1. He's a hypocrite who got filthy rich and now also wishes to indulge his jaded palate in the pleasures of self-righteousness.
2. He learned something, arguably the hard way, when becoming richer than god failed to provide any substantial hedonic benefits that merely being wealthy enough to avoid the overt pains of poverty didn't.
I suspect that direct comparisons are tricky: magnetic platter surfaces should, at least in theory, have virtually infinite read and erase capability; but every mechanical part dies a little when called on to move(and if the lubricants are a problem, when not called on to move for too long).
With SSDs, we know that the NAND dies a bit every time it is erased and rewritten; sometimes after surprisingly few cycles with contemporary high density MLC NAND; but the supporting solid state stuff should last longer that the person who owns the drive, barring firmware bugs or severe shoddiness.
I know a fair number of grad students and postdocs who (despite a recent encounter with an aspect of education notably not focused on self esteem) are still extremely pessimistic about the payoffs of working hard at STEM as compared to, say, shmoozing through an MBA and getting an honest job offshoring STEM nerds.
Doesn't it warm the free-market cockles of your heart that levels of 'market transparency' in "intellectual property", and the licensing thereof, that a regulatory action taken by commie chinese is the biggest boost it's had in years?
According to our presently available research and body of technique is there really anything on the table that 'results in outstanding academic ability'?
We know about some "Don't fuck it up" procedures (lead is not a dietary supplement, lots of early childhood stimulus is good, malnutrition stunts mind as well as body, etc.); and we know some things about getting better or worse results out of students of a given level of ability; but for anything that has some element of 'born, not made', it's a good day when we can accurately identify the good candidates, much less upgrade inadequate ones.
If your thesis is that 'difficult things are hard and most people can't do them', it wouldn't much matter if the K-12 focus is 'self-esteem', 'classical philology', or 'Measure Theory Bootcamp: No Place For The Weak.'
Never mind that, here in the USSA, despite the cries of "Double Taxation!" and "Highest Tax Rates Evar! Death Taxes!!!", if a decent size multinational is paying an effective tax rate higher than the guy who scrubs out their toilets at night, they probably just need to fire their accountants.
Given the fairly low bandwidth demands of bitcoin mining (only slightly higher than just keeping a client up to date with the blockchain, and increasing with size only by a fairly low constant factor as you need to grab more work units), I wonder what sort of connections most of the world's bitcoin hashing power is on?
Unless there is some sort of "yeah, it's just DSL speeds, but we do something really clever upstream to make it as hard to DDOS as a connection a million times as fast" service, that might actually be how Amazon, Azure, or any other web-services-oriented rental service could manipulate the bitcoin scene:
Not by computing; because CPU miners are toys; but by DDOSing all concentrations of mining power not aligned with they hypothetical attacker into smoking craters long enough to substantially magnify the effective representation of the attacker's compute assets...
Perhaps Russian Business Network wishes to consider, yes?
Malware is handy because it makes Ghashes/watt much less relevant (though, with the portion of today's computing power that is in battery powered or thermally constrained devices, you have to be careful that your compute malware doesn't cause even the most idiotic of users to notice that their laptop now scalds their flesh and lasts 45 minutes on a full charge and bring it in for repair, an uptick that vendors would probably notice relatively quickly and get actually-competent security consultants involved in); but what it doesn't do is change the fact that CPU mining is basically a toy at this point.
Sporadic amounts of time on a few hundred thousand to few million CPUs is still hardly valueless, especially if you pay none of the costs other than a command-and-control server; but you'll be a relatively small player. Not a bad gig if you can get it; but you won't be a kingmaker.
Amazon and Azure are far too expensive, unless a state-actor is willing to invest a few billions.
Unless either is somehow hiding unbelievable losses from overbuying capacity they can't sell, they'd likely get even more expensive if anyone actually had a serious try: power isn't free, so there is presumably a price for EC2 spot instances so low that they won't even fire up the server; but depreciation isn't free, so both Amazon and Microsoft presumably work rather hard to keep their supply close to the available demand and have most of the churn at a given time be give-and-take between cost-sensitive; but failure tolerant, spot customers, and less cost sensitive, less failure tolerant, reserved and on-demand customers, with very little going idle if they can help it.
If you had vast financial resources; but wanted the capacity Now, you could buy out all the spot instances in no time, and with no special fuss; but you'd have to have a special talk with Amazon about suspending normal on-damand pricing and booting all those customers, and a still touchier talk about the possibility of telling all the 'reserved instance' customers that a more important client has come up, so sorry, and booting them. Displacing the resources allocated to Amazon's own e-commerce operations would also be a touchy business.
Both Amazon and MS do a pretty good job of, at amazingly low cost, making sure that there appears to be as much capacity as you want at any given time; but they'd go bankrupt in short notice if that were actually true for radically atypical requests. Even if you paid for buildout, and were willing to wait while the contractors scrambled and shoved stuff in containers and trailers and things, the semiconductor industry is 'just-in-time' all the way down, more or less.
If a state actor (especially American) wanted to roll Bitcoin, they'd probably be better off designing a mining ASIC, or contracting Intel/IBM/some much quieter fabless contractor that usually does custom bits for NRO satellites/whoever to do it, and then using some combination of large amounts of money and insinuations of special consideration on future contracting to get it fabbed by one of the classy outfits that the plucky little ASIC startups(the ones that aren't just scams) just don't have access to.
I definitely wouldn't want to argue with the Ghashes/watt of a well designed ASIC produced by someone who has the clout and the cash to get it produced on Intel's fancy process of the moment...
If someone rents 1,000,000 Amazon severs to mine bitcoins, would you pick a DDOS fight with that. Even with special purpose HW, it's unlikely to be a small pipe.
One potential complication is that 'mining' isn't particularly bandwidth intensive (not quite zero, and more miners require more bandwidth, to cover updating all of them when it a block is found and you need to get new work units immediately since you now know that all the other candidates are incorrect; but overall quite minimal); but it is computationally demanding and very much rewards specialized hardware.
Most of the hardware you can easily rent in places with ample bandwidth will be aimed at customers who want to do web services. Comparatively anemic CPUs; plenty of storage and RAM, nice fat connection; maybe some offerings with faster CPUs or GPU compute units if the vendor has been dabbling with those sorts of customers.
I definitely wouldn't pick a bandwidth fight with such a thing, unless I had some delightfully clever amplification attack up my sleeve; but even at Amazon prices you would be burning insane amounts of money to get a hashrate that would even make you worth paying attention to.
If anything, I'd consider the reverse strategy: much of the world's crazy-ASIC capacity is probably on relatively narrow pipes because that's all they need. Don't bother trying to rent enough to out-compute them, use the rentals in high-bandwidth areas to knock as many ASIC clusters that aren't your collaborators as possible offline.
Silly consumer, the risk of 3rd parties undermining the system is why we need to make Trusted Computing mandatory! Secure remote attestation for all, only rights-management-compliant systems allowed on the network!
"If you want to eliminate special client software there, you can have this system, and run everything on the browser."
Being able to reuse more browser code might well improve the existing proprietary client; but (as with any DRM system) you can't eliminate special client software; because you would otherwise be incapable of distinguishing between a web browser that happily accepts your 'HTTPA' and then ignores your restrictions and one that accepts and obeys.
As ever, you'll either need some suitably dense obfuscated blob, or to go all-out-TPM and do secure remote attestation and 'trusted' everything from preboot on up.
Even older than that. The 'Send the data, along with a description of restrictions on the use of the data' is (along with a dash of 'semantic web' nonsense, which appears to be a new addition), classic "Trusted Computing".
Mark Stefik, and his group at Xerox PARC, were talking about 'Digital Property Rights Language' back in 1994 or so, and by 1998, if not earlier, it had metastasized into a giant chunk of XML. That later mutated into "XrML", the 'Extensible Rights Management Language', which eventually burrowed into MPEG-21/ISO/IEC 21000 as the standard's 'rights expression language'.
My understanding (from TFA and from very peripheral involvement as the support techie in a few situations involving sampled music reproduction(not for performance purposes; but it's actually a fairly popular thing in music education/guided practice, where the computer records you and plays accompaniment for whatever instrument the student is supposed to be learning)) is that the synth-orchestra is midway between a live one and a 'just a recording' background music.
With the more or less exhaustive sampling of all the instruments, you get a MIDI-on-steroids arrangement where you aren't just playing back 'a recording', you can play back any number of instruments, each playing any series of notes, on any number of physical outputs, limited only by your speakers and your software.
I don't know if anybody has done something clever with pointing a Kinect at the conductor and having the 'orchestra' respond, though that'd be a pretty natural thing to take a stab at; but if you are doing the (considerable) work of exhaustive sampling and then just doing a static playback, you are wasting your time. Simply getting a couple of high-quality recordings and playing those back would be easier, faster, and a lot less complex.
Where sampling(along with sophisticated instrument modelling/synthesis, where available) gets interesting is that it can respond in real time, if you have worked out how to measure what it is supposed to respond to.
I don't know how effective this Mr. Goldstein's setup is; but that's the somewhere-between-present-and-not-too-distant-future potential of sampling that makes it even more threatening than mere recording to the live musician. Recordings, however good, are static. Exploiting a library of samples and synthesis models, especially in real time and in response to inputs like 'conductor waving baton' or 'vocal performance' rather than 'experienced audio tech providing input designed for the computer's convenience', is very much a work in progress; but if it can be brought up to spec, it's a great deal closer to 'performance' than it is to 'playback'.
Well, parts of the genre do celebrate 'tradition', which is pretty close to stagnation with higher culture cred; but I imagine that they are also nervous because more or less any endeavor(some fields more radically than others; but all of them to a degree) tend to depend on having some minimum scale, above which economies of scale help make them viable, new people can be trained and educated fast enough that the field's unwritten knowledge(and yes, no matter how you try to document stuff, every field has a body of knowledge that is very difficult to pick up without either arduous rediscovery or learning-while-doing with people who already know) doesn't die off, and so on.
If there is plenty of demand to go around, only the crankiest of purist assholes spend their time whining about damn kids these days, and it's a 'the more, the merrier!' sort of enterprise. If people are dogged by the fear that not only is their bread and butter probably going to disappear; but their entire emotionally-invested-tradition is likely to become nonviable, things get ugly, always have.
It is worth noting that, while money is definitely on the table, odds are good that at least part of the bitterness, and conflict, stems from the fact that only opera levels of money are on the table: Opera as a genre, while it has some 'High Art' old money and classy buildings here and there, faces a dwindling and aging audience, relatively weak interest in younger demographics, and production costs that remain comparatively high.
On the one side, even people who are wholehearted opera enthusiasts, dedicated acting career to it, high enough on the food chain that they don't necessarily face immediate economic pressure, etc. can quite honestly make the argument "Well, outside of a few major cities with sufficient concentrations of potential audience and a few outfits with long histories and some generous patrons, opera will either use electronic orchestras, use (even cheaper and worse) prerecorded soundtracks piped through the speaker system, or simply nonexistent."
At the same time, anyone who has a stake in the continued existence of live orchestras(either as a performer or aesthetically) is probably entirely correct in suspecting that, if almost-as-good-and-substantially-cheaper opera starts to be produced, demand for live orchestra performers will likely decline, probably even faster than it has, and quite possibly to the point where there isn't enough demand to pay for training and ongoing practice, that keeps part-time musicians in decent shape, much less the continued availability of full-time musicians, and the field of live opera will be basically toast.
At least in fields that aren't contracting and quite possibly dying, fights over cost-cutting tend to have slightly better defined battle lines between the mercenaries and the purists.
I realize that anything that can be tenuously connected to 'unions' is intrinsically evil and all (though, while TFA had a couple of quotes from musician unions, obviously against the idea, it had no mention of any union acting against the outfit staging this sampled version, just individual nastygrams sent without any broader consultation or endorsement); but, in an enterprise like theater where individual people, at least stars, are commonly economic actors in their own right, is telling a performer "I find your willingness to replace music with samples corrosive to the medium, you should be ashamed of yourself, and I won't work with you." any different from telling a company "I find your willingness to use suppliers in exploitationstan repulsive, you should be ashamed of yourself, and I won't buy from you."?
Arguably, the 'AI curse' is only partly goalpost-moving by humans obsessed with their special uniqueness.
The other aspect of it (as well a genuinely interesting result of whatever attempt at AI was made) is the discovery that we actually handle much less of what we do in some sort of naive, idealized, "high-level, general-purpose, cognition" than we naively suspect, and rather more in specialized and unconscious mechanisms(this lesson can be learned from the other direction as well: something like face blindness, in otherwise cognitively and visually normal people, simply wouldn't be possible if object recognition were a general-purpose function handled by an 'intelligence' with access to a video stream...) And, even when we aren't crunching stuff unconsciously on quite special-purpose mechanisms, we are sometimes just dumb. Good old Youtube comments would probably allow a babbling Markov chain to pass a Turing test.
There's also the fact that humans are frequently easier to fool than they would like: It's not exactly news that people will see faces on the moon, a known sterile rock, impute emotions (and sometimes entire spirits or deities) to the weather, emotionally bond with tamagotchis and similar nonsense.
Yes, people are likely to dismiss every AI that doesn't end up murdering them all and rendering them for computronium as 'just an expert system'; but it's also arguably the case that using a Turing test to check for AI substantially rests on the assumption, more or less Cartesian and more or less nonsense, that man is first and foremost an abstract 'thinking thing', with some other stuff tacked on that philosophers and mathematicians needn't really worry about.
They also aren't exactly rolling in power in the smartphone sector. Even the most instinctively abusive people and institutions tend to avoid upsetting people when they don't have the power to avoid the consequences.
Whether or not they turn on you later, if they gain power, is a matter of disposition or institutional culture; but making enemies and alienating people when you really can't afford it is something that only the stubborn and the doomed do.
Aside from the odds that running 2.3 may mean that your last security fix was in 2011 (depends on how much the entity that built the image cared), lack of TLS SNI, VPN support, hardware accelerated UI, SELinux, and flash filesystem TRIM support are the major downers.
Application compatibility may or may not be an issue, depending on which ones you want.
At least for contract customers, and often for prepaid, carriers are usually willing to issue a replacement SIM (they might charge 5-10 dollars to do so; but especially for contract customers, why lose a customer over a dinky little crypto IC?) and invalidate the old one.
Not something you want to deal with the hassle of; but not as bad as losing an expensive phone with the SIM in it. (Plus, if your phone is cheap junk, you can still lose/destroy it on your own; but it is less likely to be stolen, which helps increase effective reliability, since most thieves aren't polite enough to pull the SIM before lifting the handset.) As for environmental hazards, SIMs are pretty tough. Unless you physically snap one, give it a particularly stiff hit of ESD, or microwave it, it'll probably be fine. A little salt water, your phone falling a couple of stories onto concrete, and similar mishaps are unlikely to do it much harm.
Even if it doesn't, my understanding is that it runs some sort of Linux kernel, with enough juice for a mostly-HTML/JS GUI to be endurable, plus it obviously includes at least some baseline GSM/GPRS (quite possibly a later standard; but it'd be hard to call it much of a phone without that), which is a notable perk for certain projects.
Interfacing a GSM module with anything from a microcontroller on up isn't all that hard (Hayes Ain't Dead, just extended a lot...); but a bare SIM900 module will run you a trifle above $10 on its own, more for something a little less spartan with a SIM slot and convenient headers. (the SIM900A is a bit cheaper; but only dual band, rather than quad, which can be an issue depending on where you live). Obviously, you can do considerably better if you score a working pull from a dead laptop or something, and a fancy arduino-shield configuration with all the connectors and such soldered on can run you ~$20 alone; but the moral of the story is that if you want to do some sort of embedded project that uses SMS or GPRS in some way, it's getting pretty hard to save money by using an add-on module for your choice of teeny computer if you can, instead, get a low-end linuxized phone that is suitably cooperative about software customization. The only real exception would be in situations where the lack of GPIO and similar microcontroller standard features is an issue, since phones tend to be fairly sparse on that score.
It's getting to be the same way with screens. Unless you are doing some 16x4 character display, or a Nokia 1100 LCD clone, or going all the way up to a cheap 'n cheerful 17 inch monitor, you can have some difficulty sourcing (and driving) moderate size LCDs for less money than you could spend to get the LCD and an entire computer, with battery, attached.
They may not make it here at $25 (a cynic would note that this is hardly the first story involving some Indian outfit that was allegedly going to work wonders in low cost electronics, who then dithers and flakes out, increasing prices, slipping schedules, or trimming specs, until, mysteriously, they eventually manage to deliver something or other, right about the time that Chinese mystery-brands are selling them on ebay for the same money, Slashdot has had at least a couple of such stories involving tablets for education); but the route taken by cheap 'n cheerful electronics to the US appears to be fairly swift and not too marked up(as long as you don't pay Belkin or somebody to write an English manual and slap their sticker on it). The US isn't a 'How do they hit that price point even with counterfeit parts and child labor? miracle of cheapness; but this isn't Brazil, or one of those lucky European countries where the price is the same as in the US, except that they want that many pounds or that many euros...
My typo, I'm afraid (I meant Mountain View, though Windows Phone devices are tied pretty closely to Redmond); but Android solves the problem rather less than it pretends to.
Unless you hack around and run bare AOSP, possibly with certain 3rd-party customizations, the percentage of 'Android' that is actually 'Google Play Services' increases with every version bump. A nontrivial percentage of even non-Google apps also build against Google-specific APIs, rather than the relatively impoverished Android ones (the rule of thumb seems to be that, once a role is added to GPS, the AOSP implementation more or less freezes at whatever state it was in and remains there), so incompatibility, even with the absolute freshest AOSP, is quite common.
Google has the right to do as they wish with their OS, and at least you can buy it on handsets that aren't crypto locked to hell and back; but you'll be talking to the mothership soon enough.
Slashdot Mirror
There are arguably two possibilities:
1. He's a hypocrite who got filthy rich and now also wishes to indulge his jaded palate in the pleasures of self-righteousness.
2. He learned something, arguably the hard way, when becoming richer than god failed to provide any substantial hedonic benefits that merely being wealthy enough to avoid the overt pains of poverty didn't.
I suspect that direct comparisons are tricky: magnetic platter surfaces should, at least in theory, have virtually infinite read and erase capability; but every mechanical part dies a little when called on to move(and if the lubricants are a problem, when not called on to move for too long).
With SSDs, we know that the NAND dies a bit every time it is erased and rewritten; sometimes after surprisingly few cycles with contemporary high density MLC NAND; but the supporting solid state stuff should last longer that the person who owns the drive, barring firmware bugs or severe shoddiness.
Yes, they are sooo reliable, every single SDD I've bought has been dead within 3 months.
A happy OCZ customer, I take it?
I know a fair number of grad students and postdocs who (despite a recent encounter with an aspect of education notably not focused on self esteem) are still extremely pessimistic about the payoffs of working hard at STEM as compared to, say, shmoozing through an MBA and getting an honest job offshoring STEM nerds.
Doesn't it warm the free-market cockles of your heart that levels of 'market transparency' in "intellectual property", and the licensing thereof, that a regulatory action taken by commie chinese is the biggest boost it's had in years?
Good work on that free market, guys.
According to our presently available research and body of technique is there really anything on the table that 'results in outstanding academic ability'?
We know about some "Don't fuck it up" procedures (lead is not a dietary supplement, lots of early childhood stimulus is good, malnutrition stunts mind as well as body, etc.); and we know some things about getting better or worse results out of students of a given level of ability; but for anything that has some element of 'born, not made', it's a good day when we can accurately identify the good candidates, much less upgrade inadequate ones.
If your thesis is that 'difficult things are hard and most people can't do them', it wouldn't much matter if the K-12 focus is 'self-esteem', 'classical philology', or 'Measure Theory Bootcamp: No Place For The Weak.'
Never mind that, here in the USSA, despite the cries of "Double Taxation!" and "Highest Tax Rates Evar! Death Taxes!!!", if a decent size multinational is paying an effective tax rate higher than the guy who scrubs out their toilets at night, they probably just need to fire their accountants.
Given the fairly low bandwidth demands of bitcoin mining (only slightly higher than just keeping a client up to date with the blockchain, and increasing with size only by a fairly low constant factor as you need to grab more work units), I wonder what sort of connections most of the world's bitcoin hashing power is on?
Unless there is some sort of "yeah, it's just DSL speeds, but we do something really clever upstream to make it as hard to DDOS as a connection a million times as fast" service, that might actually be how Amazon, Azure, or any other web-services-oriented rental service could manipulate the bitcoin scene:
Not by computing; because CPU miners are toys; but by DDOSing all concentrations of mining power not aligned with they hypothetical attacker into smoking craters long enough to substantially magnify the effective representation of the attacker's compute assets...
Perhaps Russian Business Network wishes to consider, yes?
Malware is handy because it makes Ghashes/watt much less relevant (though, with the portion of today's computing power that is in battery powered or thermally constrained devices, you have to be careful that your compute malware doesn't cause even the most idiotic of users to notice that their laptop now scalds their flesh and lasts 45 minutes on a full charge and bring it in for repair, an uptick that vendors would probably notice relatively quickly and get actually-competent security consultants involved in); but what it doesn't do is change the fact that CPU mining is basically a toy at this point.
Sporadic amounts of time on a few hundred thousand to few million CPUs is still hardly valueless, especially if you pay none of the costs other than a command-and-control server; but you'll be a relatively small player. Not a bad gig if you can get it; but you won't be a kingmaker.
Amazon and Azure are far too expensive, unless a state-actor is willing to invest a few billions.
Unless either is somehow hiding unbelievable losses from overbuying capacity they can't sell, they'd likely get even more expensive if anyone actually had a serious try: power isn't free, so there is presumably a price for EC2 spot instances so low that they won't even fire up the server; but depreciation isn't free, so both Amazon and Microsoft presumably work rather hard to keep their supply close to the available demand and have most of the churn at a given time be give-and-take between cost-sensitive; but failure tolerant, spot customers, and less cost sensitive, less failure tolerant, reserved and on-demand customers, with very little going idle if they can help it.
If you had vast financial resources; but wanted the capacity Now, you could buy out all the spot instances in no time, and with no special fuss; but you'd have to have a special talk with Amazon about suspending normal on-damand pricing and booting all those customers, and a still touchier talk about the possibility of telling all the 'reserved instance' customers that a more important client has come up, so sorry, and booting them. Displacing the resources allocated to Amazon's own e-commerce operations would also be a touchy business.
Both Amazon and MS do a pretty good job of, at amazingly low cost, making sure that there appears to be as much capacity as you want at any given time; but they'd go bankrupt in short notice if that were actually true for radically atypical requests. Even if you paid for buildout, and were willing to wait while the contractors scrambled and shoved stuff in containers and trailers and things, the semiconductor industry is 'just-in-time' all the way down, more or less.
If a state actor (especially American) wanted to roll Bitcoin, they'd probably be better off designing a mining ASIC, or contracting Intel/IBM/some much quieter fabless contractor that usually does custom bits for NRO satellites/whoever to do it, and then using some combination of large amounts of money and insinuations of special consideration on future contracting to get it fabbed by one of the classy outfits that the plucky little ASIC startups(the ones that aren't just scams) just don't have access to.
I definitely wouldn't want to argue with the Ghashes/watt of a well designed ASIC produced by someone who has the clout and the cash to get it produced on Intel's fancy process of the moment...
If someone rents 1,000,000 Amazon severs to mine bitcoins, would you pick a DDOS fight with that. Even with special purpose HW, it's unlikely to be a small pipe.
One potential complication is that 'mining' isn't particularly bandwidth intensive (not quite zero, and more miners require more bandwidth, to cover updating all of them when it a block is found and you need to get new work units immediately since you now know that all the other candidates are incorrect; but overall quite minimal); but it is computationally demanding and very much rewards specialized hardware.
Most of the hardware you can easily rent in places with ample bandwidth will be aimed at customers who want to do web services. Comparatively anemic CPUs; plenty of storage and RAM, nice fat connection; maybe some offerings with faster CPUs or GPU compute units if the vendor has been dabbling with those sorts of customers.
I definitely wouldn't pick a bandwidth fight with such a thing, unless I had some delightfully clever amplification attack up my sleeve; but even at Amazon prices you would be burning insane amounts of money to get a hashrate that would even make you worth paying attention to.
If anything, I'd consider the reverse strategy: much of the world's crazy-ASIC capacity is probably on relatively narrow pipes because that's all they need. Don't bother trying to rent enough to out-compute them, use the rentals in high-bandwidth areas to knock as many ASIC clusters that aren't your collaborators as possible offline.
Silly consumer, the risk of 3rd parties undermining the system is why we need to make Trusted Computing mandatory! Secure remote attestation for all, only rights-management-compliant systems allowed on the network!
"If you want to eliminate special client software there, you can have this system, and run everything on the browser."
Being able to reuse more browser code might well improve the existing proprietary client; but (as with any DRM system) you can't eliminate special client software; because you would otherwise be incapable of distinguishing between a web browser that happily accepts your 'HTTPA' and then ignores your restrictions and one that accepts and obeys.
As ever, you'll either need some suitably dense obfuscated blob, or to go all-out-TPM and do secure remote attestation and 'trusted' everything from preboot on up.
Even older than that. The 'Send the data, along with a description of restrictions on the use of the data' is (along with a dash of 'semantic web' nonsense, which appears to be a new addition), classic "Trusted Computing".
Mark Stefik, and his group at Xerox PARC, were talking about 'Digital Property Rights Language' back in 1994 or so, and by 1998, if not earlier, it had metastasized into a giant chunk of XML. That later mutated into "XrML", the 'Extensible Rights Management Language', which eventually burrowed into MPEG-21/ISO/IEC 21000 as the standard's 'rights expression language'.
Some terrible plans just never entirely die.
My understanding (from TFA and from very peripheral involvement as the support techie in a few situations involving sampled music reproduction(not for performance purposes; but it's actually a fairly popular thing in music education/guided practice, where the computer records you and plays accompaniment for whatever instrument the student is supposed to be learning)) is that the synth-orchestra is midway between a live one and a 'just a recording' background music.
With the more or less exhaustive sampling of all the instruments, you get a MIDI-on-steroids arrangement where you aren't just playing back 'a recording', you can play back any number of instruments, each playing any series of notes, on any number of physical outputs, limited only by your speakers and your software.
I don't know if anybody has done something clever with pointing a Kinect at the conductor and having the 'orchestra' respond, though that'd be a pretty natural thing to take a stab at; but if you are doing the (considerable) work of exhaustive sampling and then just doing a static playback, you are wasting your time. Simply getting a couple of high-quality recordings and playing those back would be easier, faster, and a lot less complex.
Where sampling(along with sophisticated instrument modelling/synthesis, where available) gets interesting is that it can respond in real time, if you have worked out how to measure what it is supposed to respond to.
I don't know how effective this Mr. Goldstein's setup is; but that's the somewhere-between-present-and-not-too-distant-future potential of sampling that makes it even more threatening than mere recording to the live musician. Recordings, however good, are static. Exploiting a library of samples and synthesis models, especially in real time and in response to inputs like 'conductor waving baton' or 'vocal performance' rather than 'experienced audio tech providing input designed for the computer's convenience', is very much a work in progress; but if it can be brought up to spec, it's a great deal closer to 'performance' than it is to 'playback'.
Well, parts of the genre do celebrate 'tradition', which is pretty close to stagnation with higher culture cred; but I imagine that they are also nervous because more or less any endeavor(some fields more radically than others; but all of them to a degree) tend to depend on having some minimum scale, above which economies of scale help make them viable, new people can be trained and educated fast enough that the field's unwritten knowledge(and yes, no matter how you try to document stuff, every field has a body of knowledge that is very difficult to pick up without either arduous rediscovery or learning-while-doing with people who already know) doesn't die off, and so on.
If there is plenty of demand to go around, only the crankiest of purist assholes spend their time whining about damn kids these days, and it's a 'the more, the merrier!' sort of enterprise. If people are dogged by the fear that not only is their bread and butter probably going to disappear; but their entire emotionally-invested-tradition is likely to become nonviable, things get ugly, always have.
It is worth noting that, while money is definitely on the table, odds are good that at least part of the bitterness, and conflict, stems from the fact that only opera levels of money are on the table: Opera as a genre, while it has some 'High Art' old money and classy buildings here and there, faces a dwindling and aging audience, relatively weak interest in younger demographics, and production costs that remain comparatively high.
On the one side, even people who are wholehearted opera enthusiasts, dedicated acting career to it, high enough on the food chain that they don't necessarily face immediate economic pressure, etc. can quite honestly make the argument "Well, outside of a few major cities with sufficient concentrations of potential audience and a few outfits with long histories and some generous patrons, opera will either use electronic orchestras, use (even cheaper and worse) prerecorded soundtracks piped through the speaker system, or simply nonexistent."
At the same time, anyone who has a stake in the continued existence of live orchestras(either as a performer or aesthetically) is probably entirely correct in suspecting that, if almost-as-good-and-substantially-cheaper opera starts to be produced, demand for live orchestra performers will likely decline, probably even faster than it has, and quite possibly to the point where there isn't enough demand to pay for training and ongoing practice, that keeps part-time musicians in decent shape, much less the continued availability of full-time musicians, and the field of live opera will be basically toast.
At least in fields that aren't contracting and quite possibly dying, fights over cost-cutting tend to have slightly better defined battle lines between the mercenaries and the purists.
I realize that anything that can be tenuously connected to 'unions' is intrinsically evil and all (though, while TFA had a couple of quotes from musician unions, obviously against the idea, it had no mention of any union acting against the outfit staging this sampled version, just individual nastygrams sent without any broader consultation or endorsement); but, in an enterprise like theater where individual people, at least stars, are commonly economic actors in their own right, is telling a performer "I find your willingness to replace music with samples corrosive to the medium, you should be ashamed of yourself, and I won't work with you." any different from telling a company "I find your willingness to use suppliers in exploitationstan repulsive, you should be ashamed of yourself, and I won't buy from you."?
Arguably, the 'AI curse' is only partly goalpost-moving by humans obsessed with their special uniqueness.
The other aspect of it (as well a genuinely interesting result of whatever attempt at AI was made) is the discovery that we actually handle much less of what we do in some sort of naive, idealized, "high-level, general-purpose, cognition" than we naively suspect, and rather more in specialized and unconscious mechanisms(this lesson can be learned from the other direction as well: something like face blindness, in otherwise cognitively and visually normal people, simply wouldn't be possible if object recognition were a general-purpose function handled by an 'intelligence' with access to a video stream...) And, even when we aren't crunching stuff unconsciously on quite special-purpose mechanisms, we are sometimes just dumb. Good old Youtube comments would probably allow a babbling Markov chain to pass a Turing test.
There's also the fact that humans are frequently easier to fool than they would like: It's not exactly news that people will see faces on the moon, a known sterile rock, impute emotions (and sometimes entire spirits or deities) to the weather, emotionally bond with tamagotchis and similar nonsense.
Yes, people are likely to dismiss every AI that doesn't end up murdering them all and rendering them for computronium as 'just an expert system'; but it's also arguably the case that using a Turing test to check for AI substantially rests on the assumption, more or less Cartesian and more or less nonsense, that man is first and foremost an abstract 'thinking thing', with some other stuff tacked on that philosophers and mathematicians needn't really worry about.
They also aren't exactly rolling in power in the smartphone sector. Even the most instinctively abusive people and institutions tend to avoid upsetting people when they don't have the power to avoid the consequences.
Whether or not they turn on you later, if they gain power, is a matter of disposition or institutional culture; but making enemies and alienating people when you really can't afford it is something that only the stubborn and the doomed do.
Aside from the odds that running 2.3 may mean that your last security fix was in 2011 (depends on how much the entity that built the image cared), lack of TLS SNI, VPN support, hardware accelerated UI, SELinux, and flash filesystem TRIM support are the major downers.
Application compatibility may or may not be an issue, depending on which ones you want.
At least for contract customers, and often for prepaid, carriers are usually willing to issue a replacement SIM (they might charge 5-10 dollars to do so; but especially for contract customers, why lose a customer over a dinky little crypto IC?) and invalidate the old one.
Not something you want to deal with the hassle of; but not as bad as losing an expensive phone with the SIM in it. (Plus, if your phone is cheap junk, you can still lose/destroy it on your own; but it is less likely to be stolen, which helps increase effective reliability, since most thieves aren't polite enough to pull the SIM before lifting the handset.) As for environmental hazards, SIMs are pretty tough. Unless you physically snap one, give it a particularly stiff hit of ESD, or microwave it, it'll probably be fine. A little salt water, your phone falling a couple of stories onto concrete, and similar mishaps are unlikely to do it much harm.
Even if it doesn't, my understanding is that it runs some sort of Linux kernel, with enough juice for a mostly-HTML/JS GUI to be endurable, plus it obviously includes at least some baseline GSM/GPRS (quite possibly a later standard; but it'd be hard to call it much of a phone without that), which is a notable perk for certain projects.
Interfacing a GSM module with anything from a microcontroller on up isn't all that hard (Hayes Ain't Dead, just extended a lot...); but a bare SIM900 module will run you a trifle above $10 on its own, more for something a little less spartan with a SIM slot and convenient headers. (the SIM900A is a bit cheaper; but only dual band, rather than quad, which can be an issue depending on where you live). Obviously, you can do considerably better if you score a working pull from a dead laptop or something, and a fancy arduino-shield configuration with all the connectors and such soldered on can run you ~$20 alone; but the moral of the story is that if you want to do some sort of embedded project that uses SMS or GPRS in some way, it's getting pretty hard to save money by using an add-on module for your choice of teeny computer if you can, instead, get a low-end linuxized phone that is suitably cooperative about software customization. The only real exception would be in situations where the lack of GPIO and similar microcontroller standard features is an issue, since phones tend to be fairly sparse on that score.
It's getting to be the same way with screens. Unless you are doing some 16x4 character display, or a Nokia 1100 LCD clone, or going all the way up to a cheap 'n cheerful 17 inch monitor, you can have some difficulty sourcing (and driving) moderate size LCDs for less money than you could spend to get the LCD and an entire computer, with battery, attached.
They may not make it here at $25 (a cynic would note that this is hardly the first story involving some Indian outfit that was allegedly going to work wonders in low cost electronics, who then dithers and flakes out, increasing prices, slipping schedules, or trimming specs, until, mysteriously, they eventually manage to deliver something or other, right about the time that Chinese mystery-brands are selling them on ebay for the same money, Slashdot has had at least a couple of such stories involving tablets for education); but the route taken by cheap 'n cheerful electronics to the US appears to be fairly swift and not too marked up(as long as you don't pay Belkin or somebody to write an English manual and slap their sticker on it). The US isn't a 'How do they hit that price point even with counterfeit parts and child labor? miracle of cheapness; but this isn't Brazil, or one of those lucky European countries where the price is the same as in the US, except that they want that many pounds or that many euros...
My typo, I'm afraid (I meant Mountain View, though Windows Phone devices are tied pretty closely to Redmond); but Android solves the problem rather less than it pretends to.
Unless you hack around and run bare AOSP, possibly with certain 3rd-party customizations, the percentage of 'Android' that is actually 'Google Play Services' increases with every version bump. A nontrivial percentage of even non-Google apps also build against Google-specific APIs, rather than the relatively impoverished Android ones (the rule of thumb seems to be that, once a role is added to GPS, the AOSP implementation more or less freezes at whatever state it was in and remains there), so incompatibility, even with the absolute freshest AOSP, is quite common.
Google has the right to do as they wish with their OS, and at least you can buy it on handsets that aren't crypto locked to hell and back; but you'll be talking to the mothership soon enough.