You can't categorically declare that Bitcoin's price is "unwarranted by the fundamentals of the asset". It's a new asset type and its fundamentals are yet to be determined. We're still feeling out Bitcoin's ultimate utility and long-term viability. Surges like this are inevitable.
We've gone through this process several times already. Each time people have declared it a "bubble", and yet... while each surge has been followed by a "crash", the average price after each crash has been significantly higher than the average price before the preceding surge. This was true at $2, $30, $200, $1200, and $4000. The long-term trend has been toward gradually increasing prices and less volatility.
CO2 is fungible, so it doesn't matter which CO2 source you offset.
It matters if your method of capturing CO2 requires releasing an equal amount of CO2 to produce the necessary ingredients. You're not "offsetting" anything in that case, just converting CaCO3 -> CaO + CO2 -> CaCO3 in a closed loop and wasting energy in the process.
You have to be a futures trader, or a research scientist for this kind of article to have any impact on you...
Everyone is a "futures trader" to some extent. For example, if you think there will be a massive improvement in commercially-available batteries within the next five years or so then it may not make sense to go out and buy an expensive electric vehicle based on soon-to-be-obsolete battery tech. Articles like this one help you estimate the likelihood of that happening and make a more informed decision, even if the commercial applications are still some years down the road.
Let's also be pessimistic and say you only get 200 miles off-road from your battery.
You call 200 miles of hard off-road driving per charge "pessimistic"? Continuous use at high torque is going to consume several times the power required simply to maintain a constant speed on a relatively flat highway, the basis for the 250-300 mile statistic. I think you'd be very lucky to get half that range. Don't forget, too, that some of the larger power draws for an all-electric vehicle (like heat/AC) do not scale linearly with distance. You're consuming the same energy to maintain climate control for an hour regardless of whether you travel 20 miles or 60 during that time.
Even using your range figures, though, we're talking about the kind of use that might see you away from the grid for several days at a time. Ten hours over a couple of days is not really so much, and that's assuming you don't leave any margin for error.
No, the abuses make it worse but the underlying principles aren't good either. Penalizing someone for using their own property in a non-violent way, just because someone else did it first, is evil. Copyright is wrong too, for many of the same reasons and others besides, but at least there you can limit your expose to things others might claim copyright on—only copies are prohibited, and you can't copy what you've never seen. That doesn't work with patents; whoever comes up with the idea first gets a monopoly, even if they only came up with the general idea and couldn't actually make it work in the real world, and it doesn't matter whether others copied them or rediscovered the principles independently. The only arguments in favor of either copyrights or patents are the same sort of "ends justify the means" social-engineering logic you find in the U.S. Constitution. I am not convinced that patents accomplish their ostensible goal of encouraging innovation, but even if they did it wouldn't make any difference; I consider people accountable for their actions, not their intentions. Your means determine your ends.
In fact, these situations are identical to running Visual Studio on a virtual machine! Any application written in Java is actually being run on its virtual machine rather than on Ubuntu, why should that situation be any different?
The most relevant differences are that these virtual machines are (a) free software and (b) included with your Ubuntu installation. Windows is neither. One might also point out that the JVM or other language-specific virtual machine is much "thinner", and better integrated with the host operating system, than the full PC hardware emulation that you need to run Windows as a guest operating system.
In contrast, free medical care requires other people to provide it, therefor it is not a right.
Because everyone is a gunsmith.
Your point? The 2nd Amendment says that you have the right to own and carry a gun, not that anyone owes you a gun. There is no tax-funded subsidy to ensure that everyone has access to a weapon regardless of ability to pay. If you want one it's up to you to acquire it at your own expense. Medical care is no different. You have the right to accept medical care, when care is available, and others have the right to provide care to you if they choose—but no one owes you medical care any more than they owe you a gun.
The way to do this is with digital signatures. You have no idea if the source server has been compromised.
Digital signatures (HMAC) are part of HTTPS. They're automatically generated on demand, so not quite as good as a manual signature generated offline after careful examination of the content and verified after every download by the end user with a proper web of trust—but still better than nothing and no worse than any other signature generated on the server itself. They at least prove that the content did originate from a system authorized to sign it on behalf of the administrator. Whether or not you can trust the server to remain under the administrator's control is a separate matter.
From the server operator's point of view, you always want to use HTTPS because you will be blamed either way for any malware people receive while visiting your site. Even if you offer signatures, which isn't a very practical solution for most of the Web. You can police your own server; you can't control what others might inject into the traffic if the connection is unsecured.
Who said anything about unsigned.deb packages? The web page you're reading contains executable code (JavaScript). If you browse the Web with JS enabled, like a normal person, then you need to be careful about provenance of the HTML and JS files which make up the pages. They run in a sandbox, true, but that doesn't prevent all issues (e.g. ad injection, tracking, cross-site attacks) and there is always the possibility of an exploit which can escape the sandbox. As a content provider, if you want people to be able to trust your site you'll deliver it over HTTPS.
if they have influenced those actions it could be seen as reasonable to (be able to) question the extent of that influence and, if 'necessary', impose repercussions (to 'train' the behaviour out of society).
Sure, that could be seen as reasonable... if you don't care about justice or proportional response, and view people as your subjects to manipulate however you want to achieve your desired ends. On the other hand, if you do care about those things and do not subscribe to the end-justifies-the-means theory, it becomes rather obvious that Person B is solely responsible for the murder. Person A's actions consist of nothing more than (a) speech and (b) giving away his own property, which are both things that he or she is perfectly entitled to do and which, in and of themselves, cause no harm to anyone. Person A did not compel Person B to do anything. The choice and the responsibility for the consequences lie with Person B alone.
To generalize, either you believe that all of your success is 100% due to your own brilliance and hard work, with almost none of it due to luck or any help you got from anyone else, in which case you end up on the Libertarian/Republican side of the chart. Or you believe that even though you worked hard you also got a lot of benefits from society and got some lucky breaks along the way... and feel like you should pay it forward, in which case you end up on the Democratic/Socialist... side of the spectrum.
Yeah, this is nonsense. You don't have to believe that you are entirely responsible for your own success to be an individualist (Libertarian or Republican). "Paying it forward" just takes a different form: private charity rather than public services. Along the same lines, collectivism does not imply that you are in favor of contributing personally; it's all about getting other people to pay for the programs you think should be available. Sure, you'll (probably) be paying those taxes you lobbied for along with everyone else... but there was nothing stopping you supporting those programs privately, out of your own funds, if that was what you wanted. The only reason to bring in the power of the State is to force others to participate.
One day, someone will explain to me why a completely public piece of information, distributed freely to everyone everywhere, needs to be delivered securely.
Because, while the information is public, the fact that you requested it should be private. Even seemingly innocuous requests can reveal personal information which can be used against you, regardless of whether you are doing anything "wrong". Limiting how much strangers know about you is a powerful first line of defense: a form of social camouflage.
Because you need to be able to trust that the information is from the expected source and hasn't been tampered with in transit. This is particularly true when the information may contain executable code—which is almost everything these days—but even theoretically non-executable data could be manipulated to exploit a bug in the receiver.
Sure, it will take some time, but I rather have reliable results slowly than wrong results fast. This is not the case where failing fast is a good thing.
On the contrary, failing fast is the right answer. That means deciding early on that the correct result cannot be determined and reporting the failure, rather than wasting resources only to fail later on (failing slowly) or silently producing the wrong result (not failing at all). Electronic voting machines are one example of a "fail fast" election system, when they work correctly: any issues with invalid selections are handled interactively, before the ballot enters the system. Electronically-scanned paper voting can be only slightly slower, if the scanned ballot is validated before it goes into the lockbox. In that case an invalid selection requires the voting process to be repeated, wasting some time, but at least the voter knows that their vote will be counted once it has been accepted. A fully-manual system is of the "not failing at all" variety—if there is an issue with a ballot it won't be discovered until the votes are tallied, and at that point the only possible resolution is to discard the ballot so in the end no one knows whether their particular vote was counted.
If Google is trying to get a patent for an already-existing invention, then it's far more likely they applied for it just in case the USPTO was stupid enough to grant it.
If they did that knowingly they would be committing fraud by claiming to be the inventor. If the point was merely to prevent someone else from getting a patent they could just ensure that the technique is clearly documented, with proper attribution, and published in a public prior-art database, which would also help anyone wanting to use it.
Without a patent he can't (successfully) sue them for using it, no, but since he released it into the public domain we must assume that he didn't intend to do that anyway. He wanted people to use it. The fact that the technique was well-known before Google attempted to patent it does mean that their patent application is invalid due to prior art. You can't just patent a technique someone else invented, even if it is in the public domain.
You have two options, buy hardware that is maxed out but leaves no room for growth, or buy hardware that is constrained by licensing/software. In the second case, you can buy an new license and get access to more hardware/performance without having to physically change anything....
You're trying to cast this as as good thing, but it really isn't. You're getting the same "maxed out" hardware in both cases. The unthrottled version doesn't cost any more to produce than the restricted model; the surcharge is purely an artifact of a non-competitive market. If you want to hold some performance in reserve for later, fine—but you should not have to pay extra when you decide you want to use the full capabilities of the hardware you've already purchased.
If I link ONE transaction to your wallet - I then have EVERY transaction in your wallet.
This is overstating matters. Modern ("hierarchical deterministic") Bitcoin wallets make it easy to use separate addresses for each incoming transfer, and this is generally the default setting. The addresses are all derived from a single seed, but unless you know that seed they appear to be independent. Incoming payments are not associated with the other addresses based on the same seed (your "wallet") until the coins are spent, and even then only to the extent that multiple inputs are combined together to fund a single transaction. Bitcoin clients routinely offer privacy-enhancing features such as control over which inputs are combined and randomizing the outputs to make it difficult to tell which is the payment and which is the "change".
And if all that isn't enough, there is no limit to the number of separate wallets you can have.
I see not enough data points if there are not enough reviews. Its quality is ill-defined, not good.
In the absence of any context I might agree with you, but we can make inferences from experience with other similar products to reason about the probable product quality even if there are no reviews yet for that particular item. The initial estimate is going to be along the lines of "average quality"—perhaps a bit lower than average if it's a new, untested item. A thousand reviews with a mean rating of one star would suggest well below average quality.
Note that I never said that a dearth of reviews mean that the product is good, just that as a relatively unknown quantity it's more likely to be good than a different one-star product with many low reviews, which is very unlikely to be good. (Neither one would be my first choice.)
What statistician says "you should study with the smallest sample size all else being equal"?!
That's not what they said. What they said was that if you have two products with the same low review, the one with fewer reviews has a higher chance of actually being good. Consider the case of a 1-star item with a single review vs. a 1-star item with 1,000 reviews. The first case could just be failure to understand the product, a bad shipping experience, or even a data entry error, whereas the second case indicates widespread dissatisfaction.
In other words, the smaller sample size is more likely to be wrong—and that is just as true of negative reviews as of positive ones.
I can't wrap my mind around two moving targets having equal odds as one moving target and one stationary target.
Imagine that instead of rolling the two (six-sided) dice together you roll them one at a time. Same result, right? The timing shouldn't make any difference. When you roll the first die any result will serve, so the probability of success is 100% and the problem can be simplified to the single-die case: you're trying to match a fixed number, which happens to be the result of the first roll. In the end it doesn't matter how the first value is selected. It wouldn't even make a difference if one die was loaded, so long as the other was fair.
The precise definition really doesn't matter. Freedom of speech is a natural consequence of the fundamental principle of proportional response. Simply avoid escalating matters by answering speech with physical violence (including "legal" violence like jail, fines, etc.) and the rest will take care of itself.
The protocol works by piggybacking a TLS connection to an unblocked host and hiding data in the ciphertext ("chosen ciphertext steganography"). This hidden data is separately encrypted with the ISP's public key and invisible to everyone else, camouflaged within the regular ciphertext which also looks like random noise to anyone without the key. All the censors see is a standard TLS connection to a perfectly normal and uncontroversial web site. An active MitM interception (with TLS proxy certificates installed on the end-users' device) might be able to tell that the plaintext is gibberish, but still wouldn't be able to decode the hidden instructions intended for the "refracting" ISP.
Of course, the software, having the recording, can "listen" to it as often as it wants. There is absolutely no "advantage" here for the human transcribers.
The advantage is that the humans are being given much more time to process the recording. While the human transcriptionists are reviewing the recording multiple times, in real-time, the speech recognition software is producing immediate results.
Ever used the free WiFi in an airport - the dopey kids sitting across from you are streaming some mind-rot and killing the bandwidth for everyone else. So the kids get the lolz, and you can barely get your work emails.
That's just poor bandwidth management. The operator shouldn't be throttling the video just because it's video, much less because they consider it "less important". Instead, available bandwidth should be divided equally between all active users. Given a properly-configured traffic-shaping router, your work e-mails should get through without any issue and without any noticeable impact to the other users of the network. If the system is simply oversubscribed then that is indeed a problem, but in that case it should be equally slow for everyone. If it's so oversubscribed that you can't even get your work e-mails then no one is going to be using it to watch video anyway, if the bandwidth allocation is equitable.
Slashdot Mirror
You can't categorically declare that Bitcoin's price is "unwarranted by the fundamentals of the asset". It's a new asset type and its fundamentals are yet to be determined. We're still feeling out Bitcoin's ultimate utility and long-term viability. Surges like this are inevitable.
We've gone through this process several times already. Each time people have declared it a "bubble", and yet... while each surge has been followed by a "crash", the average price after each crash has been significantly higher than the average price before the preceding surge. This was true at $2, $30, $200, $1200, and $4000. The long-term trend has been toward gradually increasing prices and less volatility.
CO2 is fungible, so it doesn't matter which CO2 source you offset.
It matters if your method of capturing CO2 requires releasing an equal amount of CO2 to produce the necessary ingredients. You're not "offsetting" anything in that case, just converting CaCO3 -> CaO + CO2 -> CaCO3 in a closed loop and wasting energy in the process.
added USB 3.0 ... swapped out the ExpressCard (that has proved to be rather useless to me) with an SD card slot
For less than $10 you can get an ExpressCard adapter which adds an SDHC/MMC slot. Adapters are also available to add two USB 3.0 ports.
You have to be a futures trader, or a research scientist for this kind of article to have any impact on you...
Everyone is a "futures trader" to some extent. For example, if you think there will be a massive improvement in commercially-available batteries within the next five years or so then it may not make sense to go out and buy an expensive electric vehicle based on soon-to-be-obsolete battery tech. Articles like this one help you estimate the likelihood of that happening and make a more informed decision, even if the commercial applications are still some years down the road.
Let's also be pessimistic and say you only get 200 miles off-road from your battery.
You call 200 miles of hard off-road driving per charge "pessimistic"? Continuous use at high torque is going to consume several times the power required simply to maintain a constant speed on a relatively flat highway, the basis for the 250-300 mile statistic. I think you'd be very lucky to get half that range. Don't forget, too, that some of the larger power draws for an all-electric vehicle (like heat/AC) do not scale linearly with distance. You're consuming the same energy to maintain climate control for an hour regardless of whether you travel 20 miles or 60 during that time.
Even using your range figures, though, we're talking about the kind of use that might see you away from the grid for several days at a time. Ten hours over a couple of days is not really so much, and that's assuming you don't leave any margin for error.
No, the abuses make it worse but the underlying principles aren't good either. Penalizing someone for using their own property in a non-violent way, just because someone else did it first, is evil. Copyright is wrong too, for many of the same reasons and others besides, but at least there you can limit your expose to things others might claim copyright on—only copies are prohibited, and you can't copy what you've never seen. That doesn't work with patents; whoever comes up with the idea first gets a monopoly, even if they only came up with the general idea and couldn't actually make it work in the real world, and it doesn't matter whether others copied them or rediscovered the principles independently. The only arguments in favor of either copyrights or patents are the same sort of "ends justify the means" social-engineering logic you find in the U.S. Constitution. I am not convinced that patents accomplish their ostensible goal of encouraging innovation, but even if they did it wouldn't make any difference; I consider people accountable for their actions, not their intentions. Your means determine your ends.
In fact, these situations are identical to running Visual Studio on a virtual machine! Any application written in Java is actually being run on its virtual machine rather than on Ubuntu, why should that situation be any different?
The most relevant differences are that these virtual machines are (a) free software and (b) included with your Ubuntu installation. Windows is neither. One might also point out that the JVM or other language-specific virtual machine is much "thinner", and better integrated with the host operating system, than the full PC hardware emulation that you need to run Windows as a guest operating system.
In contrast, free medical care requires other people to provide it, therefor it is not a right.
Because everyone is a gunsmith.
Your point? The 2nd Amendment says that you have the right to own and carry a gun, not that anyone owes you a gun. There is no tax-funded subsidy to ensure that everyone has access to a weapon regardless of ability to pay. If you want one it's up to you to acquire it at your own expense. Medical care is no different. You have the right to accept medical care, when care is available, and others have the right to provide care to you if they choose—but no one owes you medical care any more than they owe you a gun.
The way to do this is with digital signatures. You have no idea if the source server has been compromised.
Digital signatures (HMAC) are part of HTTPS. They're automatically generated on demand, so not quite as good as a manual signature generated offline after careful examination of the content and verified after every download by the end user with a proper web of trust—but still better than nothing and no worse than any other signature generated on the server itself. They at least prove that the content did originate from a system authorized to sign it on behalf of the administrator. Whether or not you can trust the server to remain under the administrator's control is a separate matter.
From the server operator's point of view, you always want to use HTTPS because you will be blamed either way for any malware people receive while visiting your site. Even if you offer signatures, which isn't a very practical solution for most of the Web. You can police your own server; you can't control what others might inject into the traffic if the connection is unsecured.
Who said anything about unsigned .deb packages? The web page you're reading contains executable code (JavaScript). If you browse the Web with JS enabled, like a normal person, then you need to be careful about provenance of the HTML and JS files which make up the pages. They run in a sandbox, true, but that doesn't prevent all issues (e.g. ad injection, tracking, cross-site attacks) and there is always the possibility of an exploit which can escape the sandbox. As a content provider, if you want people to be able to trust your site you'll deliver it over HTTPS.
if they have influenced those actions it could be seen as reasonable to (be able to) question the extent of that influence and, if 'necessary', impose repercussions (to 'train' the behaviour out of society).
Sure, that could be seen as reasonable... if you don't care about justice or proportional response, and view people as your subjects to manipulate however you want to achieve your desired ends. On the other hand, if you do care about those things and do not subscribe to the end-justifies-the-means theory, it becomes rather obvious that Person B is solely responsible for the murder. Person A's actions consist of nothing more than (a) speech and (b) giving away his own property, which are both things that he or she is perfectly entitled to do and which, in and of themselves, cause no harm to anyone. Person A did not compel Person B to do anything. The choice and the responsibility for the consequences lie with Person B alone.
To generalize, either you believe that all of your success is 100% due to your own brilliance and hard work, with almost none of it due to luck or any help you got from anyone else, in which case you end up on the Libertarian/Republican side of the chart. Or you believe that even though you worked hard you also got a lot of benefits from society and got some lucky breaks along the way ... and feel like you should pay it forward, in which case you end up on the Democratic/Socialist ... side of the spectrum.
Yeah, this is nonsense. You don't have to believe that you are entirely responsible for your own success to be an individualist (Libertarian or Republican). "Paying it forward" just takes a different form: private charity rather than public services. Along the same lines, collectivism does not imply that you are in favor of contributing personally; it's all about getting other people to pay for the programs you think should be available. Sure, you'll (probably) be paying those taxes you lobbied for along with everyone else... but there was nothing stopping you supporting those programs privately, out of your own funds, if that was what you wanted. The only reason to bring in the power of the State is to force others to participate.
One day, someone will explain to me why a completely public piece of information, distributed freely to everyone everywhere, needs to be delivered securely.
Because, while the information is public, the fact that you requested it should be private. Even seemingly innocuous requests can reveal personal information which can be used against you, regardless of whether you are doing anything "wrong". Limiting how much strangers know about you is a powerful first line of defense: a form of social camouflage.
Because you need to be able to trust that the information is from the expected source and hasn't been tampered with in transit. This is particularly true when the information may contain executable code—which is almost everything these days—but even theoretically non-executable data could be manipulated to exploit a bug in the receiver.
Sure, it will take some time, but I rather have reliable results slowly than wrong results fast. This is not the case where failing fast is a good thing.
On the contrary, failing fast is the right answer. That means deciding early on that the correct result cannot be determined and reporting the failure, rather than wasting resources only to fail later on (failing slowly) or silently producing the wrong result (not failing at all). Electronic voting machines are one example of a "fail fast" election system, when they work correctly: any issues with invalid selections are handled interactively, before the ballot enters the system. Electronically-scanned paper voting can be only slightly slower, if the scanned ballot is validated before it goes into the lockbox. In that case an invalid selection requires the voting process to be repeated, wasting some time, but at least the voter knows that their vote will be counted once it has been accepted. A fully-manual system is of the "not failing at all" variety—if there is an issue with a ballot it won't be discovered until the votes are tallied, and at that point the only possible resolution is to discard the ballot so in the end no one knows whether their particular vote was counted.
If Google is trying to get a patent for an already-existing invention, then it's far more likely they applied for it just in case the USPTO was stupid enough to grant it.
If they did that knowingly they would be committing fraud by claiming to be the inventor. If the point was merely to prevent someone else from getting a patent they could just ensure that the technique is clearly documented, with proper attribution, and published in a public prior-art database, which would also help anyone wanting to use it.
Without a patent he can't (successfully) sue them for using it, no, but since he released it into the public domain we must assume that he didn't intend to do that anyway. He wanted people to use it. The fact that the technique was well-known before Google attempted to patent it does mean that their patent application is invalid due to prior art. You can't just patent a technique someone else invented, even if it is in the public domain.
You have two options, buy hardware that is maxed out but leaves no room for growth, or buy hardware that is constrained by licensing/software. In the second case, you can buy an new license and get access to more hardware/performance without having to physically change anything....
You're trying to cast this as as good thing, but it really isn't. You're getting the same "maxed out" hardware in both cases. The unthrottled version doesn't cost any more to produce than the restricted model; the surcharge is purely an artifact of a non-competitive market. If you want to hold some performance in reserve for later, fine—but you should not have to pay extra when you decide you want to use the full capabilities of the hardware you've already purchased.
If I link ONE transaction to your wallet - I then have EVERY transaction in your wallet.
This is overstating matters. Modern ("hierarchical deterministic") Bitcoin wallets make it easy to use separate addresses for each incoming transfer, and this is generally the default setting. The addresses are all derived from a single seed, but unless you know that seed they appear to be independent. Incoming payments are not associated with the other addresses based on the same seed (your "wallet") until the coins are spent, and even then only to the extent that multiple inputs are combined together to fund a single transaction. Bitcoin clients routinely offer privacy-enhancing features such as control over which inputs are combined and randomizing the outputs to make it difficult to tell which is the payment and which is the "change".
And if all that isn't enough, there is no limit to the number of separate wallets you can have.
I see not enough data points if there are not enough reviews. Its quality is ill-defined, not good.
In the absence of any context I might agree with you, but we can make inferences from experience with other similar products to reason about the probable product quality even if there are no reviews yet for that particular item. The initial estimate is going to be along the lines of "average quality"—perhaps a bit lower than average if it's a new, untested item. A thousand reviews with a mean rating of one star would suggest well below average quality.
Note that I never said that a dearth of reviews mean that the product is good, just that as a relatively unknown quantity it's more likely to be good than a different one-star product with many low reviews, which is very unlikely to be good. (Neither one would be my first choice.)
What statistician says "you should study with the smallest sample size all else being equal"?!
That's not what they said. What they said was that if you have two products with the same low review, the one with fewer reviews has a higher chance of actually being good. Consider the case of a 1-star item with a single review vs. a 1-star item with 1,000 reviews. The first case could just be failure to understand the product, a bad shipping experience, or even a data entry error, whereas the second case indicates widespread dissatisfaction.
In other words, the smaller sample size is more likely to be wrong—and that is just as true of negative reviews as of positive ones.
I can't wrap my mind around two moving targets having equal odds as one moving target and one stationary target.
Imagine that instead of rolling the two (six-sided) dice together you roll them one at a time. Same result, right? The timing shouldn't make any difference. When you roll the first die any result will serve, so the probability of success is 100% and the problem can be simplified to the single-die case: you're trying to match a fixed number, which happens to be the result of the first roll. In the end it doesn't matter how the first value is selected. It wouldn't even make a difference if one die was loaded, so long as the other was fair.
Define freedom of speech.
The precise definition really doesn't matter. Freedom of speech is a natural consequence of the fundamental principle of proportional response. Simply avoid escalating matters by answering speech with physical violence (including "legal" violence like jail, fines, etc.) and the rest will take care of itself.
The protocol works by piggybacking a TLS connection to an unblocked host and hiding data in the ciphertext ("chosen ciphertext steganography"). This hidden data is separately encrypted with the ISP's public key and invisible to everyone else, camouflaged within the regular ciphertext which also looks like random noise to anyone without the key. All the censors see is a standard TLS connection to a perfectly normal and uncontroversial web site. An active MitM interception (with TLS proxy certificates installed on the end-users' device) might be able to tell that the plaintext is gibberish, but still wouldn't be able to decode the hidden instructions intended for the "refracting" ISP.
Of course, the software, having the recording, can "listen" to it as often as it wants. There is absolutely no "advantage" here for the human transcribers.
The advantage is that the humans are being given much more time to process the recording. While the human transcriptionists are reviewing the recording multiple times, in real-time, the speech recognition software is producing immediate results.
Ever used the free WiFi in an airport - the dopey kids sitting across from you are streaming some mind-rot and killing the bandwidth for everyone else. So the kids get the lolz, and you can barely get your work emails.
That's just poor bandwidth management. The operator shouldn't be throttling the video just because it's video, much less because they consider it "less important". Instead, available bandwidth should be divided equally between all active users. Given a properly-configured traffic-shaping router, your work e-mails should get through without any issue and without any noticeable impact to the other users of the network. If the system is simply oversubscribed then that is indeed a problem, but in that case it should be equally slow for everyone. If it's so oversubscribed that you can't even get your work e-mails then no one is going to be using it to watch video anyway, if the bandwidth allocation is equitable.