If I were to be using chrome, this would have been a major problem for me. My documents are on a different drive, and that drive sleeps for most of its life.
(It's actually kind of funny that in 2018, on a new and wonderful build, It takes longer for me to open an mp3 or a doc file, than it did in 1985! First access of the hour wakes the drive, and between the time-delay and the drive spinning up and the case fans spinning up at the same time, it feels and even sounds almost like a floppy disk drive. It can be up to ten full seconds, though it's usually closer to five seconds.)
This feature in chrome would cost me major money, in terms of the life of my storage drives -- both HDD and SSD -- as well as the electrical expense, and the fan noise. It would also be a major curiosity and point of confusion as I'd be wondering why my machine were so active when nothing's being accessed.
You're trying to solve a problem in hardware. We're about twenty years past that. Hardware doesn't do anything anymore.
Back in my day, "drivers" were a bad thing -- there were modems, and there were winmodems, that latter needed software drivers. That logic has flipped. Now hardware does nothing without software driving it.
You're trying to double your hardware, and then add more hardware to switch between them. That's just not the equation anymore.
And in truth, you wouldn't want that. You wouldn't want to be using your SOC to browse the web, and then not be able to get that document/data/image onto your work hardware to, you know, actually work with it.
As far as protections are concerned, you're either using your SOC to access the internet to get sensitive data anyway (like e-mail) and hence you've secured absolutely nothing, or you're getting a file to transfer to your work machine, and hence you've breached your own security anyway.
If you know what you're doing, and it sounds like you could, then it's not difficult to secure your work data from your internet connection. Think about the easy things -- like a second hdd/ssd for the work file.
Secondary storage drives are easily turned off in device manager on a whim. Don't visit terrible sites at all. Don't walk down dark alleys with your 10-year-old daughter ever. Know how to clear buffers, and generally know that all's clear before spinning up that work drive.
But most of all, know: that Ethan Hunt can always break in, that there aren't as many Ethan Hunts as you've been led to believe, that most of the time, Ethan Hunt doesn't actually harm you when he gets what he wants.
You aren't actually responsible for the edge cases, so don't expend all of your energy defending against them.
It's been a long time since Dr. Sbaitso. I don't think I've heard a digital assistant with a male voice since.
Women complain about gender bias all the time. How come men don't complain that our cars, computers, smartphones, and household assistants don't sound like us?
I cry foul sexism. We ought to be insisting that Siri come with a male-voice option. It's only fair.
Yeah, I read that, but the thing is that I don't believe the control groups are control groups if they are all people who would volunteer to be a part of such a study.
I've got a problem with measuring aggression and lack of social-ness of individuals who contribute their time and patience to research studies -- which by definition are helping other people (i.e. the researchers).
So I completely disagree with the 10%, and hence I also disagree that 3% is noise.
I think that if a person would volunteer to be a) be told what to do; b) do it every day for 60 days; c) fill out so many questionnaires; and d) get nothing out of it, that said person simply ain't pre-disposed to aggression in the first place.
I also think that any action that would affect a person's behaviour doesn't make them act differently, rather it makes them act "more" or "more often" or "more intensely" than before. (or less, as might be the case with would-be murderers who become butchers instead).
Combining those two, I think this study ought to have observed absolutely zero aggression, or even negative aggression, but wound up seeing 3% aggression.
That's the way I see it. And when I think about myself, I'm happy to participate is really quick studies, even thirty minutes of my time, but two months? Just thinking about that makes me mad.
3 of 90 is over 3% of a very huge population of millions and millions and millions of gamers. It the study's anywhere near predictive, that's a lot of damage.
I hate to say it because I spend a lot of time playing violent video games. And hey, no one's accusing me of very overly calm or patient. I'm certainly sexist, and I'm certainly not whatever interpersonally competent is.
Do I get to blame the games?
I've always wanted to be a member of the 1%. I suppose being a part of the 3% is pretty close.
I disagree. What you're saying makes sense only when you're saying it standing in one place.
The airspace rules, road rules, commercial rules, and employment rules are drastically different from place to place. So if you're going to try to describe everyone the same way, then you fall into the horrible trap familiar to most as XML. Look, a standard definition language that has no meaning in and of itself. So you spend forever defining stuff.
Spoken languages are, historically, different for those same reasons. Some cultures focus on 1 vs 2 vs many (one, both, all / individual, couple, group), others focus on 1v2v3 vs many. Some use a thousands separator for finances, (100'000.00) others use a ten-thousands separator (10'0000.00) because the culture is different. Some languages are harsh sounds vs soft sounds, because the climate is different, or the distances. Some are focused on animals vs others on legal systems, vs others on food, so the number of words by category reflects the resolution of significance to that culture. I don't know how many words for "snow" are used in Florida. I've heard the Canadian North have many. Here in Toronto (the Canadian South), I can easily count to 10, and I'd bet I could read 20 words for "snow" pretty quickly.
Here's a very quick example. Google's system doesn't have altitude. I guess Amazon won't be delivering to apartment balconies by drone -- which seams, to me at least, to be the best use-case for drone delivery. And, as I said originally, if it's by cartesian geography, that doesn't help me navigate the maze of one-way streets to get there. Nor does it help with the night-time restrictions on airspace.
One definition system can't work in multiple different environmental systems. Compatible systems need to match. It's that simple.
And that's why longitude and latitude are so perfect. Sure, they aren't easy for humans. But they weren't meant for humans. They are superbly easy for other systems to use, because they are incredibly simple.
Case in point: longitude and latitude work on other planets too. Does google's new system work on asteroids? Aircraft carriers? Beach-balls? I can use longitude and latitude on any object that fits inside an ellipsoid -- which is any object. A third dimension and I'm 3D. And in all of those cases, there's no need to define anything, no need for algorithms, no need for anything except a set of zero co-ordinates.
Specifically, Canadian postal codes are not geographic, but nothing's stopping the algorithm from being so. The first character denotes (roughly) the province, from east to west. The rest is the postal delivery route -- which means my neighbours to the left and to the right are identical or off by a single number, whereas the neighbour behind me (on a different street) is off by two characters. While not "geographic" in spatial terms, it's geographic by postal route, which is (obviously) by street. So following directions by car would actually be easier. That said, it could just as easily be adopted longitudinally. The other features are identical:
A1A 1A1 is roughly five houses big. Add one more character, and you'll divide that region into 26. A1A is roughly 100 square miles (in Ontario, but again it could be adjusted). A1A 1A* is about 100 houses.
The code alternates letter-number-letter-number-letter-number. So a) there are no words in any language; and b) B vs 8, 0 vs O are irrelevant. Type it, write it, speak it either way, it's discernible.
So I guess I'm proposing the Canadian Postal Code format, simply remapped geographically, with two or three additional characters.
Doesn't even mention the Canadian Postal Code, which seems to satisfy nearly all of google's requirements. Add two more characters, and it would satisfy, I think, each and every single one.
How about a study ranking the languages that seem to be the most reliable -- as in they've been used for decades, continue to be used, run systems un-upgraded for decades, and need little-to-no new help from random strangers?
In my career, I've been using Perl since 1997, when I abandoned lotus notes. I've still got production code from back then -- it's funny to see some of my really old comments from half-a-life ago.
I'll say this for perl: like it or hate it, the documentation is everywhere and is incredibly complete -- that's no surprise after so many decades.
I don't know of a single child who's been trained to recognize humans by being fed a million faces. I also don't know of a single child who can't recognize any human face -- black, white, healthy, disfigured, old, young, or partial.
Same goes for my dog, by the way. i don't need to see every fish in order to recognize a new fish as being a fish. Fins swim. Grade 9 science was "how to read your fish".
AI is certainly artificial, but the "intelligence" part is being taken far too literally. More intel isn't any classic definition of intelligence.
The day your AI can recognize a human based on features, and not based on templates, then maybe you'll have crossed the line into what we actually call intelligence. Psychology, not computer science.
You know, two eyes (or one eye and a patch, or sunglasses), a nose (broad, narrow, broken or bleeding), a mouth (open, closed, grinning, or covered).
I don't think it's difficult to recognize a "mouth". Whether it's human, canine, iguana, or worm. A human face has two eyes, one mouth, and a nose between them. Any kind of mouth, any kind of nose, any kind of eyes. And any of those four features can be covered, damaged, open, closed, or concealed.
You don't have an AI capable of identifying a mouth. Therefore, you have nothing.
Let me know when your AI can recognize a mouth on 90% of lizards and mammals and cartoons. Then we'll talk about AI taking over the world.
You don't have image recognition. What you have is visual pattern matching. Humans don't do visual pattern matching, just like humans don't walk around a table by using calculus. We don't walk by knowing the length of our legs. We don't drive by knowing precisely how far away the next car is. We don't use lasers.
We see through blizzards, fog, rain, ice, and sun. We don't need perfectly clean lenses to do so.
I'll say it again. If your AI needs to be fed more than 100 examples of something, then it isn't learning shit. And if it needs to remember those examples, aggregated or otherwise, then again it isn't learning anything at all.
Two eyes, a mouth, and a nose between them. You need to identify four features in relative proximity. Start there.
I've only had my schedule.todo.txt file for 19 years now.
ftp, download, edit, upload, done.
grab a quick web site, dropbox, one drive, or any other file hosting service if you don't like ftp as a protocol for transferring files.
start lines with a date stamp or two, and you can sort it in any reasonable text editor, or manually in mere moments.
make it a spreadsheet instead of a text file if you really think you're productive enough to get through that many todo items in the first place.
why are you making it difficult? whatever solution you use, it won't complete the tasks for you. Why does it need to be in active development? You think your todo list is any more complicated than your grandmother's was 50 years ago?
So you're telling me that someone thought a car owner would leave their car at home and hire a ride instead? I bought a car. It's my car. It gives me freedom. You're expecting me to give up my own car to take someone else's. That was pretty stupid of you.
For reference, owning a car was never cheap. And in a city, owning a car was always more expensive than hiring taxis. Even in suburbs, taxis are cheaper. Just about every car owner spends ~$10'000 per year on their car. That's a shitload of public transit.
Congrats on an AI that can play a game that takes decades for an adult to master.
Let me know when you have an AI that can do what an infant takes a week to master -- like distinguish a gun from an apple.
We still don't have any sort of robotics that can live in the world. Sure we have robotic assembly lines -- welcome to a controlled environment.
I spent about twenty minutes trimming the ends off of snow peas for dinner yesterday. Got a robot that can grab a kitchen knife, and a bucket of snow peas and trim them? Or hull a hundred strawberries? Or peel fifty pearl onions? Can it even core and slice an apple?
I'm just waiting for the self-driving car that so wants to avoid hitting pedestrians that it'll back up when I walk towards it. I'll be able to herd self-driving cars like herding cattle.
Right now, it sounds like AI will be great at giving orders to humans -- after humans input all of the data. GPS routing? Nope! Gotta have a human drive every road first. Stupid self-driving car can't navigate without a map.
Last I checked, maps don't exist until someone actually surveys the land. Last I checked, a compass works in most unknown territory.
If you're trying to raise your child to be the next blue-collar peon, then by all means this is how to do it. You can teach "coding", and they'll get paid minimum-wage to "code".
This is precisely the same as painting. You can teach your child to paint, and they'll be able to paint walls.
But you can't teach creativity, creative innovation, nor artistic creative innovation -- that means problem solving.
I don't think that you'll find an experienced programmer, successful and senior and making real money, who isn't mostly self-taught.
Humans learn problem solving in the only manner than any living thing has ever learned problem solving skills -- by having problems and fighting with them until someone wins.
If girls don't have the patience, or the dedication, or the motivation, or the self esteem to work a problem alone, until it's gone. ..if a person insists on direct hand-holding (as opposed to documentation or occasional guidance) to work out a solution to a problem affecting them. ..then this ain't a'gonna be their day-job, so to speak. This ain't their forte.
In the past two weeks, I've watched girls take the "shallow side" of the mountain in slope-style, get lifted in skating, and basically do push-ups from their knees -- a.k.a. "girlie push-ups". I'm no athlete, I sit at a desk 80 hours a week, but when I go to the gym next to the jocks, I play the same game they do. There's a mutual respect in that. When I golf (I don't golf) I play from the same tees as the regular golfers (the ladies tees are in-front of the amateurs, by the way).
Women don't deserve equal respect for playing a dumbed-down version, just like they wouldn't deserve equal pay for dumbed-down work.
I've recently been convinced that all of this is engrained into girls at a young age -- that they aren't as good as men, aren't as strong as men, aren't as fast as men. I have no idea if that's true of gladiator men, or hockey playing men, but I promise you that most women are faster and stronger than I am.
But I wasn't raised by Barbie. I was raised by Mr. Wizard. Maybe the hockey playing men were raised by G.I. Joe?
I am impressed by Barbie today though. She's come a long way. You wouldn't expect coding from someone who used to think that "math is hard".
So, with different hardware, anything that relies on non-ARM hardware drivers can't. Seems logical.
Also, since it's windows10, and not windows7, older graphics requires aren't there.
Again, seems logical. I'm guessing that the windows user interface has been re-built, and probably lacks a whole lot of backwards compatibility. Given that ARM devices are likely performance restricted by comparison, it's no surprise to me that there's little or no access to any ARM-specialized user interfaces (i.e. all of them). Not really disappointing, but a limitation for sure.
64-bit apps is obviously a momentary restriction -- it won't last long. That's simply proof that the project began before wide-spread 64-bit ARM adoption. It'll follow suit in short order.
HyperV, I'm sure, is similarly a few generations ahead of the windows ARM project. I'd bet that adoption of windows on ARM will dictate whether or not HyperV gets brought in at all. And reasonably so.
Interesting article, but I think my life exemplifies this particular problem, and highlights the reasons behind the problem itself.
Tools.
Imagine two plumbers. One master experienced plumber. One junior plumber. Maybe the junior helps speed up the master, maybe he doesn't. In either case, the master plumber can only do so much alone.
Then we add really good plumbing tools (welders, wrenches, et cetera) into the mix. Now the master can do a lot more. As a result, these advanced tools become justifiable. With the master using advanced tools, the gap to the junior is a) even bigger; and b) easier to close with the tools. Master teaches junior to use advanced tools, junior becomes much more valuable.
I'm not a plumber. I'm a senior (owner) web developer. Our industry is very different because our tools work very differently from welders and wrenches.
In programming, our tools consist of elements that are actually far more complicated than the programming itself. Simple programming has always been simple. But IDEs, APIs, version control systems, development mirrors, and the like are actually far more complicated than logo and javascript and html ever were.
A wrench makes turning a bolt easier. An API makes turning a thousand bolts easier than turning ten manually, but APIs are far more complicated than turning one bolt.
This leads to our senior developer's advanced experience being less about experience in programming, and more about experience in how to program. This has a few complications:
First, that kind of knowledge is much more difficult to transfer -- its conceptual, its layered, its abstract. While it's likely that any junior plumber can be taught to use a torch very quickly, it's unlikely that any junior programmer can be taught to use an IDE without months of practice.
Second, and this is back to that gap from before, the senior programmer with the senior experience is far too valuable using that experience to warrant the effort to teach it. A senior developer with senior tools and senior experience, properly motivated, might be more productive than a good team of a 100 good juniors, and that's if juniors can do the job at all. Asking a senior to teach a junior is basically like saying you have no work for the senior for the year it'll take to teach a junior anything worthwhile.
Third, and this is the problem that is especially my problem, I don't want to work that hard. It's easy for me to work as a senior programmer. It's easy for me to do the work and get paid and move on. It's easy for me to charge a lot, take a lot of responsibility, and take a lot of time off. I'm fast, I'm efficient. I really don't have any interest in teaching humans. I chose a career where I tell machines what to do, in part because I have zero interest in motivating humans. Over the years I've hired junior developers, I'm fired junior developers, I've hired my own boss, I've hired sales personnel, I've hired advertisers. In the end, I've shed them all because I simply want to program, and I don't need any help with it.
The article concluded that it's an industry problem. Maybe it is, but I see it differently. It's a senior's solution. This is what I'm doing. I like what I'm doing. What I'm doing works for me. I'll keep doing it.
I quoted a guy, who happens to be an authoritarian on the subject. That is all. I'm not citing a publication of his.
It is possible for a person to be an authority on a subject, without having ever published an article.
It is also possible for you to learn from someone by speaking with them, or by hearing them speak, forever without a recording device present.
Again, I could care less how you feed your family. I choose to learn from an expert directly. You can choose to disbelieve the second-hand information -- me.
Funny thing is that the entirety of wikipedia is second-hand information. That you would believe it, and not me, is your inconsistency. Would you feel better if I published it onto wikipedia, with a citation to a random episode where he doesn't say anything? Do you even check citations when you read?
I like how you want me to cite something that I learned over a decade ago, as though I keep a bibliography of my everyday life. That you can't find it on google is completely and utterly as meaningless as what you can find on google.
So that's my citation. Go find the episode and date on your own if you like. I'm happy eating my dinner. I really don't care if you're happy eating yours. Feed your family whatever the hell you want.
You know, that would have really helped about thirty years ago when my drive was only 500MB.
Actually, that would have sucked thirty years ago, because thirty years ago, e-mail was easily under 5MB of total disk space. I do believe it was often under 3MB, and I'm pretty sure it was possible under 1MB.
It's e-mail. It needn't be complicated.
(oooooh, fancy sorting, that probably needs another 50 KB)
Slashdot Mirror
If I were to be using chrome, this would have been a major problem for me. My documents are on a different drive, and that drive sleeps for most of its life.
(It's actually kind of funny that in 2018, on a new and wonderful build, It takes longer for me to open an mp3 or a doc file, than it did in 1985! First access of the hour wakes the drive, and between the time-delay and the drive spinning up and the case fans spinning up at the same time, it feels and even sounds almost like a floppy disk drive. It can be up to ten full seconds, though it's usually closer to five seconds.)
This feature in chrome would cost me major money, in terms of the life of my storage drives -- both HDD and SSD -- as well as the electrical expense, and the fan noise. It would also be a major curiosity and point of confusion as I'd be wondering why my machine were so active when nothing's being accessed.
I thought that when I was echo-ing AT-commands to the COM port (i.e. echo ATDT > COM1) that I wasn't going through any modem-specific drivers.
I don't believe I know your name.
I [obviously] meant don't go down dark alleys with something to lose.
You're trying to solve a problem in hardware. We're about twenty years past that. Hardware doesn't do anything anymore.
Back in my day, "drivers" were a bad thing -- there were modems, and there were winmodems, that latter needed software drivers. That logic has flipped. Now hardware does nothing without software driving it.
You're trying to double your hardware, and then add more hardware to switch between them. That's just not the equation anymore.
And in truth, you wouldn't want that. You wouldn't want to be using your SOC to browse the web, and then not be able to get that document/data/image onto your work hardware to, you know, actually work with it.
As far as protections are concerned, you're either using your SOC to access the internet to get sensitive data anyway (like e-mail) and hence you've secured absolutely nothing, or you're getting a file to transfer to your work machine, and hence you've breached your own security anyway.
If you know what you're doing, and it sounds like you could, then it's not difficult to secure your work data from your internet connection. Think about the easy things -- like a second hdd/ssd for the work file.
Secondary storage drives are easily turned off in device manager on a whim.
Don't visit terrible sites at all. Don't walk down dark alleys with your 10-year-old daughter ever.
Know how to clear buffers, and generally know that all's clear before spinning up that work drive.
But most of all, know:
that Ethan Hunt can always break in,
that there aren't as many Ethan Hunts as you've been led to believe,
that most of the time, Ethan Hunt doesn't actually harm you when he gets what he wants.
You aren't actually responsible for the edge cases, so don't expend all of your energy defending against them.
Really? Alexa comes with an Alex? I had no idea. I'll check that the next time I'm at the neighbour's.
It's been a long time since Dr. Sbaitso. I don't think I've heard a digital assistant with a male voice since.
Women complain about gender bias all the time. How come men don't complain that our cars, computers, smartphones, and household assistants don't sound like us?
I cry foul sexism. We ought to be insisting that Siri come with a male-voice option. It's only fair.
I am person, hear me roar.
Yeah, I read that, but the thing is that I don't believe the control groups are control groups if they are all people who would volunteer to be a part of such a study.
I've got a problem with measuring aggression and lack of social-ness of individuals who contribute their time and patience to research studies -- which by definition are helping other people (i.e. the researchers).
So I completely disagree with the 10%, and hence I also disagree that 3% is noise.
I think that if a person would volunteer to be a) be told what to do; b) do it every day for 60 days; c) fill out so many questionnaires; and d) get nothing out of it, that said person simply ain't pre-disposed to aggression in the first place.
I also think that any action that would affect a person's behaviour doesn't make them act differently, rather it makes them act "more" or "more often" or "more intensely" than before. (or less, as might be the case with would-be murderers who become butchers instead).
Combining those two, I think this study ought to have observed absolutely zero aggression, or even negative aggression, but wound up seeing 3% aggression.
That's the way I see it. And when I think about myself, I'm happy to participate is really quick studies, even thirty minutes of my time, but two months? Just thinking about that makes me mad.
3 of 90 is over 3% of a very huge population of millions and millions and millions of gamers. It the study's anywhere near predictive, that's a lot of damage.
I hate to say it because I spend a lot of time playing violent video games. And hey, no one's accusing me of very overly calm or patient. I'm certainly sexist, and I'm certainly not whatever interpersonally competent is.
Do I get to blame the games?
I've always wanted to be a member of the 1%. I suppose being a part of the 3% is pretty close.
I disagree. What you're saying makes sense only when you're saying it standing in one place.
The airspace rules, road rules, commercial rules, and employment rules are drastically different from place to place. So if you're going to try to describe everyone the same way, then you fall into the horrible trap familiar to most as XML. Look, a standard definition language that has no meaning in and of itself. So you spend forever defining stuff.
Spoken languages are, historically, different for those same reasons. Some cultures focus on 1 vs 2 vs many (one, both, all / individual, couple, group), others focus on 1v2v3 vs many. Some use a thousands separator for finances, (100'000.00) others use a ten-thousands separator (10'0000.00) because the culture is different. Some languages are harsh sounds vs soft sounds, because the climate is different, or the distances. Some are focused on animals vs others on legal systems, vs others on food, so the number of words by category reflects the resolution of significance to that culture. I don't know how many words for "snow" are used in Florida. I've heard the Canadian North have many. Here in Toronto (the Canadian South), I can easily count to 10, and I'd bet I could read 20 words for "snow" pretty quickly.
Here's a very quick example. Google's system doesn't have altitude. I guess Amazon won't be delivering to apartment balconies by drone -- which seams, to me at least, to be the best use-case for drone delivery. And, as I said originally, if it's by cartesian geography, that doesn't help me navigate the maze of one-way streets to get there. Nor does it help with the night-time restrictions on airspace.
One definition system can't work in multiple different environmental systems. Compatible systems need to match. It's that simple.
And that's why longitude and latitude are so perfect. Sure, they aren't easy for humans. But they weren't meant for humans. They are superbly easy for other systems to use, because they are incredibly simple.
Case in point: longitude and latitude work on other planets too. Does google's new system work on asteroids? Aircraft carriers? Beach-balls? I can use longitude and latitude on any object that fits inside an ellipsoid -- which is any object. A third dimension and I'm 3D. And in all of those cases, there's no need to define anything, no need for algorithms, no need for anything except a set of zero co-ordinates.
Specifically, Canadian postal codes are not geographic, but nothing's stopping the algorithm from being so. The first character denotes (roughly) the province, from east to west. The rest is the postal delivery route -- which means my neighbours to the left and to the right are identical or off by a single number, whereas the neighbour behind me (on a different street) is off by two characters. While not "geographic" in spatial terms, it's geographic by postal route, which is (obviously) by street. So following directions by car would actually be easier. That said, it could just as easily be adopted longitudinally. The other features are identical:
A1A 1A1 is roughly five houses big. Add one more character, and you'll divide that region into 26.
A1A is roughly 100 square miles (in Ontario, but again it could be adjusted).
A1A 1A* is about 100 houses.
The code alternates letter-number-letter-number-letter-number. So a) there are no words in any language; and b) B vs 8, 0 vs O are irrelevant. Type it, write it, speak it either way, it's discernible.
So I guess I'm proposing the Canadian Postal Code format, simply remapped geographically, with two or three additional characters.
Doesn't even mention the Canadian Postal Code, which seems to satisfy nearly all of google's requirements. Add two more characters, and it would satisfy, I think, each and every single one.
Beat me to it.
How about a study ranking the languages that seem to be the most reliable -- as in they've been used for decades, continue to be used, run systems un-upgraded for decades, and need little-to-no new help from random strangers?
In my career, I've been using Perl since 1997, when I abandoned lotus notes. I've still got production code from back then -- it's funny to see some of my really old comments from half-a-life ago.
I'll say this for perl: like it or hate it, the documentation is everywhere and is incredibly complete -- that's no surprise after so many decades.
I don't know of a single child who's been trained to recognize humans by being fed a million faces. I also don't know of a single child who can't recognize any human face -- black, white, healthy, disfigured, old, young, or partial.
Same goes for my dog, by the way. i don't need to see every fish in order to recognize a new fish as being a fish. Fins swim. Grade 9 science was "how to read your fish".
AI is certainly artificial, but the "intelligence" part is being taken far too literally. More intel isn't any classic definition of intelligence.
The day your AI can recognize a human based on features, and not based on templates, then maybe you'll have crossed the line into what we actually call intelligence. Psychology, not computer science.
You know, two eyes (or one eye and a patch, or sunglasses), a nose (broad, narrow, broken or bleeding), a mouth (open, closed, grinning, or covered).
I don't think it's difficult to recognize a "mouth". Whether it's human, canine, iguana, or worm. A human face has two eyes, one mouth, and a nose between them. Any kind of mouth, any kind of nose, any kind of eyes. And any of those four features can be covered, damaged, open, closed, or concealed.
You don't have an AI capable of identifying a mouth. Therefore, you have nothing.
Let me know when your AI can recognize a mouth on 90% of lizards and mammals and cartoons. Then we'll talk about AI taking over the world.
You don't have image recognition. What you have is visual pattern matching. Humans don't do visual pattern matching, just like humans don't walk around a table by using calculus. We don't walk by knowing the length of our legs. We don't drive by knowing precisely how far away the next car is. We don't use lasers.
We see through blizzards, fog, rain, ice, and sun. We don't need perfectly clean lenses to do so.
I'll say it again. If your AI needs to be fed more than 100 examples of something, then it isn't learning shit. And if it needs to remember those examples, aggregated or otherwise, then again it isn't learning anything at all.
Two eyes, a mouth, and a nose between them. You need to identify four features in relative proximity. Start there.
I've only had my schedule.todo.txt file for 19 years now.
ftp, download, edit, upload, done.
grab a quick web site, dropbox, one drive, or any other file hosting service if you don't like ftp as a protocol for transferring files.
start lines with a date stamp or two, and you can sort it in any reasonable text editor, or manually in mere moments.
make it a spreadsheet instead of a text file if you really think you're productive enough to get through that many todo items in the first place.
why are you making it difficult? whatever solution you use, it won't complete the tasks for you. Why does it need to be in active development? You think your todo list is any more complicated than your grandmother's was 50 years ago?
So you're telling me that someone thought a car owner would leave their car at home and hire a ride instead? I bought a car. It's my car. It gives me freedom. You're expecting me to give up my own car to take someone else's. That was pretty stupid of you.
For reference, owning a car was never cheap. And in a city, owning a car was always more expensive than hiring taxis. Even in suburbs, taxis are cheaper. Just about every car owner spends ~$10'000 per year on their car. That's a shitload of public transit.
Way to solve a problem by destroying the age-old solution to a long-past problem. The circle of forgetfulness.
Show of hands: who remembers why we have disposable plastic products in the first place? Anyone? Anyone from Taiwan?
Congrats on an AI that can play a game that takes decades for an adult to master.
Let me know when you have an AI that can do what an infant takes a week to master -- like distinguish a gun from an apple.
We still don't have any sort of robotics that can live in the world. Sure we have robotic assembly lines -- welcome to a controlled environment.
I spent about twenty minutes trimming the ends off of snow peas for dinner yesterday. Got a robot that can grab a kitchen knife, and a bucket of snow peas and trim them? Or hull a hundred strawberries? Or peel fifty pearl onions? Can it even core and slice an apple?
I'm just waiting for the self-driving car that so wants to avoid hitting pedestrians that it'll back up when I walk towards it. I'll be able to herd self-driving cars like herding cattle.
Right now, it sounds like AI will be great at giving orders to humans -- after humans input all of the data. GPS routing? Nope! Gotta have a human drive every road first. Stupid self-driving car can't navigate without a map.
Last I checked, maps don't exist until someone actually surveys the land. Last I checked, a compass works in most unknown territory.
If you're trying to raise your child to be the next blue-collar peon, then by all means this is how to do it. You can teach "coding", and they'll get paid minimum-wage to "code".
This is precisely the same as painting. You can teach your child to paint, and they'll be able to paint walls.
But you can't teach creativity, creative innovation, nor artistic creative innovation -- that means problem solving.
I don't think that you'll find an experienced programmer, successful and senior and making real money, who isn't mostly self-taught.
Humans learn problem solving in the only manner than any living thing has ever learned problem solving skills -- by having problems and fighting with them until someone wins.
If girls don't have the patience, or the dedication, or the motivation, or the self esteem to work a problem alone, until it's gone. . .if a person insists on direct hand-holding (as opposed to documentation or occasional guidance) to work out a solution to a problem affecting them. . .then this ain't a'gonna be their day-job, so to speak. This ain't their forte.
In the past two weeks, I've watched girls take the "shallow side" of the mountain in slope-style, get lifted in skating, and basically do push-ups from their knees -- a.k.a. "girlie push-ups". I'm no athlete, I sit at a desk 80 hours a week, but when I go to the gym next to the jocks, I play the same game they do. There's a mutual respect in that. When I golf (I don't golf) I play from the same tees as the regular golfers (the ladies tees are in-front of the amateurs, by the way).
Women don't deserve equal respect for playing a dumbed-down version, just like they wouldn't deserve equal pay for dumbed-down work.
I've recently been convinced that all of this is engrained into girls at a young age -- that they aren't as good as men, aren't as strong as men, aren't as fast as men. I have no idea if that's true of gladiator men, or hockey playing men, but I promise you that most women are faster and stronger than I am.
But I wasn't raised by Barbie. I was raised by Mr. Wizard. Maybe the hockey playing men were raised by G.I. Joe?
I am impressed by Barbie today though. She's come a long way. You wouldn't expect coding from someone who used to think that "math is hard".
So, with different hardware, anything that relies on non-ARM hardware drivers can't. Seems logical.
Also, since it's windows10, and not windows7, older graphics requires aren't there.
Again, seems logical. I'm guessing that the windows user interface has been re-built, and probably lacks a whole lot of backwards compatibility. Given that ARM devices are likely performance restricted by comparison, it's no surprise to me that there's little or no access to any ARM-specialized user interfaces (i.e. all of them). Not really disappointing, but a limitation for sure.
64-bit apps is obviously a momentary restriction -- it won't last long. That's simply proof that the project began before wide-spread 64-bit ARM adoption. It'll follow suit in short order.
HyperV, I'm sure, is similarly a few generations ahead of the windows ARM project. I'd bet that adoption of windows on ARM will dictate whether or not HyperV gets brought in at all. And reasonably so.
Interesting article, but I think my life exemplifies this particular problem, and highlights the reasons behind the problem itself.
Tools.
Imagine two plumbers. One master experienced plumber. One junior plumber. Maybe the junior helps speed up the master, maybe he doesn't. In either case, the master plumber can only do so much alone.
Then we add really good plumbing tools (welders, wrenches, et cetera) into the mix. Now the master can do a lot more. As a result, these advanced tools become justifiable. With the master using advanced tools, the gap to the junior is a) even bigger; and b) easier to close with the tools. Master teaches junior to use advanced tools, junior becomes much more valuable.
I'm not a plumber. I'm a senior (owner) web developer. Our industry is very different because our tools work very differently from welders and wrenches.
In programming, our tools consist of elements that are actually far more complicated than the programming itself. Simple programming has always been simple. But IDEs, APIs, version control systems, development mirrors, and the like are actually far more complicated than logo and javascript and html ever were.
A wrench makes turning a bolt easier. An API makes turning a thousand bolts easier than turning ten manually, but APIs are far more complicated than turning one bolt.
This leads to our senior developer's advanced experience being less about experience in programming, and more about experience in how to program. This has a few complications:
First, that kind of knowledge is much more difficult to transfer -- its conceptual, its layered, its abstract. While it's likely that any junior plumber can be taught to use a torch very quickly, it's unlikely that any junior programmer can be taught to use an IDE without months of practice.
Second, and this is back to that gap from before, the senior programmer with the senior experience is far too valuable using that experience to warrant the effort to teach it. A senior developer with senior tools and senior experience, properly motivated, might be more productive than a good team of a 100 good juniors, and that's if juniors can do the job at all. Asking a senior to teach a junior is basically like saying you have no work for the senior for the year it'll take to teach a junior anything worthwhile.
Third, and this is the problem that is especially my problem, I don't want to work that hard. It's easy for me to work as a senior programmer. It's easy for me to do the work and get paid and move on. It's easy for me to charge a lot, take a lot of responsibility, and take a lot of time off. I'm fast, I'm efficient. I really don't have any interest in teaching humans. I chose a career where I tell machines what to do, in part because I have zero interest in motivating humans. Over the years I've hired junior developers, I'm fired junior developers, I've hired my own boss, I've hired sales personnel, I've hired advertisers. In the end, I've shed them all because I simply want to program, and I don't need any help with it.
The article concluded that it's an industry problem. Maybe it is, but I see it differently. It's a senior's solution. This is what I'm doing. I like what I'm doing. What I'm doing works for me. I'll keep doing it.
I quoted a guy, who happens to be an authoritarian on the subject. That is all. I'm not citing a publication of his.
It is possible for a person to be an authority on a subject, without having ever published an article.
It is also possible for you to learn from someone by speaking with them, or by hearing them speak, forever without a recording device present.
Again, I could care less how you feed your family. I choose to learn from an expert directly. You can choose to disbelieve the second-hand information -- me.
Funny thing is that the entirety of wikipedia is second-hand information. That you would believe it, and not me, is your inconsistency. Would you feel better if I published it onto wikipedia, with a citation to a random episode where he doesn't say anything? Do you even check citations when you read?
I like how you want me to cite something that I learned over a decade ago, as though I keep a bibliography of my everyday life. That you can't find it on google is completely and utterly as meaningless as what you can find on google.
This man:
https://en.wikipedia.org/wiki/...
On this television show:
https://en.wikipedia.org/wiki/...
Across the street from this deli:
http://schwartzsdeli.com/
From inside this deli:
http://www.maindelisteakhouse....
Said so.
So that's my citation. Go find the episode and date on your own if you like. I'm happy eating my dinner. I really don't care if you're happy eating yours. Feed your family whatever the hell you want.
Yeah, I figured if I said it was a 20MB disk, most would be shocked that I had any internet at all.
E-mail in only 25MB? Wow! That's amazing!
You know, that would have really helped about thirty years ago when my drive was only 500MB.
Actually, that would have sucked thirty years ago, because thirty years ago, e-mail was easily under 5MB of total disk space. I do believe it was often under 3MB, and I'm pretty sure it was possible under 1MB.
It's e-mail. It needn't be complicated.
(oooooh, fancy sorting, that probably needs another 50 KB)