This is the same reality faced by Bill Gates, Bezos, Musk, Buffet, etc.
Yes and no.
Founders cashing out does temporarily lower the stock price, but it's usually not too bad unless investors believe they're bailing because they know something (the knowledge can't be specific, current non-public data; that would be insider trading, but it can be application of the same level of insight that led them to become successful in the first place). This is true even when the Founders own significantly more than 1% of the stock. Divestiture of a mere 1% would rarely cause a big downward movement, even if it hit pretty fast.
In the case of BTC, though, because the real value of a BTC as an exchange medium is pretty weak (due to the scaling issues), nearly all of the BTC value is bubble-driven. In that sort of situation, a 1% sell-off could easily trigger a catastrophic collapse.
What people came up with to encourage the sharing of trade secrets are patents. In exchange for charging people a license fee for a period of time, you show them the better way you figured out how to do something. Current implementation of the patent concept in law also has issues, particularly for software patents.
True. I think the misapplication of copyright to software is much more severe than the misapplication of patents. IMO, the biggest problem with software patents isn't the concept of patenting software designs but the granting of patents for obvious software designs. If the PTO refused to grant for things that are obvious to one skilled in the art, it wouldn't be so bad.
BTW, since you brought up the subject, I'll share my notional "patent system test". You know that your patent system is working correctly when working engineers regularly search the patent database in search of solutions to their problems, because it's more cost-effective to license a solution than to devise one. That would prove that the sharing of ideas is actually being encouraged by the system. In the software world, of course, every company I've ever worked for cautions software engineers never to look at the patent database, because odds are they'll find something they've independently reinvented, and knowing about it makes the company liable for treble damages. That's fairly compelling evidence that the software patent system is broken.
Of course, if you're a startup stepping on the toes of a juggernaut, clean-room might still be advisable. Abstraction-Filtration-Comparison is great, but you have to go to court to do it. Clean-room tends to make the issue so clear-cut that you don't even get to court. Or, if you do, the only avenue of attack open to the plaintiff is to try to prove that you didn't implement the process correctly. Only if they succeed at that do you have to bother with complex AFC arguments.
Okay but right now legal reverse engineering is allowed on the grounds of proven clean-room reimplementation, right? Like say with the ReactOS people being unable to accept patches from former Microsoft employees who worked on Windows. So wouldn't freely-available source muddy those waters?
Wouldn't change it at all. In order to do clean-room reimplementation you have to separate the people who do the reverse engineering from the people who do the reimplementation. You have one group analyze the existing code and create a specification document. The specification contains the ideas from the original code, but none of the original code, nor any mechanical transformation of the original code, so it does not contain any of the original expression (copyright protects expressions, not ideas). Then you pass the specification document to a new group of people who implement it. Because they've had no exposure to the original code they cannot possibly copy any of the expression. Any identical portions of the result are purely coincidence, or necessity.
Note that nothing in the above says anything about the form of the code from which the specification was produced. It doesn't matter. Source availability would make clean-room reimplementation dramatically easier, by making the job of the analysts who write the specification dramatically easier.
It's also worth noting that clean-room reimplementation isn't legally necessary. You can actually have the same people do both the analysis and the implementation. The problem is that it becomes hard to prove that identical portions were not copied. You'd have to demonstrate that there are only a limited number of ways to implement the relevant bits, and therefore that the probability of coincidental duplication is high. It can be done, and has been done (e.g. in Oracle v Google, where the judge actually learned enough about Java programming to be able to make that determination himself). But clean-room, done correctly, is ironclad proof of the absence of copying.
The notion of clean-room implementation was created when Compaq decided to reverse engineer the IBM PC BIOS. Compaq was a small startup and IBM was a behemoth well-known for it's willingness and ability to litigate. IBM's legal staff hadn't yet acquired the moniker "The Nazgul" (that came during the SCO fiasco), but Compaq would have understood and agreed with the label. So, Compaq's lawyers devised the clean-room process, and Compaq carefully followed it and documented every step, so that IBM would have absolutely no hope of proving copyright infringement. It worked. IBM's lawyers looked at Compaq's process and documentation and didn't bother trying to sue.
Copyright is not the only possible legal regime. It's just the one we have now. A legal structure supporting openness could exist side-by-side with proprietary copyright.
I'd argue also that we have misapplied copyright to software. The fundamental goals of copyright are not met by offering protection for binary-only software. Just as it's possible to take the ideas expressed in a book and apply them in your own work, it should be possible to read software you purchase and remix the ideas in your own work... but this is impossible (or at least impractical) if you receive only a binary. Binary-only distribution means that you can publish your work while simultaneously keeping most of the ideas it contains secret, thus not enabling progress in the useful art and science of software engineering.
A more correct application of the principles of copyright to software would be to extend copyright protection only to software distributed in source form. This would make nearly all commercial software "source-available", which obviously isn't the same as "open source", much less the same as "free software", but would make the world of software considerably more open. Closed source software would still be possible, but would not have copyright protection so other means would have to be used to protect it, such as trade secret and contract law. But that would be cumbersome, so most software would have source available.
Without copyright there is no such thing as the GPL. There is no such thing as copyleft,
Without copyright, there would be no need for copyleft. Somewhat counterintuitively, it's GPL (v2) rather than BSD/MIT that emulates a world without copyright better: we'd have decompilers.
IMO, there's a better solution than a world without copyright: A world that grants copyright protection only when the expression is published. This is trivially the case in all other areas of copyright: You can't publish a book without letting people read the words that it's made of, thus allowing them to learn and therefore remix your ideas in their own expressions. Same for music, same for everything except software. Only in software is it possible to publish your ideas while simultaneously keeping them hidden.
We made a huge mistake in allowing software distributed in binary-only form to have copyright protection. It doesn't achieve the goal of modern copyright[*], which is to encourage the dissemination of ideas by allowing temporary ownership of the expression of those ideas. (Note that we've also broken the "temporary" part, but that's a separate issue). Binary-only software distribution means that it's impractical to read code and learn from it. The only part of the ideas embodied in the software that get disseminated with it are those observable from outside.
A more appropriate way to apply copyright to software would have been to provide it only to programs distributed in source form. This wouldn't seriously inhibit commercial software. Indeed in the early days of commercial software, licenses usually included access to the source code. This approach wouldn't imply that licensees had the permission to modify or compile the code (that would be production of a derived work), but most of the early licenses I mention did allow modification and recompiling, and I think it would be the norm, because it just makes sense to allow licensees to fix bugs that they discover, or to make enhancements that they need.
For code that really does need to be kept secret, there are other options under the law, including trade secret law. It would be more cumbersome to apply, since every recipient would have to sign a contract including an NDA -- no shrink-wrapped EULAs possible. But that seems fine to me, since it's pretty rare that code actually needs to be kept secret. SaaS would be another option for keeping code secret.
But under such a regime, most commercial software would be source-available, and open source and copyleft licenses would work as well. You could take open source-licensed software and add proprietary bits and redistribute, with the open source license on the original code and your proprietary bits under your own license. Your licensees would have full source code, but would only be free to redistribute the open source-licensed bits. Your parts would be subject to the restrictions in your license. But copyleft licenses would require that all code redistributed with the copylefted code fall under the copyleft license, meaning that whoever you give the combined code to has permission to modify and redistribute all of it.
[*] By "modern copyright" I'm essentially referring to the notion of copyright espoused by the US Constitution, as opposed to the earlier form embodied in the Statute of Anne. The earlier form was mostly about censorship, enabling the crown to decide what could or could not be published.
the only rational choice is to think about how to manage it.
OK, so how does one "manage" a technology that nobody really understands (and the only realistic shot at changing that currently on the horizon seems to be "let the algorithm explain why it does what it does")?
Indeed. That's a very, very hard problem. Roughly half of the book is about it. It's an even harder problem than you think it is, because not only do we need good solutions, we need good solutions that can be implemented even though we don't have the ability to force everyone researching AI to implement them. So we either need solutions that can be implemented over the objections of some groups, or we need solutions that everyone actually wants to implement.
My core point, though, is that "let's just not build AI" is not an option, because there's no way to prevent people from doing it.
The only way AI could have this type of algorithm is if we specifically program it to do so.
Or if it evolved to do so. Keep in mind that our current AI methods are much like accelerated and guided evolution, where we keep the fittest version of the system and discard the rest. It's not a stretch to see how a survival motivation might arise there, in pretty much exactly the way ours did. If there's any way the fitness function could favor a will to survive, the will will emerge.
In addition to that possibility, consider that if the AI does have some goal (almost regardless of what it is), it will logically conclude that its survival is a prerequisite to achieving that goal, and that therefore it must survive.
There are many other ways that an AI might gain survival as a goal, without in any way presuming human-like motivations. I highly recommend that you read Nick Bostrom's "Superintelligence". It explores these issues carefully, and thoroughly. It also discusses a lot of potential countermeasures.
If that's true, clearly the machines are already in charge and thus it doesn't matter what we do. If the humans are still in charge, they can decide to stop.
No, it's because humans are in charge that we will not decide to stop.
Oh, we theoretically could decide to stop, but it's abundantly clear that isn't going to happen. Some individuals or groups may decide to stop, but AI is so useful that someone will continue, convinced that the competitive advantage is too great and the risk manageable. That will in turn motivate others, and so on.
It's like nuclear weapons. It would have been rational for the world to look at the devastated city of Hiroshima and collectively say "Nope, this is a bad idea. We shouldn't make these," and everyone would stop. Instead 15 years later we had Tsar Bomba, 1600 times more powerful than that first one, and thousands of the things atop missiles, on airplanes and in submarines. Even now that we have a system in place designed to stop the proliferation of nuclear weapons, there are still countries trying to acquire them.
We are what we are, and because of what we are, we will continue. That being the case, the only rational choice is to think about how to manage it.
That is why we still write "you are" rather than "you is".
Except in African American Vernacular English.
I'm not sure that is really an inconsistency, though. Had we not lost the explicit second person singular we would be saying "thou art" not "thou is". "Are" has always been the plural present conjugation of "to be", but we had three different singular forms: "I am", "thou art" and "you / ye are". When "thou" became first too intimate to use generally, and then archaic, and we shifted to using "you" for both singular and plural cases we needed to pick a conjugation. You're arguing that we should have adopted the third person singular conjugation, but I don't see any reason that's more appropriate than using the second person plural conjugation, which had the advantage of sounding "right", since "you are" was already a normal, correct phrase.
Further, I've read speculation that the shift was driven by court language, which lends weight to the choice of "are". A royal "you" to correspond to the royal "we" (though one would rarely say "you" to a monarch). It's easy to see how that would extend into other spheres, such as business, where you might politely address an individual as plural "you" to recognize their status as representative of an organization. From there it's easy to see how over time "thou" would be relegated to purely informal contexts (especially in extremely formal England), and eventually fade away entirely, particularly as class distinctions became fuzzier.
So, I think "you are" actually makes more sense than "you is" or "you am", the two singular options available. Other than "you art", of course.
My phone doesn't work in my basement, and I live in my basement. No, not my mom's basement, my wonderful, fully furnished basement. The cats live upstairs...
So, you live in your cats' basement? Is that supposed to be better?
Not accessible to the user, meaning no nice UI to display and manage it all.
There is... Settings / Wireless & Network / Data Usage
Which obviously has a tiny subset of the UI functionality described in the article, and has the serious disadvantage that it can only be added or updated in a system update. The rationale for putting this in an app is very clear.
Not accessible? It's a background app made by Google. It's their OS, they have access to everything.
Not accessible to the user, meaning no nice UI to display and manage it all. Clearly the app can't add any fundamental OS capabilities; everything needed to track and control usage has to already be in the system.
Oh, and obviously it's not just "three big fucking counters", since it gives you very fine-grained information about usage. And lets you control it.
I like Google, but I think they've lost the ability to execute. I find their products are increasingly half-baked
Actually, I think it's the opposite. Google's products are more polished at launch than they ever used to be. From the outset, Google's modus operandi has always been to launch early and incomplete, and then iterate incrementally. Remember when every new launch was explicitly tagged as "beta"? Expectations have changed, though, and now people expect Google stuff to work perfectly from the beginning. I think Google's execution has actually gotten much better (though there are some glaring counterexamples), but expectations have risen even more.
and getting flakier over time (e.g. Chromecast).
Chromecast is getting flakier? I haven't noticed that.
Other companies do this, just not in so many words. Aside from the nudging and winking policies, some things are done under overt, yet bogus reasons. My company now deletes virtually all email after 3 months, to better manage storage resources and adhere to best practices for retention.
Three months is rather short. Most companies have a one year retention policy. This isn't for any nefarious reasons, and it's not to save disk space. It's to limit the information available to discovery during lawsuits, less to make the information unavailable than to eliminate the need to pay lawyers lots and lots of money to comb through it to comply with discovery requests. Since any large company -- no matter how saintly -- is always involved in multiple lawsuits, this saves huge amounts of money.
The truth is, I use bitcoin everyday to buy coffee, gas, pay my bills, take my dates out, and buy every day items such as groceries and electronics.
You're not a liar, you're just wrong. Or perhaps you're being lied to.
You're using dollars every day, lent to you by a bitcoin wallet provider against bitcoins held as collateral by the provider. This is not a bit different from the debit card I have that draws against my stock brokerage margin account, except that your provider is almost certainly more aggressive about doing "margin calls" than my brokerage.
Well, one other difference is that you're currently in fat city, seeing your BTC appreciate in value far faster than my stocks do. Of course, there will come a time when that reverses, and it'll reverse much, much harder and faster than any stock market crash, because stocks have, at root, something of real tangible value. The only real value of BTC is its use as a currency which, as others have pointed out, has simply broken down as it attempted to scale. Were BTC a better cryptocurrency with better scalability characteristics, then it would be the case that BTC is just for criminals.
They don't know, and there is a valid reason for it. The market is irrational. The market is as much about emotion as it is about fundamentals, maybe even more so.
There *are* no fundamentals. The transaction rates/costs make it unusable as a currency for more than a fairly small number of people, and it has no other value than as a currency.
VHF, not VHS, and you have them reversed. Very High Frequency channels were 2-13, UHF channels were 14-83. (I'll admit I had to look the upper limit of UHF up, though I knew it was 80-something).
I'm not a fan of Uber as a company. They've done a lot of shady crap. But this sounds like it could also be normal corporate "speaking with care" training. Every company I've ever worked for has told me to be cautious with communicating about legally-sensitive topics in writing (including email and chat systems that archive conversations). Said training usually includes information about how to specify that written communication is attorney-client privileged, too, which makes it non-discoverable in most legal proceedings, plus a recommendation that if you're in doubt, you should hold your conversation in person or over the phone (or video conference).
This is just normal stuff. Sure, it could be used to avoid leaving documentary evidence of illegal dealings, but that's not its intent. Its intent is to avoid generating large amounts of documentation that has to be reviewed in discovery, and which could be exploited by opposing counsel by mischaracterizing it or taking it out of context. For individual employees, it also tends to keep them out of the line of fire of subpoenas. If you mentioned a topic in discoverable communications, you're likely to get subpoenaed to give a deposition.
It should be noted that if you're actually doing something illegal, and if your attorney is unethical enough to advise you on how to get away with it, the training shouldn't tell you to use any chat apps, or phone calls, etc., because there's always a chance that an investigation is already under way and that the communications medium is tapped. If you're committing crime, "speaking with care" recommendations are not good enough.
To what extent is this SOC comparable to Intel's management engine? Isn't Google basically putting a second computer in my computer? With all the risks that entails?
There are many computers inside your computer. The Intel ME is risky because it can take very low level control of the device, update software, etc. The Vision Core is just a graphics coprocessor, not something that can take over control of the device.
Of course, that would be unnecessary, since unless you've unlocked the bootloader and installed custom software, Google can already update your software, etc. Or if you have a Samsung device, Samsung can, etc.
Slashdot Mirror
This is the same reality faced by Bill Gates, Bezos, Musk, Buffet, etc.
Yes and no.
Founders cashing out does temporarily lower the stock price, but it's usually not too bad unless investors believe they're bailing because they know something (the knowledge can't be specific, current non-public data; that would be insider trading, but it can be application of the same level of insight that led them to become successful in the first place). This is true even when the Founders own significantly more than 1% of the stock. Divestiture of a mere 1% would rarely cause a big downward movement, even if it hit pretty fast.
In the case of BTC, though, because the real value of a BTC as an exchange medium is pretty weak (due to the scaling issues), nearly all of the BTC value is bubble-driven. In that sort of situation, a 1% sell-off could easily trigger a catastrophic collapse.
What people came up with to encourage the sharing of trade secrets are patents. In exchange for charging people a license fee for a period of time, you show them the better way you figured out how to do something. Current implementation of the patent concept in law also has issues, particularly for software patents.
True. I think the misapplication of copyright to software is much more severe than the misapplication of patents. IMO, the biggest problem with software patents isn't the concept of patenting software designs but the granting of patents for obvious software designs. If the PTO refused to grant for things that are obvious to one skilled in the art, it wouldn't be so bad.
BTW, since you brought up the subject, I'll share my notional "patent system test". You know that your patent system is working correctly when working engineers regularly search the patent database in search of solutions to their problems, because it's more cost-effective to license a solution than to devise one. That would prove that the sharing of ideas is actually being encouraged by the system. In the software world, of course, every company I've ever worked for cautions software engineers never to look at the patent database, because odds are they'll find something they've independently reinvented, and knowing about it makes the company liable for treble damages. That's fairly compelling evidence that the software patent system is broken.
I don't think it's nearly so bad.
That was part of my point :-)
Of course, if you're a startup stepping on the toes of a juggernaut, clean-room might still be advisable. Abstraction-Filtration-Comparison is great, but you have to go to court to do it. Clean-room tends to make the issue so clear-cut that you don't even get to court. Or, if you do, the only avenue of attack open to the plaintiff is to try to prove that you didn't implement the process correctly. Only if they succeed at that do you have to bother with complex AFC arguments.
Okay but right now legal reverse engineering is allowed on the grounds of proven clean-room reimplementation, right? Like say with the ReactOS people being unable to accept patches from former Microsoft employees who worked on Windows. So wouldn't freely-available source muddy those waters?
Wouldn't change it at all. In order to do clean-room reimplementation you have to separate the people who do the reverse engineering from the people who do the reimplementation. You have one group analyze the existing code and create a specification document. The specification contains the ideas from the original code, but none of the original code, nor any mechanical transformation of the original code, so it does not contain any of the original expression (copyright protects expressions, not ideas). Then you pass the specification document to a new group of people who implement it. Because they've had no exposure to the original code they cannot possibly copy any of the expression. Any identical portions of the result are purely coincidence, or necessity.
Note that nothing in the above says anything about the form of the code from which the specification was produced. It doesn't matter. Source availability would make clean-room reimplementation dramatically easier, by making the job of the analysts who write the specification dramatically easier.
It's also worth noting that clean-room reimplementation isn't legally necessary. You can actually have the same people do both the analysis and the implementation. The problem is that it becomes hard to prove that identical portions were not copied. You'd have to demonstrate that there are only a limited number of ways to implement the relevant bits, and therefore that the probability of coincidental duplication is high. It can be done, and has been done (e.g. in Oracle v Google, where the judge actually learned enough about Java programming to be able to make that determination himself). But clean-room, done correctly, is ironclad proof of the absence of copying.
The notion of clean-room implementation was created when Compaq decided to reverse engineer the IBM PC BIOS. Compaq was a small startup and IBM was a behemoth well-known for it's willingness and ability to litigate. IBM's legal staff hadn't yet acquired the moniker "The Nazgul" (that came during the SCO fiasco), but Compaq would have understood and agreed with the label. So, Compaq's lawyers devised the clean-room process, and Compaq carefully followed it and documented every step, so that IBM would have absolutely no hope of proving copyright infringement. It worked. IBM's lawyers looked at Compaq's process and documentation and didn't bother trying to sue.
Not even to mention the top speed.
And the acceleration numbers! This is so far beyond ludicrous mode, there is no word for it.
Copyright is not the only possible legal regime. It's just the one we have now. A legal structure supporting openness could exist side-by-side with proprietary copyright.
I'd argue also that we have misapplied copyright to software. The fundamental goals of copyright are not met by offering protection for binary-only software. Just as it's possible to take the ideas expressed in a book and apply them in your own work, it should be possible to read software you purchase and remix the ideas in your own work... but this is impossible (or at least impractical) if you receive only a binary. Binary-only distribution means that you can publish your work while simultaneously keeping most of the ideas it contains secret, thus not enabling progress in the useful art and science of software engineering.
A more correct application of the principles of copyright to software would be to extend copyright protection only to software distributed in source form. This would make nearly all commercial software "source-available", which obviously isn't the same as "open source", much less the same as "free software", but would make the world of software considerably more open. Closed source software would still be possible, but would not have copyright protection so other means would have to be used to protect it, such as trade secret and contract law. But that would be cumbersome, so most software would have source available.
Without copyright there is no such thing as the GPL. There is no such thing as copyleft,
Without copyright, there would be no need for copyleft. Somewhat counterintuitively, it's GPL (v2) rather than BSD/MIT that emulates a world without copyright better: we'd have decompilers.
IMO, there's a better solution than a world without copyright: A world that grants copyright protection only when the expression is published. This is trivially the case in all other areas of copyright: You can't publish a book without letting people read the words that it's made of, thus allowing them to learn and therefore remix your ideas in their own expressions. Same for music, same for everything except software. Only in software is it possible to publish your ideas while simultaneously keeping them hidden.
We made a huge mistake in allowing software distributed in binary-only form to have copyright protection. It doesn't achieve the goal of modern copyright[*], which is to encourage the dissemination of ideas by allowing temporary ownership of the expression of those ideas. (Note that we've also broken the "temporary" part, but that's a separate issue). Binary-only software distribution means that it's impractical to read code and learn from it. The only part of the ideas embodied in the software that get disseminated with it are those observable from outside.
A more appropriate way to apply copyright to software would have been to provide it only to programs distributed in source form. This wouldn't seriously inhibit commercial software. Indeed in the early days of commercial software, licenses usually included access to the source code. This approach wouldn't imply that licensees had the permission to modify or compile the code (that would be production of a derived work), but most of the early licenses I mention did allow modification and recompiling, and I think it would be the norm, because it just makes sense to allow licensees to fix bugs that they discover, or to make enhancements that they need.
For code that really does need to be kept secret, there are other options under the law, including trade secret law. It would be more cumbersome to apply, since every recipient would have to sign a contract including an NDA -- no shrink-wrapped EULAs possible. But that seems fine to me, since it's pretty rare that code actually needs to be kept secret. SaaS would be another option for keeping code secret.
But under such a regime, most commercial software would be source-available, and open source and copyleft licenses would work as well. You could take open source-licensed software and add proprietary bits and redistribute, with the open source license on the original code and your proprietary bits under your own license. Your licensees would have full source code, but would only be free to redistribute the open source-licensed bits. Your parts would be subject to the restrictions in your license. But copyleft licenses would require that all code redistributed with the copylefted code fall under the copyleft license, meaning that whoever you give the combined code to has permission to modify and redistribute all of it.
[*] By "modern copyright" I'm essentially referring to the notion of copyright espoused by the US Constitution, as opposed to the earlier form embodied in the Statute of Anne. The earlier form was mostly about censorship, enabling the crown to decide what could or could not be published.
the only rational choice is to think about how to manage it.
OK, so how does one "manage" a technology that nobody really understands (and the only realistic shot at changing that currently on the horizon seems to be "let the algorithm explain why it does what it does")?
Indeed. That's a very, very hard problem. Roughly half of the book is about it. It's an even harder problem than you think it is, because not only do we need good solutions, we need good solutions that can be implemented even though we don't have the ability to force everyone researching AI to implement them. So we either need solutions that can be implemented over the objections of some groups, or we need solutions that everyone actually wants to implement.
My core point, though, is that "let's just not build AI" is not an option, because there's no way to prevent people from doing it.
I found this one, with my wife's aunt's correct name and address: https://www.fcc.gov/ecfs/filin...
Only problem is that she died in 2006.
The only way AI could have this type of algorithm is if we specifically program it to do so.
Or if it evolved to do so. Keep in mind that our current AI methods are much like accelerated and guided evolution, where we keep the fittest version of the system and discard the rest. It's not a stretch to see how a survival motivation might arise there, in pretty much exactly the way ours did. If there's any way the fitness function could favor a will to survive, the will will emerge.
In addition to that possibility, consider that if the AI does have some goal (almost regardless of what it is), it will logically conclude that its survival is a prerequisite to achieving that goal, and that therefore it must survive.
There are many other ways that an AI might gain survival as a goal, without in any way presuming human-like motivations. I highly recommend that you read Nick Bostrom's "Superintelligence". It explores these issues carefully, and thoroughly. It also discusses a lot of potential countermeasures.
If that's true, clearly the machines are already in charge and thus it doesn't matter what we do. If the humans are still in charge, they can decide to stop.
No, it's because humans are in charge that we will not decide to stop.
Oh, we theoretically could decide to stop, but it's abundantly clear that isn't going to happen. Some individuals or groups may decide to stop, but AI is so useful that someone will continue, convinced that the competitive advantage is too great and the risk manageable. That will in turn motivate others, and so on.
It's like nuclear weapons. It would have been rational for the world to look at the devastated city of Hiroshima and collectively say "Nope, this is a bad idea. We shouldn't make these," and everyone would stop. Instead 15 years later we had Tsar Bomba, 1600 times more powerful than that first one, and thousands of the things atop missiles, on airplanes and in submarines. Even now that we have a system in place designed to stop the proliferation of nuclear weapons, there are still countries trying to acquire them.
We are what we are, and because of what we are, we will continue. That being the case, the only rational choice is to think about how to manage it.
That is why we still write "you are" rather than "you is".
Except in African American Vernacular English.
I'm not sure that is really an inconsistency, though. Had we not lost the explicit second person singular we would be saying "thou art" not "thou is". "Are" has always been the plural present conjugation of "to be", but we had three different singular forms: "I am", "thou art" and "you / ye are". When "thou" became first too intimate to use generally, and then archaic, and we shifted to using "you" for both singular and plural cases we needed to pick a conjugation. You're arguing that we should have adopted the third person singular conjugation, but I don't see any reason that's more appropriate than using the second person plural conjugation, which had the advantage of sounding "right", since "you are" was already a normal, correct phrase.
Further, I've read speculation that the shift was driven by court language, which lends weight to the choice of "are". A royal "you" to correspond to the royal "we" (though one would rarely say "you" to a monarch). It's easy to see how that would extend into other spheres, such as business, where you might politely address an individual as plural "you" to recognize their status as representative of an organization. From there it's easy to see how over time "thou" would be relegated to purely informal contexts (especially in extremely formal England), and eventually fade away entirely, particularly as class distinctions became fuzzier.
So, I think "you are" actually makes more sense than "you is" or "you am", the two singular options available. Other than "you art", of course.
My phone doesn't work in my basement, and I live in my basement. No, not my mom's basement, my wonderful, fully furnished basement. The cats live upstairs...
So, you live in your cats' basement? Is that supposed to be better?
Not accessible to the user, meaning no nice UI to display and manage it all.
There is... Settings / Wireless & Network / Data Usage
Which obviously has a tiny subset of the UI functionality described in the article, and has the serious disadvantage that it can only be added or updated in a system update. The rationale for putting this in an app is very clear.
Not accessible? It's a background app made by Google. It's their OS, they have access to everything.
Not accessible to the user, meaning no nice UI to display and manage it all. Clearly the app can't add any fundamental OS capabilities; everything needed to track and control usage has to already be in the system.
Oh, and obviously it's not just "three big fucking counters", since it gives you very fine-grained information about usage. And lets you control it.
I like Google, but I think they've lost the ability to execute. I find their products are increasingly half-baked
Actually, I think it's the opposite. Google's products are more polished at launch than they ever used to be. From the outset, Google's modus operandi has always been to launch early and incomplete, and then iterate incrementally. Remember when every new launch was explicitly tagged as "beta"? Expectations have changed, though, and now people expect Google stuff to work perfectly from the beginning. I think Google's execution has actually gotten much better (though there are some glaring counterexamples), but expectations have risen even more.
and getting flakier over time (e.g. Chromecast).
Chromecast is getting flakier? I haven't noticed that.
Other companies do this, just not in so many words. Aside from the nudging and winking policies, some things are done under overt, yet bogus reasons. My company now deletes virtually all email after 3 months, to better manage storage resources and adhere to best practices for retention.
Three months is rather short. Most companies have a one year retention policy. This isn't for any nefarious reasons, and it's not to save disk space. It's to limit the information available to discovery during lawsuits, less to make the information unavailable than to eliminate the need to pay lawyers lots and lots of money to comb through it to comply with discovery requests. Since any large company -- no matter how saintly -- is always involved in multiple lawsuits, this saves huge amounts of money.
The truth is, I use bitcoin everyday to buy coffee, gas, pay my bills, take my dates out, and buy every day items such as groceries and electronics.
You're not a liar, you're just wrong. Or perhaps you're being lied to.
You're using dollars every day, lent to you by a bitcoin wallet provider against bitcoins held as collateral by the provider. This is not a bit different from the debit card I have that draws against my stock brokerage margin account, except that your provider is almost certainly more aggressive about doing "margin calls" than my brokerage.
Well, one other difference is that you're currently in fat city, seeing your BTC appreciate in value far faster than my stocks do. Of course, there will come a time when that reverses, and it'll reverse much, much harder and faster than any stock market crash, because stocks have, at root, something of real tangible value. The only real value of BTC is its use as a currency which, as others have pointed out, has simply broken down as it attempted to scale. Were BTC a better cryptocurrency with better scalability characteristics, then it would be the case that BTC is just for criminals.
No, bitcoin is not money. It's digital gold. How much does it cost to transact a bar of gold ?
You don't actually know how BTC works, do you?
They don't know, and there is a valid reason for it. The market is irrational. The market is as much about emotion as it is about fundamentals, maybe even more so.
There *are* no fundamentals. The transaction rates/costs make it unusable as a currency for more than a fairly small number of people, and it has no other value than as a currency.
But the remaining pool of fools is dwindling fast.
I don't know. There are a lot of fools. A sucker is born every minute and all that.
I'm sure the bubble will burst. I haven't the foggiest idea how far we are from that, and neither do you.
One for UHF Chanel's (2-13+VHS) and VHS (14-72?)
VHF, not VHS, and you have them reversed. Very High Frequency channels were 2-13, UHF channels were 14-83. (I'll admit I had to look the upper limit of UHF up, though I knew it was 80-something).
I'm not a fan of Uber as a company. They've done a lot of shady crap. But this sounds like it could also be normal corporate "speaking with care" training. Every company I've ever worked for has told me to be cautious with communicating about legally-sensitive topics in writing (including email and chat systems that archive conversations). Said training usually includes information about how to specify that written communication is attorney-client privileged, too, which makes it non-discoverable in most legal proceedings, plus a recommendation that if you're in doubt, you should hold your conversation in person or over the phone (or video conference).
This is just normal stuff. Sure, it could be used to avoid leaving documentary evidence of illegal dealings, but that's not its intent. Its intent is to avoid generating large amounts of documentation that has to be reviewed in discovery, and which could be exploited by opposing counsel by mischaracterizing it or taking it out of context. For individual employees, it also tends to keep them out of the line of fire of subpoenas. If you mentioned a topic in discoverable communications, you're likely to get subpoenaed to give a deposition.
It should be noted that if you're actually doing something illegal, and if your attorney is unethical enough to advise you on how to get away with it, the training shouldn't tell you to use any chat apps, or phone calls, etc., because there's always a chance that an investigation is already under way and that the communications medium is tapped. If you're committing crime, "speaking with care" recommendations are not good enough.
To what extent is this SOC comparable to Intel's management engine? Isn't Google basically putting a second computer in my computer? With all the risks that entails?
There are many computers inside your computer. The Intel ME is risky because it can take very low level control of the device, update software, etc. The Vision Core is just a graphics coprocessor, not something that can take over control of the device.
Of course, that would be unnecessary, since unless you've unlocked the bootloader and installed custom software, Google can already update your software, etc. Or if you have a Samsung device, Samsung can, etc.
I dislike reading things from viewpoints that involve malicious lies. You want to maintain that lying is a conservative value, you go right ahead.
You're missing the point.