There will never be one that works well until companies stop rolling their own and use a standard. The most important aspect of working well is not having to ask someone whether they have or can join a service.
Google seems to finally be taking the high road by supporting RCS or "chat", the only viable SMS replacement I've heard of. It would be nice to see telecom standards for video calls come about too.
We seem to have thoroughly given up personal and home-based computing. Almost every decent app today is cloud-based, and the devices in our hands are just terminals. Personal SW innovation has stalled for so long that my mother can actually run today's programs including Win10 and Office on her 2006 vintage 17" Core Duo laptop with 4GB RAM. The only upgrade I've put in it is an SSD. That's a far cry from the days when the pace of progress required a 3 y/o PC to be re-tasked as a router or something.
I'd love a return to a world where I felt justified to spend $3K on a home server and $2K on my desktop and $1K on displays every 3 years because what I had was totally obsolete.
If Google sold the ability to set up a home server, run their software locally, and then use that as the basis for all of my phone's abilities no matter where I'm at, I would be willing to put down that kind of change instead of doing everything on cloud machines. That kind of move could reignite personal computing, probably opening the door to an explosion in AI hardware diversity that can't happen with the large datacenter approach to computing.
All that said, I won't give up the capabilities I now have without a reasonably equal alternative. Yes, I know of the various open source projects trying to do things like this, have loaded some up, and found them to be less than alpha IMO.
I also realize some phones are trying to do some of this on the phone, but I find that very distasteful. It adds a lot of cost to the phone and only benefits the phone. If I'm going to pay for that kind of AI, I won't it to benefit every device in every room in our home, our PCs, our vehicles, and all of our phones.
The home server approach would balance my privacy concerns with my engineering sense of cost-effectiveness. That could probably also be cost-effectively augmented with a car-based server as the autos get more built-in AI.
Personally, the main area of growth I've seen is that I rarely have to pick up my phone and use the keypad or, much of the time, even look at the screen anymore. I can and do accomplish almost everything conversationally. I set appointments, add things to lists, have lists read to me, send messages, perform all my navigation, make all calls, play music, start and stop my runs, set timers and query timers and alarms, set reminders both at times and locations, etc. conversationally. And I never have to worry whether the people I'm interacting with have some special proprietary or non-standard app or not like I see many others doing.
You are correct but you can compare Intel to TSMC plus its customers who are doing their own design work. The article did so.
Historically, the company has squashed rivals using a research budget that dwarfed anything else in the industry. But TSMC’s approach is even undermining this advantage.
While Intel still outguns TSMC in capital spending on new plants and equipment, the tables are turned when you combine the research budgets of TSMC customers like Qualcomm, Apple, Nvidia Corp. and Huawei Technologies Co.
According to Goldman Sachs, the combined budgets of TSMC’s customers are not only larger than Intel but the gap is increasing. By 2020, they will spend almost $20 billion, according to its estimate, at least $4 billion more than Intel.
IMO, the rise to dominance of TSMC's business model is inevitable and probably being driven by the industry's fall off of the Moore's Law curve.
For decades, companies have been able to keep increasing the capabilities of their product by just buying the next-generation general-purpose chip. They got lazy in the process. I'd say this transition occurred in the '87-'97 time frame, a time when the need for engineers to design custom hardware plummeted in favor of buying COTS. But the general purpose approach is starting to fall short of the increases necessary to drive new consumer purchases.
But innovation is still possible. Our laziness has created a deep untapped well of performance growth that can be had by equipping the domains to create domain specific designs. If we can reignite domain-specific engineering, many domains can achieve order of magnitude changes in performance by rolling their own designs.
TSMC is enabling the larger of these domains to achieve purpose-built silicon designed by the domain's engineers for the domain.
Granted, 44.5% is not precisely "most", but certainly not limited to the top 10% types that can afford it. In order to have that kind of market share, there has to be a lot of people with below average incomes with iPhones.
And, as others have indicated, there is no way a tariff on smartphones would be limited to iPhones.
I feel like I've entered the twilight zone when I read an argument that storing my work on my machine is dangerous while storing it anywhere else is considered safe. Security has been compromised the moment my data isn't stored on my machine. Virtually every internet service today is a major security breach. And when someone tries to come up with something to reduce the near-requirement that all data be given up from inception, people call it a security threat? WTF? That's some pretty rich spin control.
I have Spectrum and am finding that my speeds for streaming are going up dramatically (triple or more) when I stream through a VPN. This is while always measuring the 100 MB speed I pay for or a touch more when running any speed checking service. So we have entered an age where it is useless to say you have a 100 MB connection without saying what data type you're getting that speed with.
It has only been a quarter century or so since virtually all map data came from governments. As computers reached a level of being able to pick routes and GPS became cheap, both needs and wants arose for expanded data. I can remember a few years in the 90s where applications that were technically possible were being held back for lack of that data. Many knew we needed it. It would be nice if the governments had stepped up and expanded their data for everyone's use in a timely fashion. They didn't.
The companies who had applications that could utilize the data to enough benefit to justify the billions in continuous expenditures necessary to collect it and keep it up to date did so. It would not be the least bit fair to then turn around and wipe out the benefits derived from their efforts by making the same data available to all.
A government grab of this data would be little different from a government announcing tomorrow that all entertainment in the country it oversees is now public domain and free for all to view or listen to at will. The public might cheer initially, but the industry would collapse. It would kill the goose that laid the golden egg.
At this point, it would not even be fair for the governments to step in with their own money, create an equivalent dataset including all of the imagery and analysis of that imagery, and then give it to the general public. The public wouldn't even desire the data if it weren't for the industry's risk-taking and developments that created the need for it.
Until an update made it unbootable again a few weeks ago, my Linux partition updated almost every day. Reboots were required in these updates once every couple of weeks or so. Several times a year, I'd have to spend significant time fixing a problem caused by an update. I imagine updates cost me around a week a year on average. This time, I decided to go back to my Windows partition. It just works better with my hardware - a laptop with the NVidia Optimus graphics configuration that Linux has never supported well.
I could have configured my Linux installation to be more stable by only allowing security updates, but that's not what I like as a user. I opt in to every beta I can. I've always liked being on the bleeding edge.
Windows needs to offer every user a choice in the frequency of their updates. Many are like me and want them sooner than later. I am in the preview program. Others want more stability. One size doesn't fit all.
They said they were sensitive to it, not that they like it.
I am very sensitive to bitter tastes, do not like them, avoid most bitter foods, and drink a lot of black coffee despite the fact that I don't like the bitterness. In an odd way, that is sort of why I drink it. I posted below that that sort of makes it worthwhile in a medicinal way.
My wife doesn't seem to have the ability to taste "bitter" and so would be determined to not be sensitive to bitterness. She likes a lot of foods that I consider very bitter and can't stand. She describes the taste of those foods with terms other than bitter. I can't taste those other flavors through the overwhelming bitterness and so can't appreciate what she is tasting. And, yes, she hates coffee.
I suspect these findings on coffee would hold for many alcoholic beverages that are bitter as well. There are quite a few that are considered acquired tastes that I find to be extremely bitter.
In general, I think this falls into our "no pain, no gain" prejudice. People routinely feel more like they are doing something worthwhile to solve a problem when the something they are doing is unpleasant in some fashion - be it exercise, diet, medicine, work, etc.
Caller ID "spoofing" is a major feature that many people often use. For example, Google Home uses it when you set it up to utilize your cell's number when you call out. There are also services that allow you to send and receive work calls using your work number from your personal cell.
But, behind the scenes, the real device making the call is always known. What is needed is a trivial means of letting private attorneys pursue the civil fines - something that these traffic ticket type shops can handle. You could dial their number after getting a robocall, they'd initiate an automatic trace, group it with others that trace to the same organization, and pursue the fines. An attorney would work pretty hard for a percentage of $10K x 100 calls or so. Push that to 1000 calls and they'll gladly start trying to go after the foreign ones too. Perhaps they could find American assets to grab.
I get a junk robocall almost every day and most who I know get more. 100 million a day in America is probably a conservative estimate. That would make $1500 a call a $150 billion per day fine rate if the fines were effective. Clearly they are not being utilized.
I have written an operating system and drivers in Ada that ran in a very complex embedded system that had a mix of over a hundred CPUs of different types. Elsewhere in that same system I also wrote an operating system with C. The interesting thing is that the one written in Ada was running on a CPU with 8 times the speed and 4 times the memory of the one written in C, but the one written in C outran it dramatically. Ada's performance sucks.
I've been running with just Windows Defender for years. As part of my work, I visit several hundred unique new internet sites every week. I haven't gotten a virus since the 90s. On the other hand, I have seen many serious system performance problems solved by removing antivirus software. I'd say that removing AV software is the second biggest performance increase you can have on a modern PC after switching to an SSD. Upgrading to a lower latency internet connection might beat it, but often isn't available (though I have found that using a VPN multiplied my internet throughput in many applications).
Which is just to say that they now have other ways of confirming your ID than paper. When all of the biometric techs get good enough, we'll be able to drop the paper again - but that is only because it has been replaced with something more reliable.
I think the shocking concept to me at the time was that anyone would ever need to prove their identity by any means to travel within the confines of our country. That was a totally foreign concept in the 60s and 70s.
True. At the time, I felt that the two big ones were FBI and Homeland Security. I remember the odor one caught my attention because I had just read an article on the sniffers that many were starting to have to walk through in the airports in those years. Those systems are not at all limited to explosive compounds and collect more than enough data to identify a person. Of course there are also many cameras in the airports that could be analyzed in multiple ways with gait being likely.
It was always shocking to me as a kid to read about how Nazi Germany actually required people to have papers to travel - unimaginable and a prime example of a dictatorial government run amok. Fast forward a few decades... try to get on a plane in the US without papers.
The National Information Exchange Model (NIEM) is an XML-based information exchange framework used by the US government to grease the sharing of information between departments. It was started in the Bush administration in 2005. I first noticed it around 2008 and it contained fields to define PersonGait at that time (version 2 I think).
They have deleted some of these fields since then. I suspect they did so because it gave out too much information about what they are doing though it is also possible the new administration rolled back the push into some of these areas.
The rules for getting data components added to the model include a requirement that the data components must already be in use by at least 2 different departments. So, in 2008, at least 2 departments of the government were using a person's gait as an identifying characteristic.
I have included a list of the fields in the "PersonAugmentationType" data type for the V2.0 version of the spec from a decade ago below. Many are interesting including: body odor, ear shape, finger geometry, gait, hand geometry, keystroke dynamics, lip movement, urine, vein pattern, etc.
The NIEM is publicly available and published on Github. I highly recommend downloading one of the older versions before they sanitized it, like V2.0 or so, and spending some time looking at the spreadsheets that describe it. The insight to be gained in what the government has reason to store about us is extensive.
Of course we need better pattern recognition, but I don't think we get there through larger nets and better training. I suspect we've reached a level of training and individual net size that is already adequate. It is actually very surprising that the individual nets we've created can compete as well as they can with humans because they are doing it without the feedback of thousands of other nets that the human brain has.
What is most needed is not better trained specialized nets. We need many nets trained in different things and connected in ways that allow them to correct each other either directly or indirectly through arbitration nets. The feedback from other nets with different specializations would have corrected the interpretation in the examples given because the interpretation didn't make sense.
I've long felt that what we are missing is the OS. We've figured out how to create the subroutines, but little is in place to put the many different subroutines into parallel contact with each other in a useful way.
My gut says this problem is not a big one. As we learn to connect large numbers of nets together in a manner that allows the whole community of nets to settle on a truth and describe it and act on it without any one central thing holding all of the knowledge of the truth, what we think of as understanding will emerge.
I get a lot of value out of this when using Google to search. If the search is for hard to find or describe data and I'm spending over half an hour searching and entering searches that approach the question from many angles, I definitely want to see the many links I've already visited in old searches highlighted in the new ones. I also research many subjects again and again over time (days, months, years, etc.) and would like to be able to distinguish previously unseen information.
In fact, it would be awesome to have a feature in Google search that I could flip on and off with a single click to just filter out previously seen information on the server side! Maybe it's there and I just haven't looked for it.
I'm not sure I see a direct threat from this for myself. I would think it would be used to inform phishing attacks and ad placement, but I'm not vulnerable to either.
On the other hand, the vast majority of internet users are not as informed as most tech users. They are vulnerable to attacks like this and we should be concerned about that because that vulnerability does affect the internet as a whole in ways that splash back on everyone by inciting regulation, limiting services, etc.
In a very, tenuously related theme, the feature I would most like added to Netflix is the ability to remove everything I've already watched from any suggestions as well as to allow me to tell it that I don't ever want to watch a particular video and have that removed from suggestions also. Of the shows and movies I've watched in my life, I doubt I watched more than 1% twice. I usually hate viewing or reading the same thing twice. Oddly, that doesn't carry to music. There is something fundamentally different there.
If that is the case, then it has become to specialized. I switched from computer science to computer engineering mid-program back in the 80s, but both had multiple statistics classes as well as database classes (different but related).
Data Science should be part of the fundamental base for any science or engineering curriculum. It is kind of hard to perform any science or engineering without it.
My kid (~ 3) has an Amazon Fire 8 for Kids. It has a huge selection of apps and videos. I have never seen an ad on it or any inappropriate apps or videos. It requires zero policing on my part.
What kid's device has this problem? Are parents actually letting their kids use adult devices or something?
Of course. But that requires a large space to walk around in and all new media. Down the road, sure. As a launch minimum, my minimum is to significantly improve on the cost, flexibility, capability and convenience provided by fixed flat panel display technology.
I've never felt that gaming is the right app for this, probably because I don't game but have wanted AR since first experiencing crude forms of it in the 90s when it was used to tremendously cut costs in aircraft wiring bundle production. I much prefer apps that offer real-life advantages.
But AR devices are still short on FOV, resolution, and brightness.
If AR could give me a full FOV with as many virtual displays as I want wherever I place them around the room and in whatever shapes I choose with no apparent pixelation and good brightness, I'd pay up to $1K for it. That would be less than what I paid for large displays in the early 90s (much less accounting for inflation). AR has the advantage that it doesn't have to render everything in full fidelity all of the time even within your FOV if it knows where your eye is pointed, so it should be achievable from a GPU POV. We just need much higher resolution full FOV AR displays that optimize rendering per where the eye is looking.
The killer launch application from my POV would be to give me banks of monitors at my desk in my living room without having to have banks of monitors. I'd love it for watching movies with virtual screens as large as I want too. But the images need to be solid with HDR and apparent resolution in the center of my vision similar to my 4K TV.
That's the minimum launch point. It can then evolve to being something portable that could eliminate the need for any product that just exists to be looked at or allow anyone in my home to see it as "decorated" in any way they like and changed with their mood.
Slashdot Mirror
There will never be one that works well until companies stop rolling their own and use a standard. The most important aspect of working well is not having to ask someone whether they have or can join a service.
Google seems to finally be taking the high road by supporting RCS or "chat", the only viable SMS replacement I've heard of. It would be nice to see telecom standards for video calls come about too.
I wish it did. But, beggars can't be choosers.
We seem to have thoroughly given up personal and home-based computing. Almost every decent app today is cloud-based, and the devices in our hands are just terminals. Personal SW innovation has stalled for so long that my mother can actually run today's programs including Win10 and Office on her 2006 vintage 17" Core Duo laptop with 4GB RAM. The only upgrade I've put in it is an SSD. That's a far cry from the days when the pace of progress required a 3 y/o PC to be re-tasked as a router or something.
I'd love a return to a world where I felt justified to spend $3K on a home server and $2K on my desktop and $1K on displays every 3 years because what I had was totally obsolete.
If Google sold the ability to set up a home server, run their software locally, and then use that as the basis for all of my phone's abilities no matter where I'm at, I would be willing to put down that kind of change instead of doing everything on cloud machines. That kind of move could reignite personal computing, probably opening the door to an explosion in AI hardware diversity that can't happen with the large datacenter approach to computing.
All that said, I won't give up the capabilities I now have without a reasonably equal alternative. Yes, I know of the various open source projects trying to do things like this, have loaded some up, and found them to be less than alpha IMO.
I also realize some phones are trying to do some of this on the phone, but I find that very distasteful. It adds a lot of cost to the phone and only benefits the phone. If I'm going to pay for that kind of AI, I won't it to benefit every device in every room in our home, our PCs, our vehicles, and all of our phones.
The home server approach would balance my privacy concerns with my engineering sense of cost-effectiveness. That could probably also be cost-effectively augmented with a car-based server as the autos get more built-in AI.
Personally, the main area of growth I've seen is that I rarely have to pick up my phone and use the keypad or, much of the time, even look at the screen anymore. I can and do accomplish almost everything conversationally. I set appointments, add things to lists, have lists read to me, send messages, perform all my navigation, make all calls, play music, start and stop my runs, set timers and query timers and alarms, set reminders both at times and locations, etc. conversationally. And I never have to worry whether the people I'm interacting with have some special proprietary or non-standard app or not like I see many others doing.
You are correct but you can compare Intel to TSMC plus its customers who are doing their own design work. The article did so.
Historically, the company has squashed rivals using a research budget that dwarfed anything else in the industry. But TSMC’s approach is even undermining this advantage.
While Intel still outguns TSMC in capital spending on new plants and equipment, the tables are turned when you combine the research budgets of TSMC customers like Qualcomm, Apple, Nvidia Corp. and Huawei Technologies Co.
According to Goldman Sachs, the combined budgets of TSMC’s customers are not only larger than Intel but the gap is increasing. By 2020, they will spend almost $20 billion, according to its estimate, at least $4 billion more than Intel.
IMO, the rise to dominance of TSMC's business model is inevitable and probably being driven by the industry's fall off of the Moore's Law curve.
For decades, companies have been able to keep increasing the capabilities of their product by just buying the next-generation general-purpose chip. They got lazy in the process. I'd say this transition occurred in the '87-'97 time frame, a time when the need for engineers to design custom hardware plummeted in favor of buying COTS. But the general purpose approach is starting to fall short of the increases necessary to drive new consumer purchases.
But innovation is still possible. Our laziness has created a deep untapped well of performance growth that can be had by equipping the domains to create domain specific designs. If we can reignite domain-specific engineering, many domains can achieve order of magnitude changes in performance by rolling their own designs.
TSMC is enabling the larger of these domains to achieve purpose-built silicon designed by the domain's engineers for the domain.
You're right and wrong. They can't afford them and are buying them anyway.
iPhones may only have about 24% market share worldwide, but the US market share is 44.5%.
Granted, 44.5% is not precisely "most", but certainly not limited to the top 10% types that can afford it. In order to have that kind of market share, there has to be a lot of people with below average incomes with iPhones.
And, as others have indicated, there is no way a tariff on smartphones would be limited to iPhones.
I feel like I've entered the twilight zone when I read an argument that storing my work on my machine is dangerous while storing it anywhere else is considered safe. Security has been compromised the moment my data isn't stored on my machine. Virtually every internet service today is a major security breach. And when someone tries to come up with something to reduce the near-requirement that all data be given up from inception, people call it a security threat? WTF? That's some pretty rich spin control.
I have Spectrum and am finding that my speeds for streaming are going up dramatically (triple or more) when I stream through a VPN. This is while always measuring the 100 MB speed I pay for or a touch more when running any speed checking service. So we have entered an age where it is useless to say you have a 100 MB connection without saying what data type you're getting that speed with.
It has only been a quarter century or so since virtually all map data came from governments. As computers reached a level of being able to pick routes and GPS became cheap, both needs and wants arose for expanded data. I can remember a few years in the 90s where applications that were technically possible were being held back for lack of that data. Many knew we needed it. It would be nice if the governments had stepped up and expanded their data for everyone's use in a timely fashion. They didn't.
The companies who had applications that could utilize the data to enough benefit to justify the billions in continuous expenditures necessary to collect it and keep it up to date did so. It would not be the least bit fair to then turn around and wipe out the benefits derived from their efforts by making the same data available to all.
A government grab of this data would be little different from a government announcing tomorrow that all entertainment in the country it oversees is now public domain and free for all to view or listen to at will. The public might cheer initially, but the industry would collapse. It would kill the goose that laid the golden egg.
At this point, it would not even be fair for the governments to step in with their own money, create an equivalent dataset including all of the imagery and analysis of that imagery, and then give it to the general public. The public wouldn't even desire the data if it weren't for the industry's risk-taking and developments that created the need for it.
Until an update made it unbootable again a few weeks ago, my Linux partition updated almost every day. Reboots were required in these updates once every couple of weeks or so. Several times a year, I'd have to spend significant time fixing a problem caused by an update. I imagine updates cost me around a week a year on average. This time, I decided to go back to my Windows partition. It just works better with my hardware - a laptop with the NVidia Optimus graphics configuration that Linux has never supported well.
I could have configured my Linux installation to be more stable by only allowing security updates, but that's not what I like as a user. I opt in to every beta I can. I've always liked being on the bleeding edge.
Windows needs to offer every user a choice in the frequency of their updates. Many are like me and want them sooner than later. I am in the preview program. Others want more stability. One size doesn't fit all.
They said they were sensitive to it, not that they like it.
I am very sensitive to bitter tastes, do not like them, avoid most bitter foods, and drink a lot of black coffee despite the fact that I don't like the bitterness. In an odd way, that is sort of why I drink it. I posted below that that sort of makes it worthwhile in a medicinal way.
My wife doesn't seem to have the ability to taste "bitter" and so would be determined to not be sensitive to bitterness. She likes a lot of foods that I consider very bitter and can't stand. She describes the taste of those foods with terms other than bitter. I can't taste those other flavors through the overwhelming bitterness and so can't appreciate what she is tasting. And, yes, she hates coffee.
I suspect these findings on coffee would hold for many alcoholic beverages that are bitter as well. There are quite a few that are considered acquired tastes that I find to be extremely bitter.
therefore it is doing something.
In general, I think this falls into our "no pain, no gain" prejudice. People routinely feel more like they are doing something worthwhile to solve a problem when the something they are doing is unpleasant in some fashion - be it exercise, diet, medicine, work, etc.
Caller ID "spoofing" is a major feature that many people often use. For example, Google Home uses it when you set it up to utilize your cell's number when you call out. There are also services that allow you to send and receive work calls using your work number from your personal cell.
But, behind the scenes, the real device making the call is always known. What is needed is a trivial means of letting private attorneys pursue the civil fines - something that these traffic ticket type shops can handle. You could dial their number after getting a robocall, they'd initiate an automatic trace, group it with others that trace to the same organization, and pursue the fines. An attorney would work pretty hard for a percentage of $10K x 100 calls or so. Push that to 1000 calls and they'll gladly start trying to go after the foreign ones too. Perhaps they could find American assets to grab.
I get a junk robocall almost every day and most who I know get more. 100 million a day in America is probably a conservative estimate. That would make $1500 a call a $150 billion per day fine rate if the fines were effective. Clearly they are not being utilized.
I have written an operating system and drivers in Ada that ran in a very complex embedded system that had a mix of over a hundred CPUs of different types. Elsewhere in that same system I also wrote an operating system with C. The interesting thing is that the one written in Ada was running on a CPU with 8 times the speed and 4 times the memory of the one written in C, but the one written in C outran it dramatically. Ada's performance sucks.
I've been running with just Windows Defender for years. As part of my work, I visit several hundred unique new internet sites every week. I haven't gotten a virus since the 90s. On the other hand, I have seen many serious system performance problems solved by removing antivirus software. I'd say that removing AV software is the second biggest performance increase you can have on a modern PC after switching to an SSD. Upgrading to a lower latency internet connection might beat it, but often isn't available (though I have found that using a VPN multiplied my internet throughput in many applications).
Which is just to say that they now have other ways of confirming your ID than paper. When all of the biometric techs get good enough, we'll be able to drop the paper again - but that is only because it has been replaced with something more reliable.
I think the shocking concept to me at the time was that anyone would ever need to prove their identity by any means to travel within the confines of our country. That was a totally foreign concept in the 60s and 70s.
True. At the time, I felt that the two big ones were FBI and Homeland Security. I remember the odor one caught my attention because I had just read an article on the sniffers that many were starting to have to walk through in the airports in those years. Those systems are not at all limited to explosive compounds and collect more than enough data to identify a person. Of course there are also many cameras in the airports that could be analyzed in multiple ways with gait being likely.
It was always shocking to me as a kid to read about how Nazi Germany actually required people to have papers to travel - unimaginable and a prime example of a dictatorial government run amok. Fast forward a few decades... try to get on a plane in the US without papers.
The National Information Exchange Model (NIEM) is an XML-based information exchange framework used by the US government to grease the sharing of information between departments. It was started in the Bush administration in 2005. I first noticed it around 2008 and it contained fields to define PersonGait at that time (version 2 I think).
They have deleted some of these fields since then. I suspect they did so because it gave out too much information about what they are doing though it is also possible the new administration rolled back the push into some of these areas.
The rules for getting data components added to the model include a requirement that the data components must already be in use by at least 2 different departments. So, in 2008, at least 2 departments of the government were using a person's gait as an identifying characteristic.
I have included a list of the fields in the "PersonAugmentationType" data type for the V2.0 version of the spec from a decade ago below. Many are interesting including: body odor, ear shape, finger geometry, gait, hand geometry, keystroke dynamics, lip movement, urine, vein pattern, etc.
The NIEM is publicly available and published on Github. I highly recommend downloading one of the older versions before they sanitized it, like V2.0 or so, and spending some time looking at the spreadsheets that describe it. The insight to be gained in what the government has reason to store about us is extensive.
Of course we need better pattern recognition, but I don't think we get there through larger nets and better training. I suspect we've reached a level of training and individual net size that is already adequate. It is actually very surprising that the individual nets we've created can compete as well as they can with humans because they are doing it without the feedback of thousands of other nets that the human brain has.
What is most needed is not better trained specialized nets. We need many nets trained in different things and connected in ways that allow them to correct each other either directly or indirectly through arbitration nets. The feedback from other nets with different specializations would have corrected the interpretation in the examples given because the interpretation didn't make sense.
I've long felt that what we are missing is the OS. We've figured out how to create the subroutines, but little is in place to put the many different subroutines into parallel contact with each other in a useful way.
My gut says this problem is not a big one. As we learn to connect large numbers of nets together in a manner that allows the whole community of nets to settle on a truth and describe it and act on it without any one central thing holding all of the knowledge of the truth, what we think of as understanding will emerge.
I get a lot of value out of this when using Google to search. If the search is for hard to find or describe data and I'm spending over half an hour searching and entering searches that approach the question from many angles, I definitely want to see the many links I've already visited in old searches highlighted in the new ones. I also research many subjects again and again over time (days, months, years, etc.) and would like to be able to distinguish previously unseen information.
In fact, it would be awesome to have a feature in Google search that I could flip on and off with a single click to just filter out previously seen information on the server side! Maybe it's there and I just haven't looked for it.
I'm not sure I see a direct threat from this for myself. I would think it would be used to inform phishing attacks and ad placement, but I'm not vulnerable to either.
On the other hand, the vast majority of internet users are not as informed as most tech users. They are vulnerable to attacks like this and we should be concerned about that because that vulnerability does affect the internet as a whole in ways that splash back on everyone by inciting regulation, limiting services, etc.
In a very, tenuously related theme, the feature I would most like added to Netflix is the ability to remove everything I've already watched from any suggestions as well as to allow me to tell it that I don't ever want to watch a particular video and have that removed from suggestions also. Of the shows and movies I've watched in my life, I doubt I watched more than 1% twice. I usually hate viewing or reading the same thing twice. Oddly, that doesn't carry to music. There is something fundamentally different there.
If that is the case, then it has become to specialized. I switched from computer science to computer engineering mid-program back in the 80s, but both had multiple statistics classes as well as database classes (different but related).
Data Science should be part of the fundamental base for any science or engineering curriculum. It is kind of hard to perform any science or engineering without it.
My kid (~ 3) has an Amazon Fire 8 for Kids. It has a huge selection of apps and videos. I have never seen an ad on it or any inappropriate apps or videos. It requires zero policing on my part.
What kid's device has this problem? Are parents actually letting their kids use adult devices or something?
Of course. But that requires a large space to walk around in and all new media. Down the road, sure. As a launch minimum, my minimum is to significantly improve on the cost, flexibility, capability and convenience provided by fixed flat panel display technology.
I've never felt that gaming is the right app for this, probably because I don't game but have wanted AR since first experiencing crude forms of it in the 90s when it was used to tremendously cut costs in aircraft wiring bundle production. I much prefer apps that offer real-life advantages.
But AR devices are still short on FOV, resolution, and brightness.
If AR could give me a full FOV with as many virtual displays as I want wherever I place them around the room and in whatever shapes I choose with no apparent pixelation and good brightness, I'd pay up to $1K for it. That would be less than what I paid for large displays in the early 90s (much less accounting for inflation). AR has the advantage that it doesn't have to render everything in full fidelity all of the time even within your FOV if it knows where your eye is pointed, so it should be achievable from a GPU POV. We just need much higher resolution full FOV AR displays that optimize rendering per where the eye is looking.
The killer launch application from my POV would be to give me banks of monitors at my desk in my living room without having to have banks of monitors. I'd love it for watching movies with virtual screens as large as I want too. But the images need to be solid with HDR and apparent resolution in the center of my vision similar to my 4K TV.
That's the minimum launch point. It can then evolve to being something portable that could eliminate the need for any product that just exists to be looked at or allow anyone in my home to see it as "decorated" in any way they like and changed with their mood.