecute the crime. I'd bet the majority of the phones they've complained about not being able to open were taken in an investigation that resulted in a conviction without the phone's data.
People will drop off the list. Others who didn't want to put money down that far in advance, have since had a car wear out, or since become more successful will gladly throw in. As the car becomes more available, moves out of the beta phase it has obviously been in, people see them, people ride in ones their friends have, etc., reservation numbers will increase, not decrease.
I believe it is possible with current technology to "read a mind" - perhaps not reliably, but enough of the time to make it useful. This would involve something like placing an individual in an fMRI machine, projecting images of people, things, symbols, etc. and observing the mind's reaction to them. I bet you could involuntarily extract a password this way - one symbol at a time. The opening of whatever vault the password protects would be your proof of its correctness so you wouldn't need to worry about misinterpretation.
However, I think it should be obvious that this would represent a violation of a person's 5th amendment rights.
A future scenario we need to start preparing for though is accessing an implanted memory device other than the individual's "brain". We are already interfacing chips to brains. I'd be surprised if some of those devices don't have memories, though perhaps they are all still external to the person with the interface chip. Regardless of whether they are internal or external, I believe those memories contained in a personal extension are also deserving of 5th amendment protection. You shouldn't be able to access my pacemaker to see if I had an elevated heart rate during the time of a crime without my explicit consent regardless of warrant.
If we don't take this route of protecting personal electronic memories by the 5th amendment, a day will come when the 5th amendment is worthless.
If we do protect them, we need to consider that, initially, implanted personal augmentations are going to be more available to the rich than the poor. Those that don't have the money will "continue" to augment their capabilities using external devices. They should not have lesser rights just because their augmentation is external.
I say "continue" because that is exactly what my smartphone is to me today. It is a personal augmentation. I have an atrocious memory. Instead of trying to keep my calendar, appointments, reminders, personal communications, etc. in my head, they are in my phone's store which in many cases is extended to the cloud.
Regardless of where those memories physically reside, they are my memories and nowhere near as "readable" as a piece of paper in a filing cabinet. In fact, the tech necessary to read and access the memories from the chips is much closer to that of the tech necessary to read my biological memory without my permission than the tech necessary to read a piece of paper.
In short, I believe the law has erred in comparing smartphone memories to filing cabinets to find precedent. They should have compared them to the memory in our brains and considered their contents to be under 5th amendment protection. They should not be legally accessible, much less admissible, without my permission - even if unencrypted - unless I say so, not some judge. We need to do some backtracking and fix it now or face a future where users of augmentation tech - eventually everyone - give up their 5th amendment rights.
An inability to access the phone means nothing if prosecution was successful for other reasons. A more useful statistic would be how many phones do they have that couldn't be opened that were evidence in crimes that have not been successfully prosecuted. But, that is probably far, far beyond their math skills.
Slightly in the editor's defense, Reuters did use "dark side" in their title translation.
However, I just checked and many other sources got it right. The editor should have corrected this by finding a better source article. Shame on Slashdot. Double shame on Reuters - they really, really should know better.
Out of curiosity, I just did a quick survey by searching Google news for both "china dark side of the moon" and "china far side of the moon". It appears both that "far side" does win and that most that said "dark side" are just repeats of the Reuters mistake. The world is not as dumb as I feared.
The highlight of my Digital Electronics II Lab course in college was to design and prototype a PDP5 using 74xx series logic chips on breadboards. In comparison, the PDP 11/70 was an advanced supercomputer, but the task of designing even the simplest computer at the gate level really created an appreciation for the complexity of processors. It took 20 breadboards to prototype and worked for just a few minutes before a chip lost its smoke somewhere. Fun days!
I'm not talking about a fictional person at all. Rather, I'd like the concept of a person to be extended to include their augmentations acting as their proxies. In other words, AIs acting to fill the express requests of a person are that person.
Complaints about AIs "recording" things will come back to haunt us. When the day comes that someone like Larry Ellis is uploaded into a machine, will it always have to announce that it is recording? What about when we finally get the ability to augment our memory with electronic implants?
We need to figure out how to embrace this into our system now or face discrimination later. We need to clear the path for augmenting ourselves so that we can keep up.
One approach that could satisfy all is to just blanket label all recordings made without permission as hearsay - i.e. treat one form of a person's memory (tech enabled recordings) as no more admissable than the other (unaltered biological memory enabled recordings). Removing the requirement that "person" be limited to a person's natural physical body or that "home" be limited to within the physical walls of a home would help a lot of legal situations that have evolved that our founders could never have thought of. We just need an interpretation expansion to match how technology is expanding us.
I was thinking about Google Search too, but I was considering the early PC farm version running fully in RAM on a bunch of COTS machines. That required more of what I consider "sophistication" relative to the general environment at the time than the modern day version. I can appreciate the headaches of getting that to happen.
Anyone who really programmed in the day knows that the original Plug and Play effort that was almost universally hated for the cases it didn't work with was unbelievably successful given the problem. Figuring out which of several thousand common cards was present when the cards had not been designed with that in mind was an impossible problem. It involved testing for and documenting quirks of every card and then finding the right set of tests and pattern of performing them to sense those quirks without actually causing things to happen on whatever card might happen to be there. In some ways it was similar to a travelling salesman problem with 5,000 locations and roads with landmines that trigger if the right pattern wasn't followed. The audacity of even trying was as remarkable as the fact that they were as successful as they were.
The crux of this question is the interpretation of the phrase "most sophisticated". I feel it has a density of complexity component. So I'd lean towards candidates that must perform complex tasks under difficult constraints, either physical or virtual.
It actually makes me think back to bygone days of trying to cram complex tasks into 8-bit embedded controllers.
A representative case that comes to mind was a function in an armament controller - the computer that controlled dropping dumb bombs from a fighter travelling 500+ miles an hour. The processor was an Intel 8080 in the days when 64K was a lot of memory. The specification we had to hit was for the bombs to hit the ground at the spacing dialed in, typically 50 or 100 ft, with +/- 1 foot of accuracy. We were not allowed to require that the plane be flying level or at constant speed. Also, when you drop a 2000 pound bomb off a wingtip, the plane lurches in the opposite direction so that the next one dropping from the other side a second later has an additional acceleration. These and other factors required that we be able to perform a multivariable integration problem in real time on an 8 bit processor running something like 5MHz with no floating point capability. It took a lot of thought, creativity, and simulation. Carefully constructed tables were used to speed the integration and a tremendous amount of trial and error to make it always converge. All of the code was in assembly language though it had been prototyped in a slightly higher level language that is likely long dead. But, the specification was met. That software was sophisticated. I've worked multi-million line projects since that didn't begin to approach the art that went into those KBs of assembly.
Other examples I'd think of are in the device logic arena, which I also consider software. Getting PLAs to perform more sophisticated operations often involved dispensing with synchronous logic and working in the asynchronous realm. Getting that to be stable across devices with gate speed variations could be pretty tricky, but the end result of having functions performed at throughput levels that others considered impossible at the time was oh so satisfying.
I can understand the anonymous reader's thoughts of the complexity of the worm. It has constraints that fall in that "virtual" arena. It must do its job without being detected until it is too late or having a signature that indisputably reveals its creator. That is very challenging. The task of creating software like that is more art than science. Requiring "art" is also a very necessary component of "most sophisticated" IMO.
Slashdot Mirror
We need to encourage a future in which the vast amount of charging occurs at home. Slow charges are much better for the batteries and the impact on the grid is less if charging is spread out, coordinated with home systems, and occurs largely overnight. The positive impact to the grid is also maximized in a future where these vehicles are connected to the grid when parked for long periods of time and can be utilized to stabilize it.
Of these three initiatives, New Jersey's sounds best due to more prominent support for large numbers of home installations.
Governments do have a place in this. The place is to clear the regulatory hurdles to installations.
For example, most states will need a law clearing HOA hurdles, especially for those living in condominium and townhome situations who need to be given the right to run electric to their parking spaces without getting HOA permission which is almost always impossible, even if the HOA boards wish to do so. Their hands are usually tied by bylaws that can't be altered without impossible to attain vote percentages. Our HOA has had trouble even electing board members to abandoned spots because they can't gather a quorum. Homeowners are just completely apathetic to the process.
Another is to ensure that new major household systems such as ovens, water heaters, and air conditioners include some communication protocols to help manage household load. This can allow vehicle charging to work around A/C and water heater usage and keep peak draws no higher than they currently are. Since the A/C and water heaters could work around each other as well, it could even lower peak draws versus today. Regulating production of household systems to ensure a healthy grid is nothing new. Without existing regulations, the grid would have collapsed long ago.
The leading edge is moving away from this. Training is perhaps the hottest area of AI research. I just read an article about one that could learn to play a game by watching videos of the games being played. There are robotics systems that can train from having an action demonstrated. And there is research into methods of training that reduce the dataset sizes that has produced better networks.
I don't think you're entirely off on "innate knowledge". I think the training will become easier as we move away from the idea of training stand-alone networks to perform well-defined tasks, start connecting many trained networks together, and then train the whole in new skills. The system must build on knowledge, not relearn everything every time. Networks need to be highly interconnected "subroutines" though that is a limiting analogy because they must self-activate when needed as opposed to being invoked.
Absolutely agree that your scenario is most likely. I just haven't given up hope that tech can be a force that enables and frees us instead of one that enslaves us.
The reason the cloud has been able to take over is because the PC stopped developing technologies that serve the user and require more real-time bandwidth than is available on a home internet connection.
I want two big things to happen in PC development.
First, bring the peripherals into this century. Free it from fixed displays and support mixed voice, gesture and keyboard input in everything. Most importantly, I want wireless A/R based displays that allow me to see many virtual displays, sheets of paper, talking avatars, or whatever I want to use to represent data naturally appearing in the environment around me. Blow up the limited "desktop" metaphor and expand it to my whole world. I want to be able to create display walls like those that would be in an advanced military command center and have everything updating in real time.
Second, give it native AI capabilities that can operate at the level of Deepmind's best and be trained at home. This undoubtedly means giving it a separate AI processor that is not just a tweaked GPU and probably operates in the analog domain to pack the power of hundreds of Google TPUs into a single chip. It should be able to play games, screen callers not on my contact list, screen people at the door (announcing them and letting them in if necessary), continuously monitor news for my interests, help me to automate my stock trading activities, help me to write my software, monitor the baby room, monitor my vitals and adjust the environment for my comfort at all times, order my groceries, suggest meals based on its knowledge of what I like and my mood, and many other things in a 24x7 simultaneous fashion.
That being said, is there an indie/homegrown market for home automation?
Most smart devices can function without phoning home though updates would still require it.
There are several open source central servers in development that can run at home and control these. Here is an article that reviews some.
However, all of these seem to be requiring the users to memorize special control phrases. There doesn't seem to be any effort to create a deep neural network based open source assistant that recognizes users and understands context. Without this, it is really not a smart device. It is based on keyword recognition, not natural language processing.
The whole point of this is to keep the data at home and cut out the vendor. It doesn't work without open source AI including automated machine learning.
The open source community needs to come together and create an AI system that competes with the best and learns from its everyday users and the data they feed it without the help of an expert. That is the only way we can both enjoy the advances that AI assistants can bring to our lives and free our data from the vendors.
Of course, even if that happens, it is an open question as to whether we will find a hardware vendor willing to support home-based AI. It is in direct conflict with their lucrative data center business.
That probably means that we need a disruptive development either in AI software or hardware to achieve the coup - something from a startup (that chooses not to be bought) that enables AIs in the same class as Deepmind's best to be trained (not just executed) at home in a continuous real-time fashion.
Most of the problems with this law would be corrected if you ban or severely limit profit off of it at all levels from clinic to manufacturer. Require that it be provided at cost of production, shipping, and administration of the therapy. Do not allow any fraction of R&D costs to be billed, and do not allow clinical profits that would cause clinics to push it.
Drug companies would still support it for the sake of their development process.
weed killer != pesticide. Weed killer is designed to kill plants with a preference hopefully for weeds. Pesticide is designed to kill bugs. The dominant weed killers in the city are concrete and asphalt.
At one point we had the delusion that weed killers don't have an effect on humans. We've since found that our bodies do often have responses. For example, those based on plant estrogens have caused early puberty in some females - as early as toddler ages in cases where they got really stupid and put plant estrogens in a shampoo.
I worked several years at a software company whose 180 or so engineers mostly worked from home. The main office contained a number of high tech conference rooms of varying sizes with the largest one able to hold around 50 people at a time. There were also a few small offices available for those that just needed to get away from home for a few days to concentrate on something. Most teams would meet about once a week in one of the conference rooms.
The conference rooms all had WiFi projectors that we could share from our laptops, high-end teleconference systems, smartboards, and a few computers.
This setup allowed us to have the best of both worlds. We worked about 80-90% at home and the rest at the office. We only drove about one day a week so we saved hugely on vehicle wear and tear also. The office was much smaller than would have been necessary if everyone worked there, so the company saved a lot of overhead which was generously reflected in our salaries and in the overhead rates we charged to the customer.
Co-working spaces just take this concept and make a separate business out of it.
The idea that these are "papers" is exactly what I think we should rethink. Initially, yes, they followed that model. But, the model of what they store and how they connect into our lives is changing. We are already connecting devices into our nervous system. We could never do that with papers and gain any usefulness from it. They are becoming personal augmentations. We need to realize that even the human brain is just a device and expand to allow both device upgrades within the body and some device peripherals outside to fall under the 5th's protections as part of "self".
Tech reporting has really, really gone downhill. And Slashdot editors chose to quote it?
57% better than [615ppi]
Let's see. That would be 615 * (1 + 0.57) = 966 ppi, not 1443 ppi.
1443 ppi is (1443 / 615) - 1 = 135% better than 615 ppi. Apparently this author thinks that because 615 is about 43% of 1443 that the 1443 ppi display is 57% better. Wow. Percentages are taught in 5th grade math.
there's already one display that's better than anything on offer, and that's your own vision
WTF? The human eye is a display? Does the author maybe think that some people really can read other people's eyes? Like you look in there and there is a tiny little screen?
The approach used isn't to have the person think about it and read what they are thinking about. You tell them that you're seeking the password but you don't ask them to think about it. Then you show letters and look for involuntary recognition responses (or even looking away responses) when you show them symbols.
An fMRI is actually overkill. It's better to start with precision eye tracking, one of the new thought "microphones" that read the weak involuntary nerve signals going to your speech muscles as you think (this is not subvocal and works without any attempt to actually say a word), and perhaps similar sensors on the nerves controlling the fingers. Then show fields of symbols with a blank up above them and see what happens. If you want to scare the heck out of them first, you do a "tune up" by asking them to think about their name and gradually fill in the letters making sure to miss a couple just for show.
Most have a target mileage. My target is about 180K miles these days since reliability has increased. After that point, I'm going to consider mine worn out and get a newer one. Someone else will buy mine and drive it till it is worn out by their standard. My mom's target is about 100K miles.
All I was saying was that people who were initially 100K away from their target mileage and thought it silly to make a reservation are getting there by now.
The question is whether there is evidence needed to prosecute the crime whose investigation justified the taking of the phone. If the crime was successfully prosecuted without it, then there was no evidence on the phone needed to prosecute the crime. I'd bet the majority of the phones they've complained about not being able to open were taken in an investigation that resulted in a conviction without the phone's data.
People will drop off the list. Others who didn't want to put money down that far in advance, have since had a car wear out, or since become more successful will gladly throw in. As the car becomes more available, moves out of the beta phase it has obviously been in, people see them, people ride in ones their friends have, etc., reservation numbers will increase, not decrease.
I believe it is possible with current technology to "read a mind" - perhaps not reliably, but enough of the time to make it useful. This would involve something like placing an individual in an fMRI machine, projecting images of people, things, symbols, etc. and observing the mind's reaction to them. I bet you could involuntarily extract a password this way - one symbol at a time. The opening of whatever vault the password protects would be your proof of its correctness so you wouldn't need to worry about misinterpretation.
However, I think it should be obvious that this would represent a violation of a person's 5th amendment rights.
A future scenario we need to start preparing for though is accessing an implanted memory device other than the individual's "brain". We are already interfacing chips to brains. I'd be surprised if some of those devices don't have memories, though perhaps they are all still external to the person with the interface chip. Regardless of whether they are internal or external, I believe those memories contained in a personal extension are also deserving of 5th amendment protection. You shouldn't be able to access my pacemaker to see if I had an elevated heart rate during the time of a crime without my explicit consent regardless of warrant.
If we don't take this route of protecting personal electronic memories by the 5th amendment, a day will come when the 5th amendment is worthless.
If we do protect them, we need to consider that, initially, implanted personal augmentations are going to be more available to the rich than the poor. Those that don't have the money will "continue" to augment their capabilities using external devices. They should not have lesser rights just because their augmentation is external.
I say "continue" because that is exactly what my smartphone is to me today. It is a personal augmentation. I have an atrocious memory. Instead of trying to keep my calendar, appointments, reminders, personal communications, etc. in my head, they are in my phone's store which in many cases is extended to the cloud.
Regardless of where those memories physically reside, they are my memories and nowhere near as "readable" as a piece of paper in a filing cabinet. In fact, the tech necessary to read and access the memories from the chips is much closer to that of the tech necessary to read my biological memory without my permission than the tech necessary to read a piece of paper.
In short, I believe the law has erred in comparing smartphone memories to filing cabinets to find precedent. They should have compared them to the memory in our brains and considered their contents to be under 5th amendment protection. They should not be legally accessible, much less admissible, without my permission - even if unencrypted - unless I say so, not some judge. We need to do some backtracking and fix it now or face a future where users of augmentation tech - eventually everyone - give up their 5th amendment rights.
An inability to access the phone means nothing if prosecution was successful for other reasons. A more useful statistic would be how many phones do they have that couldn't be opened that were evidence in crimes that have not been successfully prosecuted. But, that is probably far, far beyond their math skills.
Good journalism seeks to communicate clearly and effectively. Choosing "dark" instead of "far" in this case does not achieve that.
Slightly in the editor's defense, Reuters did use "dark side" in their title translation.
However, I just checked and many other sources got it right. The editor should have corrected this by finding a better source article. Shame on Slashdot. Double shame on Reuters - they really, really should know better.
Out of curiosity, I just did a quick survey by searching Google news for both "china dark side of the moon" and "china far side of the moon". It appears both that "far side" does win and that most that said "dark side" are just repeats of the Reuters mistake. The world is not as dumb as I feared.
The highlight of my Digital Electronics II Lab course in college was to design and prototype a PDP5 using 74xx series logic chips on breadboards. In comparison, the PDP 11/70 was an advanced supercomputer, but the task of designing even the simplest computer at the gate level really created an appreciation for the complexity of processors. It took 20 breadboards to prototype and worked for just a few minutes before a chip lost its smoke somewhere. Fun days!
I'm not talking about a fictional person at all. Rather, I'd like the concept of a person to be extended to include their augmentations acting as their proxies. In other words, AIs acting to fill the express requests of a person are that person.
Complaints about AIs "recording" things will come back to haunt us. When the day comes that someone like Larry Ellis is uploaded into a machine, will it always have to announce that it is recording? What about when we finally get the ability to augment our memory with electronic implants?
We need to figure out how to embrace this into our system now or face discrimination later. We need to clear the path for augmenting ourselves so that we can keep up.
One approach that could satisfy all is to just blanket label all recordings made without permission as hearsay - i.e. treat one form of a person's memory (tech enabled recordings) as no more admissable than the other (unaltered biological memory enabled recordings). Removing the requirement that "person" be limited to a person's natural physical body or that "home" be limited to within the physical walls of a home would help a lot of legal situations that have evolved that our founders could never have thought of. We just need an interpretation expansion to match how technology is expanding us.
I was thinking about Google Search too, but I was considering the early PC farm version running fully in RAM on a bunch of COTS machines. That required more of what I consider "sophistication" relative to the general environment at the time than the modern day version. I can appreciate the headaches of getting that to happen.
Anyone who really programmed in the day knows that the original Plug and Play effort that was almost universally hated for the cases it didn't work with was unbelievably successful given the problem. Figuring out which of several thousand common cards was present when the cards had not been designed with that in mind was an impossible problem. It involved testing for and documenting quirks of every card and then finding the right set of tests and pattern of performing them to sense those quirks without actually causing things to happen on whatever card might happen to be there. In some ways it was similar to a travelling salesman problem with 5,000 locations and roads with landmines that trigger if the right pattern wasn't followed. The audacity of even trying was as remarkable as the fact that they were as successful as they were.
The crux of this question is the interpretation of the phrase "most sophisticated". I feel it has a density of complexity component. So I'd lean towards candidates that must perform complex tasks under difficult constraints, either physical or virtual.
It actually makes me think back to bygone days of trying to cram complex tasks into 8-bit embedded controllers.
A representative case that comes to mind was a function in an armament controller - the computer that controlled dropping dumb bombs from a fighter travelling 500+ miles an hour. The processor was an Intel 8080 in the days when 64K was a lot of memory. The specification we had to hit was for the bombs to hit the ground at the spacing dialed in, typically 50 or 100 ft, with +/- 1 foot of accuracy. We were not allowed to require that the plane be flying level or at constant speed. Also, when you drop a 2000 pound bomb off a wingtip, the plane lurches in the opposite direction so that the next one dropping from the other side a second later has an additional acceleration. These and other factors required that we be able to perform a multivariable integration problem in real time on an 8 bit processor running something like 5MHz with no floating point capability. It took a lot of thought, creativity, and simulation. Carefully constructed tables were used to speed the integration and a tremendous amount of trial and error to make it always converge. All of the code was in assembly language though it had been prototyped in a slightly higher level language that is likely long dead. But, the specification was met. That software was sophisticated. I've worked multi-million line projects since that didn't begin to approach the art that went into those KBs of assembly.
Other examples I'd think of are in the device logic arena, which I also consider software. Getting PLAs to perform more sophisticated operations often involved dispensing with synchronous logic and working in the asynchronous realm. Getting that to be stable across devices with gate speed variations could be pretty tricky, but the end result of having functions performed at throughput levels that others considered impossible at the time was oh so satisfying.
I can understand the anonymous reader's thoughts of the complexity of the worm. It has constraints that fall in that "virtual" arena. It must do its job without being detected until it is too late or having a signature that indisputably reveals its creator. That is very challenging. The task of creating software like that is more art than science. Requiring "art" is also a very necessary component of "most sophisticated" IMO.