That argument is bullshit from someone who doesn't understand machine learning (let alone "deep" learning).
You can use a machine learning face recognition system to identify "face geometry" features trivially, simply by using the output of the system with a postprocessing layer. The combined system is a system which collects "face geometry". As such, the black box nature of Shutterfly's system doesn't preclude the claimed illegal behaviour. Frankly, the only evidence that Shutterfly and Facebook don't break the law is their word, which suggests the lawsuit is entirely reasonable so that we may get to the bottom of this.
An irrevocable patent pledge is intended to be precisely that; it's a legal document that is written to
carry weight regardless of changes of ownership or management. Whether it will stand in court when tested
remains to be seen, but that problem applies also to free content and open source licenses and other
legal tools for sharing of information and ideas.
Such a patent pledge document is pure theatre, and should not be confused with open source licenses.
The former relies on *the* issuing company behaving in a certain way, whereas the latter relies on *some* licensees behaving in a certain way.
Taking as a given that people change, their motivations, circumstances, relationships change over time, you can and should expect that both KA and open source licensees will change their minds and act contrary to the legal documents they agreed to initially. When that happens, the difference is stark: in the case of the patent pledge, everyone who built their products using said patent becomes an infringer, and the potential for damage and misuse is huge and worldwide (eg SCO). In the case of the open source license, those who misuse the code according to the license are merely stealing code that doesn't belong to them, and this has no effect on the other open source recipients, it's on a case by case basis.
In other words, a patent pledge breaks if one entity changes its mind, whereas an open source license doesn't break if one entity changes its mind. And expecting one entity to change their mind is a very low bar.
What KA should be doing is creating a document, similar to the BSD license, which automatically licenses their patents to anyone for free and forever. That will not happen, because the true motivation of buying dubious patents is to build a war chest that can be used to threaten others, and there would be absolutely no point in paying for a patent if it had no threat value.
I don't think that calling projects started by women "segregated" is helpful. You are looking at the state of open source through racial conflict tinted glasses. This is not the same problem as black/white integration at all.
What is open source about? It's about scratching your own itch. If some mainstream windows like system doesn't work for you, you don't complain and you don't propose to force windows to do things your way. No, you fork a project and do your own thing. If other people like where you are going with this, they'll join and the project grows. If not, there's no harm in having a project of one.
Either way, nobody is being forced to do anything, people CHOOSE freely to do what works for them at the time.
THAT is the open source way, and that is what is being suggested to you IMHO. If someone wants to fork an open source project with the special feature that cussing and disrespecting women members is not acceptable, they should do exactly that. Fork and run with the idea. Make the rules clear on the mailing list and enforce them. THAT is the open source way.
And people will come, and people will choose, and the new projects will grow and take over mindshare, like Linux and GNU grew, and took over mind share, and 30 years later we have a thriving ecosystem. Linux didn't grow fully formed overnight. Neither will your kinder, gentler open source community. So relax, don't tell the community what it should do, don't be impatient. Start by forking an interesting project, experiment with social rules, and see if the idea works out. If you truly have a good idea, people will come and help out. If not, keep going.
That doesn't work. What if Khan gets a new management and/or new owners? They can change the rules and repudiate
all those promises, no matter what.
For example, Google used to be trustworthy in the early days, now they are an evil spying organisation who takes so much heat that even the founders are distancing themselves with a nondescript holding company called alphabet.
If it can happen to Google, who had so much promise in 1999, then you betcha it will happen to Khan once the pressure of economics start to bite them.
Trusting the intentions of companies is simply a woeful approach to economic decision making.
Patents should be granted and immediately sold exclusively to the government, who can then open them up for everyone to use. A good example of how this works is the photography patent which was sold by Louis Daguerre to the French government, who opened the technology to the world.
In this way we fix all the greatest problems with patents in one swoop. No more patent portfolios that create barriers to entry for innovative new startups, no more frivolous patent lawsuits about square looking smartphones, no more chilling effects based on the threat of being sued, no more protection money (aka licensing/royalties) rackets, and if the government must buy all potential patents then there is a balance between income (from patent examination fees) and expenditure (from buying the granted patents) so that the worst excesses of the current system are automatically held in check.
When a guy kills a random person just to point out that the police can't be everywhere at the same time, he's not offering a valuable insight into society to change it, he's just a murderer being cheeky.
This guy went around literally condemning people to death because he was a greedy fuck, he doesn't deserve any sympathy and your argument is bogus.
A rational person who truly wanted to help people when he has a monopoly on fully developed drugs worth $1 a pill, bought from a company who already jacked up the price 10fold just one year earlier out of greed, doesn't jack the price up 700 times to make a point about the state of the industry. A rational person who wants to help sells the pills for $1 a pop, so as to maximize the good they do to individual suffering people.
That's utter nonsense. All you need is to feed a pig a single old leaf of lettuce, and now you've ensured that a single bacon rasher is worse for the environment than a single head of lettuce.
It never ceases to amaze me how American students are so much dumber than their European counterparts, even though they have so much better facilities and supposedly world class teachers. Sigh.
In a traffic jam, drivers can frustrated and bump in each other. I highly
doubt self driving car would do that.
True. In a bad traffic jam, how would frustrated people who have nothing to do at all but look out the windows react? Let's hoe we don't find out, because 88% of those people have guns.
No, you're being disingenuous. There's a world of difference between an organization keeping tons of data in a random hodgepodge of logs and temporary files with incomplete and missing data mixed, and effectively no ability to datamine except by painstaking manual ad hoc methods, against an organization that purposely builds state of the art data retrieval systems and AI algorithms to maximize the information gain from every collected tidbit.
It's the difference between some company keeping a basement full of old paperwork and other junk they just don't know what to do with, and a stasi-like multi-billion dollar spying bureaucracy that has up to date files on every citizen, dead or alive. In both cases they can find out if you paid off that bill in Augst 2011, but in the first case it takes three days for an underpaid intern to open thousands of folders and flip pages for that one piece of data, and in the other it takes a button press and you not only get the exact information, but you also get a two hundred page dossier with the names of your neighbours, children's teacher, the brand of food your cat eats and how many years you've been secretly gay.
It's the difference between some company for whom the data is not related to the business model, and some company for whom the data IS the business model.
It's not a cool idea, it's analog flawed: Using 3 words as a tuple to represent squares is a naive hashing scheme with all the problems this entails.
Unlike numerical coordinates, word coordinates in this scheme don't have a natural topology. So when you search table.book.lamp and you find that it's a square in Johannesburg, there's no way to know what the neighbouring square is called.
Worse, you might try to guess table.book.football, and that could be some place in Antarctica.
Suppose you communicate your 3-word coordinates with a Southern Texas accent to a Canadian. They might not hear the second word properly, and now you've sent them to Tokyo instead of New York.
Suppose you live in Idaho, and you need a package delivered but you type your address in with a typo in one of the words. Normally, with a proper human address like 1234 Pennsylvania Avenue, the truck driver first finds Pennsylvania Avenue, and then if 1234 turns out to be a typo
and the correct address is 1235, the delivery will be close by and the neighbours might bring the package over or tell the driver it's next door. With 3 random words the cost of redelivery skyrockets to a distance of half of the world's circumference, on average.
Furthermore, the uniform grid sizing causes more problems than it solves. The world is not full of uniformly sized, 3x3 apartments which can be uniquely identified with a single address. Using this system, many places of interest in the world receive hundreds or thousands of effectively randomly assigned addresses which all refer to the same logical unit of location in the human world. That makes communication less precise.
Suppose you want to meet your friend in the park. How do you communicate the idea of "park", which is an aggreegate of so many addresses? You might simplify and say let's meet at the bench in front of the fountain. What if that bench is occupied? You won't be able to meet at those coordinates. Now suppose there are hundreds of people near there, you might not be able to meet your friend unless you agree on some square coordinates that are much further away. For example, you might want to meet at a concert, but where in the hall would that be, and how would you know an hour before then while sending an SMS?
Suppose you want to describe the idea of a kid skateboarding in that same park. With a traditional human addressing system, you can say there's a kid in the park, and that is valid for the whole duration of the event . With this system you'd have to say there's a kid at bridge.pool.steak, fifty.moron.quantum, blah.blah.truck, etc. since the kid is moving around the whole time.
Similarly you wouldn't be able to warn people about disasters efficiently. Say there's a flooded street or a shooting, you'd have to say don't go to a.a.a, b.b.b, c.c.c,..., z.z.z etc By the time you've listed all the coordinates it might be too late, and never mind the world salad. It would also be a problem for television news presenters. Hash function sequences don't compress well.
There's a reason why the world uses many different coordinate systems simultaneously, inluding ultra precise ones like GPS. They all have their purposes and are the simplest systems for their purpose, respectively. This 3 word system however doesn't have good geographic locality because it's based on hashing, and suffers from complex aggregation issues. It tries to do too much and yet fails the analog test. It's badly flawed for human use.
It makes perfect sense if Blackberry's main customer is the US govt. In that case, they're saying that the US doesn't have to install third party spying apps to keep tabs on their employees, the hardware vendor will do this service for them. Sounds like a useful differentiating feature from the competition. Moreover, if the govt likes it, they might require all the contractors in the military industrial complex to use Blackberries too, purely for interoperability of course. If that happens, then Blackberry are saved.
The IS wankers think they are Hobbits. Now as you all know, the Hobbits win in the final battle against the armies of Mordor (that's us). But this is the 21st century, and those guys live on Internet Time, and they don't feel like sitting through 20 years of skirmishes and drama just to win in the end anyhow. So they're pressing the fast forward button and attacking everyone at once in the hope that we'll hurry up and gather the troops for the Final Battle, hopefully in March next year, so that they can win.
Because they're Hobbits. And the last battle HAS to be against EVERYONE.
Well put. Frankly, I don't know why he even wanted to campaign in the first place, since he made it clear he didn't want to be the prez. Wasting people's time much? He thought he was going to game the system in a trivial way to get his message heard, and now he's upset that his intention was transparently thwarted.
1) Tell your best friend (must be black!) that you bought a used TV on ebay from some guy, and would he mind going to the guy's house to pick it up for you?
Why not? Google is way overstaffed for what they accomplish. Suppose they reduce the workforce to about 1000 people, and leverage machine learning fully? Now that would be a company whose stock goes through the roof!
NO ads are reasonable. The purpose of an ad is to steal a small part of a person's attention while they are doing something else. It is evil, pure and simple.
(and BTW, humans have lived on earth for thousands of years, and 99% of that time ads didn't exist.)
Meh, that ship has sailed. All modern programs are full of GOTOs. They're just called EXCEPTIONs.
An exception is a nonlocal jump, where the destination is in some totally unrelated section of the codebase, typically written by some other programmer even. Everything that's said to be wrong with GOTOs is often said to be right by exception supporters.
And once again I'm just telling you that what you are saying is beside the point. The point isn't that sleep is vital, or that a certain flow rate measured in some individuals occurs, if that were so you'd be perfectly on topic, and great work btw.
The point is that the limits of healthy sleep durations is necessarily different among individuals trained to sleep more efficiently versus individuals who don't train their sleep patterns in any way. Even among random individuals the variance is large already.
If you're a scientist or engineer asking the specific question what is the minimum healthy sleep time for human beings, then you have to measure exceptional trained humans, not average patients. In particular, you should be devising training programs to beat the limits safely, just like athletes have training programs to beat the olympics.
That's not answering the question at all. Let me use a peeing analogy (w00t!).
TFA: Humans can probably get by on less than two piss breaks a day!
You: But [CITATION] says humans need to pee!
Me: What gives? Here's an analogy...
The point of the analogy is that individuals can train themselves to pee once a day at most, without ill effects. They can also train themselves to not have stage fright, so that what "normally" takes a few minutes can be done in much less time, with deliberate practice. Some people have stronger bladders, and you can trin yourself to have a strong bladder.
Humans practice things all the time. Athletes show what the human body is capable of, through deliberately programmed activities. When compared to the average joe, that can seem amazing.
There's no reason to believe that 8 hours sleep (say) is required, just because lots of people end up sleeping around that long. It's plausible that people who are "fit" in the sleep sense can do in 4 hours all that you or I could do in 8 because we're not sleeping fit, and it's plausible that people can train themselves to achieve more of their sleep activities in less time, without ill effects.
Most biological models of the body are one size fits all. At best, they represent an idealized average body, which is great, but doesn't answer what's *possible*. For that, we need to learn how to *train* people to sleep more efficiently.
You don't need markets for that, you're asking for regulation (which incidentally is exactly why handicapped people have more choices today than 50 years ago)
Markets are fundamentally unfair. They discriminate against those who can afford to pay a little less. (That is why one percenters love capitalism, because they know that as long as there's a market for some good, *they* can outbid everyone else if they badly want to have it - preference for the rich when they feel like it).
A low cost uniformly priced taxi service is much more fair to the population than Uber. The cost of the service should only cover the actual operating costs with a small overhead. This maximizes affordability for all. And access should not be on a market bidding system, it should be exclusively on a first click first served basis.
Slashdot Mirror
I don't really like their articles very much, but I'm lazy and I click on TFA's links simply because it's what slashdot serves for dinner.
You can use a machine learning face recognition system to identify "face geometry" features trivially, simply by using the output of the system with a postprocessing layer. The combined system is a system which collects "face geometry". As such, the black box nature of Shutterfly's system doesn't preclude the claimed illegal behaviour. Frankly, the only evidence that Shutterfly and Facebook don't break the law is their word, which suggests the lawsuit is entirely reasonable so that we may get to the bottom of this.
Such a patent pledge document is pure theatre, and should not be confused with open source licenses. The former relies on *the* issuing company behaving in a certain way, whereas the latter relies on *some* licensees behaving in a certain way.
Taking as a given that people change, their motivations, circumstances, relationships change over time, you can and should expect that both KA and open source licensees will change their minds and act contrary to the legal documents they agreed to initially. When that happens, the difference is stark: in the case of the patent pledge, everyone who built their products using said patent becomes an infringer, and the potential for damage and misuse is huge and worldwide (eg SCO). In the case of the open source license, those who misuse the code according to the license are merely stealing code that doesn't belong to them, and this has no effect on the other open source recipients, it's on a case by case basis.
In other words, a patent pledge breaks if one entity changes its mind, whereas an open source license doesn't break if one entity changes its mind. And expecting one entity to change their mind is a very low bar.
What KA should be doing is creating a document, similar to the BSD license, which automatically licenses their patents to anyone for free and forever. That will not happen, because the true motivation of buying dubious patents is to build a war chest that can be used to threaten others, and there would be absolutely no point in paying for a patent if it had no threat value.
I don't think that calling projects started by women "segregated" is helpful. You are looking at the state of open source through racial conflict tinted glasses. This is not the same problem as black/white integration at all. What is open source about? It's about scratching your own itch. If some mainstream windows like system doesn't work for you, you don't complain and you don't propose to force windows to do things your way. No, you fork a project and do your own thing. If other people like where you are going with this, they'll join and the project grows. If not, there's no harm in having a project of one. Either way, nobody is being forced to do anything, people CHOOSE freely to do what works for them at the time. THAT is the open source way, and that is what is being suggested to you IMHO. If someone wants to fork an open source project with the special feature that cussing and disrespecting women members is not acceptable, they should do exactly that. Fork and run with the idea. Make the rules clear on the mailing list and enforce them. THAT is the open source way. And people will come, and people will choose, and the new projects will grow and take over mindshare, like Linux and GNU grew, and took over mind share, and 30 years later we have a thriving ecosystem. Linux didn't grow fully formed overnight. Neither will your kinder, gentler open source community. So relax, don't tell the community what it should do, don't be impatient. Start by forking an interesting project, experiment with social rules, and see if the idea works out. If you truly have a good idea, people will come and help out. If not, keep going.
That doesn't work. What if Khan gets a new management and/or new owners? They can change the rules and repudiate all those promises, no matter what. For example, Google used to be trustworthy in the early days, now they are an evil spying organisation who takes so much heat that even the founders are distancing themselves with a nondescript holding company called alphabet. If it can happen to Google, who had so much promise in 1999, then you betcha it will happen to Khan once the pressure of economics start to bite them. Trusting the intentions of companies is simply a woeful approach to economic decision making.
Patents should be granted and immediately sold exclusively to the government, who can then open them up for everyone to use. A good example of how this works is the photography patent which was sold by Louis Daguerre to the French government, who opened the technology to the world. In this way we fix all the greatest problems with patents in one swoop. No more patent portfolios that create barriers to entry for innovative new startups, no more frivolous patent lawsuits about square looking smartphones, no more chilling effects based on the threat of being sued, no more protection money (aka licensing/royalties) rackets, and if the government must buy all potential patents then there is a balance between income (from patent examination fees) and expenditure (from buying the granted patents) so that the worst excesses of the current system are automatically held in check.
When a guy kills a random person just to point out that the police can't be everywhere at the same time, he's not offering a valuable insight into society to change it, he's just a murderer being cheeky.
This guy went around literally condemning people to death because he was a greedy fuck, he doesn't deserve any sympathy and your argument is bogus.
A rational person who truly wanted to help people when he has a monopoly on fully developed drugs worth $1 a pill, bought from a company who already jacked up the price 10fold just one year earlier out of greed, doesn't jack the price up 700 times to make a point about the state of the industry. A rational person who wants to help sells the pills for $1 a pop, so as to maximize the good they do to individual suffering people.
It never ceases to amaze me how American students are so much dumber than their European counterparts, even though they have so much better facilities and supposedly world class teachers. Sigh.
True. In a bad traffic jam, how would frustrated people who have nothing to do at all but look out the windows react? Let's hoe we don't find out, because 88% of those people have guns.
It's the difference between some company keeping a basement full of old paperwork and other junk they just don't know what to do with, and a stasi-like multi-billion dollar spying bureaucracy that has up to date files on every citizen, dead or alive. In both cases they can find out if you paid off that bill in Augst 2011, but in the first case it takes three days for an underpaid intern to open thousands of folders and flip pages for that one piece of data, and in the other it takes a button press and you not only get the exact information, but you also get a two hundred page dossier with the names of your neighbours, children's teacher, the brand of food your cat eats and how many years you've been secretly gay.
It's the difference between some company for whom the data is not related to the business model, and some company for whom the data IS the business model.
ISP != Google
Unlike numerical coordinates, word coordinates in this scheme don't have a natural topology. So when you search table.book.lamp and you find that it's a square in Johannesburg, there's no way to know what the neighbouring square is called. Worse, you might try to guess table.book.football, and that could be some place in Antarctica.
Suppose you communicate your 3-word coordinates with a Southern Texas accent to a Canadian. They might not hear the second word properly, and now you've sent them to Tokyo instead of New York.
Suppose you live in Idaho, and you need a package delivered but you type your address in with a typo in one of the words. Normally, with a proper human address like 1234 Pennsylvania Avenue, the truck driver first finds Pennsylvania Avenue, and then if 1234 turns out to be a typo and the correct address is 1235, the delivery will be close by and the neighbours might bring the package over or tell the driver it's next door. With 3 random words the cost of redelivery skyrockets to a distance of half of the world's circumference, on average.
Furthermore, the uniform grid sizing causes more problems than it solves. The world is not full of uniformly sized, 3x3 apartments which can be uniquely identified with a single address. Using this system, many places of interest in the world receive hundreds or thousands of effectively randomly assigned addresses which all refer to the same logical unit of location in the human world. That makes communication less precise.
Suppose you want to meet your friend in the park. How do you communicate the idea of "park", which is an aggreegate of so many addresses? You might simplify and say let's meet at the bench in front of the fountain. What if that bench is occupied? You won't be able to meet at those coordinates. Now suppose there are hundreds of people near there, you might not be able to meet your friend unless you agree on some square coordinates that are much further away. For example, you might want to meet at a concert, but where in the hall would that be, and how would you know an hour before then while sending an SMS?
Suppose you want to describe the idea of a kid skateboarding in that same park. With a traditional human addressing system, you can say there's a kid in the park, and that is valid for the whole duration of the event . With this system you'd have to say there's a kid at bridge.pool.steak, fifty.moron.quantum, blah.blah.truck, etc. since the kid is moving around the whole time.
Similarly you wouldn't be able to warn people about disasters efficiently. Say there's a flooded street or a shooting, you'd have to say don't go to a.a.a, b.b.b, c.c.c, ..., z.z.z etc By the time you've listed all the coordinates it might be too late, and never mind the world salad. It would also be a problem for television news presenters. Hash function sequences don't compress well.
There's a reason why the world uses many different coordinate systems simultaneously, inluding ultra precise ones like GPS. They all have their purposes and are the simplest systems for their purpose, respectively. This 3 word system however doesn't have good geographic locality because it's based on hashing, and suffers from complex aggregation issues. It tries to do too much and yet fails the analog test. It's badly flawed for human use.
Looks like it's time to short Mattel stock.
It makes perfect sense if Blackberry's main customer is the US govt. In that case, they're saying that the US doesn't have to install third party spying apps to keep tabs on their employees, the hardware vendor will do this service for them. Sounds like a useful differentiating feature from the competition. Moreover, if the govt likes it, they might require all the contractors in the military industrial complex to use Blackberries too, purely for interoperability of course. If that happens, then Blackberry are saved.
On the contrary. I wish them success, so that they can build and sell a beacon I can put in my backyard to fence off my home.
The IS wankers think they are Hobbits. Now as you all know, the Hobbits win in the final battle against the armies of Mordor (that's us). But this is the 21st century, and those guys live on Internet Time, and they don't feel like sitting through 20 years of skirmishes and drama just to win in the end anyhow. So they're pressing the fast forward button and attacking everyone at once in the hope that we'll hurry up and gather the troops for the Final Battle, hopefully in March next year, so that they can win.
Because they're Hobbits. And the last battle HAS to be against EVERYONE.
Well put. Frankly, I don't know why he even wanted to campaign in the first place, since he made it clear he didn't want to be the prez. Wasting people's time much? He thought he was going to game the system in a trivial way to get his message heard, and now he's upset that his intention was transparently thwarted.
1) Tell your best friend (must be black!) that you bought a used TV on ebay from some guy, and would he mind going to the guy's house to pick it up for you?
2) Wait for your friend to return.
Why not? Google is way overstaffed for what they accomplish. Suppose they reduce the workforce to about 1000 people, and leverage machine learning fully? Now that would be a company whose stock goes through the roof!
It's a self inflicted predicament. They shouldn't have gone into bed with evil Doubleclick.
(and BTW, humans have lived on earth for thousands of years, and 99% of that time ads didn't exist.)
An exception is a nonlocal jump, where the destination is in some totally unrelated section of the codebase, typically written by some other programmer even. Everything that's said to be wrong with GOTOs is often said to be right by exception supporters.
The point is that the limits of healthy sleep durations is necessarily different among individuals trained to sleep more efficiently versus individuals who don't train their sleep patterns in any way. Even among random individuals the variance is large already.
If you're a scientist or engineer asking the specific question what is the minimum healthy sleep time for human beings, then you have to measure exceptional trained humans, not average patients. In particular, you should be devising training programs to beat the limits safely, just like athletes have training programs to beat the olympics.
TFA: Humans can probably get by on less than two piss breaks a day!
You: But [CITATION] says humans need to pee!
Me: What gives? Here's an analogy...
The point of the analogy is that individuals can train themselves to pee once a day at most, without ill effects. They can also train themselves to not have stage fright, so that what "normally" takes a few minutes can be done in much less time, with deliberate practice. Some people have stronger bladders, and you can trin yourself to have a strong bladder.
Humans practice things all the time. Athletes show what the human body is capable of, through deliberately programmed activities. When compared to the average joe, that can seem amazing.
There's no reason to believe that 8 hours sleep (say) is required, just because lots of people end up sleeping around that long. It's plausible that people who are "fit" in the sleep sense can do in 4 hours all that you or I could do in 8 because we're not sleeping fit, and it's plausible that people can train themselves to achieve more of their sleep activities in less time, without ill effects.
Most biological models of the body are one size fits all. At best, they represent an idealized average body, which is great, but doesn't answer what's *possible*. For that, we need to learn how to *train* people to sleep more efficiently.
You don't need markets for that, you're asking for regulation (which incidentally is exactly why handicapped people have more choices today than 50 years ago)
A low cost uniformly priced taxi service is much more fair to the population than Uber. The cost of the service should only cover the actual operating costs with a small overhead. This maximizes affordability for all. And access should not be on a market bidding system, it should be exclusively on a first click first served basis.