Extensions to the browser? This used to be called "programs you install on your computer". Weird.
I'm pretty sure that 'extensions' 'plugins' and the like have been recognized as distinct from full 'programs'(even if, in practice, the line is architecturally somewhat blurry, depending on how elegant the extension mechanism provided by the main program is).
I know that I was using Winamp plugins well before I had a remotely reliable internet connection, and most sufficiently sprawling programs tend to spawn a plugin or extension mechanism sooner or later. In fact, some are barely more than a default set of plugins lodged in a labyrinthine mechanism for adding more(looking at you, Lotus Notes...)
It's the failure to analyze risk with some degree of perspective and avoidance of hysteria that really annoys me. Many aspects of the good old days were, in fact, bad (eg. the fact that the went around huffing tetraethyl lead combustion products basically all the time as though that were a good idea). Others offend the squeamish but start to look pretty harmless when you ask around about any actual, verified, injuries that they are supposed to have caused.
Mindless celebration of risk as some sort of virtue is puerile nonsense; but flinching before you've even determined that a risk is real is even worse.
The fact that it is even called 'the most dangerous toy' is evidence that some people need a serious beating.
Yeah, sure, "Radiation!!!" is scary; but low level sources are pretty tepid unless you do your best to consume them(and sometimes even then) and 'uranium ore' can(depending on source and quality) be about as geiger-counter-clicking as a nice, safe, granite building. It's trivially apparent that any toy that constituted a decent choking hazard was orders of magnitude more hazardous.
I'm not privy to the details of the implementation; but it sounds like they are aiming to solve that problem by just brutalizing the permutation space. It's probably not as good as a really good human; but hand an expert system a thesaurus and it can probably spew 'equivalent' implementation claims at an unbelievable pace, even if many of them are of low quality or not really worth mentioning.
Quantity has a quality all its own, as one Mr. Stalin is alleged to have said.
It looks like a very effective denial of service attack, to me. No law school is shitty enough to puke out patent examiners that cost less per hour than some glorified Markov chain, so barring some sort of rate limiting, with teeth, you should be able to shove a lot of utter shit through by sheer brute force.
Even better, if the patent examiners don't just crack under the strain, then they do the job of distinguishing vaguely worthwhile patents from algorithmic word salad for you!
Isn't the whole family of alloys we call 'steel' essentially a testament to the fact that certain nanoscale structures(that, conveniently for us, can be produced by comparatively primitive methods) in iron can radically improve its properties compared the the (actually pretty lousy) pure metal?
Evolutionary pressure ultimately acts on genes; but (especially among species where 'horizontal gene transfer' is a weak sex joke, rather than a routine genomic reshuffling strategy), a lot of selection happens to organism-level bundles of genes, with all of them going down with the ship at the same time.
I wonder if this has any relation to the role of 'spite', which is sort of altruism's barbed cousin: (for my purposes, 'spite' in the sense of 'inflicting injury on someone else without benefit, or even at direct cost, to oneself')?
Spite is hardly a directly rational response, especially if you are inflicting injury as a reprisal for something that somebody has done to a third party, rather than to you; but it is clearly something that humans do(or, even if they don't, they often fume indignantly and wish that they could); and seems that it could provide additional incentive to behave altruistically, or at least within the bounds of 'fairness'
"AT&T may collect and use web browsing information for other purposes, as described in our Privacy Policy, even if you do not participate in the Internet Preferences program."
So, there's the $100/month 'Yup, definitely spying on you' tier where "your Internet traffic is routed to AT&T's Internet Preferences web browsing and analytics platform"(good luck finding out exactly what that entails; but it's probably bad); or the $70/month 'Ominous and vague "other purposes"' tier.
How much evil do they manage into their 'browsing and analytics platform' to be $30 worse than their baseline level of spying?
Nuclear power plants generate net positive energy(or they wouldn't be power plants at all); but they still have numerous systems(from relatively important sensors, control systems, pumps, etc. down to boring stuff like the bathroom lights) that need to be powered to work properly. They do have backup generators on site; but, for reasons of safety(and because it's hard to sell the output of a plant that isn't connected to the grid) you aren't supposed to run them when they have been disconnected or are expected to be disconnected.
Nuclear plants are a trifle more dramatic, because shutting them down isn't just a matter of not shovelling more coal into the furnace; but it's my understanding that no power plant types are really supposed to be run off grid. Even a perfectly well behaved design is still a waste of money if the power can't reach customers, and nobody likes depending on backup power to keep every electric device in the facility online.
They might well need to review the qualify of the grid connection, if only because an idling nuclear reactor is a punchy opportunity cost; but there isn't much reason to leave it on. (Unless you managed to get the NRC drunk and obtain approval for a madcap scheme to convert the reactor to a snow-melter for the duration of the storm. We really are starting to run out of places to put the stuff, and a nuclear reactor would be very well qualified for melting duties.)
On the plus side, early acclimatization to tapping symbol-coated touchscreens may be actively harmful; but it does prepare you to adeptly navigate the POS system of any fast food chain, giving you a leg up on one of the undesirable, economically tenuous, careers of the future!
It's particularly 'optimistic' given that we already have a fair amount of experience with what does (and doesn't) happen to children with access to books. With comparatively rare exceptions, mostly in slightly older children, not all that much.
It's pretty obvious that networked computers are, sooner or later, going to beat out printed textbooks(if only because it's getting cheaper to transmit a few megabytes to the ass end of nowhere than it is to ship tens of kilograms there, not necessarily because they are better, especially with the hardware in the cheap seats).
It is radically less obvious that our tiny monkey-spawn, with their few-hundred-million-years of experience in absorbing knowledge into their sponge-like brains by demanding interaction with nearby group members, are on the cusp of successfully being tutored by expert systems with some animated cartoon characters tossed on top.
If the trouble with teaching were a matter of text scarcity, Gutenberg would have mostly kicked its ass. That's not exactly what happened.
It certainly appears that the cheap gear is crap; but it isn't actually obvious why such hardware should be expensive, save that the assorted rushed-to-market new entrants are immature and terrible.
These aren't machine tools or mechanical watches or something, where high quality materials and precision workmanship are Just Plain Expensive, take it or leave it. It's all dirt-cheap-and-even-cheaper-tomorrow commodity silicon running bad software. With maturity, the floor price of implementing a basic control SoC competently enough that it isn't usually hardware faults that cause the software to crash should be relatively trivial. Inconveniently, maturity is approximately as far away as fusion power when it comes to the companies in the cheap seats.
Given that 'dumb' thermostats are very good at what they do, just not very flexible, the savings of going 'smart' for heating/cooling depend pretty substantially on how much you do or don't bother to adjust thermostats manually and how erratic your comings and goings are(as well as how much control the climate in your area requires, obviously).
A good old mercury-and-bimetallic strip device can keep the temperature stable about as well as anything else; but you have to adjust it. A cheapy digital unit will have basic support for 'morning/day/evening/weekday/weekend' time programs, a 'smart' unit will hopefully be more elegant and accurate in building a schedule. If you made frequent manual adjustments to a conventional thermostat, or have such a predictable schedule that a rudimentary digital unit can capture it, the change will be relatively minimal. If less sophisticated gear was doing a bad job of capturing your actual demand schedule, you might save a bundle.
It's not as though the current offerings leave you with much of a choice; but 'smart' is never going to be worth it if it is merely a high tech re-implementation of what you can already do with a few bucks worth of mechanical switches. Even if it works flawlessly, it's still going to be expensive and unexciting.
The only real shot for 'smart' is to do things that conventional systems cannot or do not. Exactly what those things are is a bit vague(lighting and drapes that automatically adjust to available sunlight? automatic dimming when you fire up the TV? subtle color temperature modifications to facilitate greater alertness or easier sleep depending on time of day?); but unless they figure those out, there simply isn't any any way that 'smart' could possibly be worth the trouble. If they do, then we can talk; but 're-implementation of a light switch by dragging an entire wireless LAN and more computing power than existed on earth in the early 80s' is just dumb, even if you polish it properly.
There is no direct causal connection, as you say, embedded security has been pretty much crap for ages, particularly in the cheap seats; but it is the case that 'IoT' manages to combine a disturbing enthusiasm for giving anything and everything firmware and an IP address with a security record at least as slapdash and atrocious, if not more, as other low-end embedded vendors, which makes them a particularly messy case.
Why would they shut it down? Clearly this 'feature' is just there to help more things connect themselves to the IoT without inconveniencing the consumer by bothering them for a password!
This seems like a rather pointless question, since 'reviews' and 'review scores' serve somewhat different purposes.
If you want a comparatively deep examination of a game, strengths, weaknesses, what is it trying to do?, does it succeed?, who is it aimed at?, etc. an answer like "65" or "8" is practically useless. If you want to do a metacritic-style survey(or decide what long-form reviews to read when faced with 2,000 games), though, 3 pages of prose and musing, each, from two dozen sources isn't going to cut it.
Anyone pretending that a hundred-point score is actually that precise is likely fooling themselves; but there's a much stronger argument that you can get at least a 1-10 or so scoring system unless you are a pure, handwaving "It's all, like, intersubjective, man..." type.
I'd be very interested to see how such a design would stack up in practice(and, one assumes, with practice).
With an ordinary keyboard, your fingers do have to move; but each keystroke provides multiple bits worth of input(exactly how many varying with input language and a bunch of other fiddly stuff).
Would you be able to cycle all ten fingers fast enough to achieve the same number of effective bits per minute, or would the advantage of having multiple-bit input in a single keypress(and keys chosen to be the most commonly used multiple bit sequences) outweigh the need to move a bit to reach them all?
A test involving lots of freaky unicode that doesn't have dedicated keys would allow the bit-based layout to utterly crush an ordinary board; but a test weighted toward the probabilities of various characters in the language the keyboard was designed for might well go the other way.
Obviously, some details of a keyboard's operation must be mechanical(ergonomics, key travel, the Absolute and Unquestionable Superiority of Buckling Spring Designs, etc.); but other details can be addressed at the hardware level, the firmware level, or the OS/userland level. Tons of function keys, say: you can physically add the additional buttons for all sorts of functions(volume control, start/stop/play, application specific shortcuts and macros), you can have some sort of firmware-level capability for assigning unusual keycodes or keycode sequences to specific keys or modifier key+key combinations, or you can have the keyboard remain relatively dumb, with most of the work happening at the level of software on the host machine(as with IMEs that use English keyboards to handle input of East Asian languages, or programs that execute complex macros on demand).
In your ideal world, how would you divide the work up? Eat the extra space and cost, and break out a ton of dedicated buttons? Endure the rather limited user interface that keyboard firmware can allow, in order to be able to do complex macros, custom keycodes, and other fun stuff even in the BIOS, on another computer, in an OS without software support? Go for a fairly minimal-complexity keyboard; but use software that provides a great deal of power from the keycodes you can generate?
There's also the concern for legitimate 3rd-party firmware. There likely are ways to build fairly robust kill switches without destroying the ability to replace (most of) the firmware(you'd presumably need to have the small amount that stores the kill bit and validates the unlock code be hard to replace; but everything on top of that could still be replaceable); but they aren't necessarily the easiest ways, or the ways that manufacturers who don't necessarily much care about 3rd party firmware, or even actively dislike it, will choose.
It's analogous to 'secure boot' in the PC space. In principle, an OEM that cares can build a 'secure boot' implementation that gives the user full control over keying, root of trust, etc. (and, if memory serves, some corporate-focused models do offer the ability to slave the 'secure boot' stuff to your own in-house PKI setup); but it's a whole hell of a lot easier just to burn Microsoft's keys into the firmware and call it a day. With phones, the main risk is probably all the OEMs who have historically not given a damn, and implemented weak or nonexistent bootloader lockdown; but, if mandated, will respond with stronger, equally apathetic, lockdown instead.
Cryptographic security mechanisms are a good thing; but it's usually easier to build ones that are also control/lockdown mechanisms than it is to ensure that the user retains control, especially if you are targeting unsophisticated users.
Given the proportion of the value that is represented in the(almost invariably fused together in newer and higher end models) LCD/OLED and capacitive sensor, parting out seems like a fairly pragmatic strategy, especially when that's the part that everybody breaks, so the demand is there. Not quite as trivial as wipe 'n flip; but not substantially greater sophistication, and the skills and tools are common and not in themselves illicit or suspicious.
Any word on whether more sophisticated 'parting'/remanufacturing is going on as well? It's not something you'd want to do as a one-off, or without proper tools; but for a 400-600 dollar device, it seems easily conceivable that actual board-level rework(especially for phone models where the silicon where the 'kill switch' data actually live is available on the open market) to produce working and unlocked units from a mixture of stolen and locked, licit but damaged, and new or salvaged ICs would be economically viable.
That, and rats are better at trench warfare than humans will ever be, or would ever want to be. Plus, they are alarmingly good(by mammalian standards, nobody has anything on the unicellular guys in this area) at resisting poisoning attempts. Wolves, by all accounts, are quite vulnerable to the M44 and conceptually similar chemical traps.
Deer, on the other hand, are successful as a pest species almost entirely because of people getting jumpy about gunnmen running around the suburbs and shooting them.
With the larger mammals, we've actually done so well that they are either extinct, domesticated, or have to be legally protected as endangered species. It's the smaller ones that are pretty much unstoppable. Compare wolves to rats, say.
In the case of advertisers, their metabolic and phenotypic similarity to humans would afford them a degree of concealment in settled areas and make mass poisoning impractical; but they don't reproduce particularly quickly or have large litter sizes and are large enough to hunt from aircraft, so I suspect control would be possible.
Slashdot Mirror
Extensions to the browser? This used to be called "programs you install on your computer". Weird.
I'm pretty sure that 'extensions' 'plugins' and the like have been recognized as distinct from full 'programs'(even if, in practice, the line is architecturally somewhat blurry, depending on how elegant the extension mechanism provided by the main program is).
I know that I was using Winamp plugins well before I had a remotely reliable internet connection, and most sufficiently sprawling programs tend to spawn a plugin or extension mechanism sooner or later. In fact, some are barely more than a default set of plugins lodged in a labyrinthine mechanism for adding more(looking at you, Lotus Notes...)
It's the failure to analyze risk with some degree of perspective and avoidance of hysteria that really annoys me. Many aspects of the good old days were, in fact, bad (eg. the fact that the went around huffing tetraethyl lead combustion products basically all the time as though that were a good idea). Others offend the squeamish but start to look pretty harmless when you ask around about any actual, verified, injuries that they are supposed to have caused.
Mindless celebration of risk as some sort of virtue is puerile nonsense; but flinching before you've even determined that a risk is real is even worse.
The fact that it is even called 'the most dangerous toy' is evidence that some people need a serious beating.
Yeah, sure, "Radiation!!!" is scary; but low level sources are pretty tepid unless you do your best to consume them(and sometimes even then) and 'uranium ore' can(depending on source and quality) be about as geiger-counter-clicking as a nice, safe, granite building. It's trivially apparent that any toy that constituted a decent choking hazard was orders of magnitude more hazardous.
I'm not privy to the details of the implementation; but it sounds like they are aiming to solve that problem by just brutalizing the permutation space. It's probably not as good as a really good human; but hand an expert system a thesaurus and it can probably spew 'equivalent' implementation claims at an unbelievable pace, even if many of them are of low quality or not really worth mentioning.
Quantity has a quality all its own, as one Mr. Stalin is alleged to have said.
It looks like a very effective denial of service attack, to me. No law school is shitty enough to puke out patent examiners that cost less per hour than some glorified Markov chain, so barring some sort of rate limiting, with teeth, you should be able to shove a lot of utter shit through by sheer brute force.
Even better, if the patent examiners don't just crack under the strain, then they do the job of distinguishing vaguely worthwhile patents from algorithmic word salad for you!
Isn't the whole family of alloys we call 'steel' essentially a testament to the fact that certain nanoscale structures(that, conveniently for us, can be produced by comparatively primitive methods) in iron can radically improve its properties compared the the (actually pretty lousy) pure metal?
Evolutionary pressure ultimately acts on genes; but (especially among species where 'horizontal gene transfer' is a weak sex joke, rather than a routine genomic reshuffling strategy), a lot of selection happens to organism-level bundles of genes, with all of them going down with the ship at the same time.
I wonder if this has any relation to the role of 'spite', which is sort of altruism's barbed cousin: (for my purposes, 'spite' in the sense of 'inflicting injury on someone else without benefit, or even at direct cost, to oneself')?
Spite is hardly a directly rational response, especially if you are inflicting injury as a reprisal for something that somebody has done to a third party, rather than to you; but it is clearly something that humans do(or, even if they don't, they often fume indignantly and wish that they could); and seems that it could provide additional incentive to behave altruistically, or at least within the bounds of 'fairness'
"AT&T may collect and use web browsing information for other purposes, as described in our Privacy Policy, even if you do not participate in the Internet Preferences program."
So, there's the $100/month 'Yup, definitely spying on you' tier where "your Internet traffic is routed to AT&T's Internet Preferences web browsing and analytics platform"(good luck finding out exactly what that entails; but it's probably bad); or the $70/month 'Ominous and vague "other purposes"' tier.
How much evil do they manage into their 'browsing and analytics platform' to be $30 worse than their baseline level of spying?
Nuclear power plants generate net positive energy(or they wouldn't be power plants at all); but they still have numerous systems(from relatively important sensors, control systems, pumps, etc. down to boring stuff like the bathroom lights) that need to be powered to work properly. They do have backup generators on site; but, for reasons of safety(and because it's hard to sell the output of a plant that isn't connected to the grid) you aren't supposed to run them when they have been disconnected or are expected to be disconnected.
Nuclear plants are a trifle more dramatic, because shutting them down isn't just a matter of not shovelling more coal into the furnace; but it's my understanding that no power plant types are really supposed to be run off grid. Even a perfectly well behaved design is still a waste of money if the power can't reach customers, and nobody likes depending on backup power to keep every electric device in the facility online.
They might well need to review the qualify of the grid connection, if only because an idling nuclear reactor is a punchy opportunity cost; but there isn't much reason to leave it on. (Unless you managed to get the NRC drunk and obtain approval for a madcap scheme to convert the reactor to a snow-melter for the duration of the storm. We really are starting to run out of places to put the stuff, and a nuclear reactor would be very well qualified for melting duties.)
On the plus side, early acclimatization to tapping symbol-coated touchscreens may be actively harmful; but it does prepare you to adeptly navigate the POS system of any fast food chain, giving you a leg up on one of the undesirable, economically tenuous, careers of the future!
It's particularly 'optimistic' given that we already have a fair amount of experience with what does (and doesn't) happen to children with access to books. With comparatively rare exceptions, mostly in slightly older children, not all that much.
It's pretty obvious that networked computers are, sooner or later, going to beat out printed textbooks(if only because it's getting cheaper to transmit a few megabytes to the ass end of nowhere than it is to ship tens of kilograms there, not necessarily because they are better, especially with the hardware in the cheap seats).
It is radically less obvious that our tiny monkey-spawn, with their few-hundred-million-years of experience in absorbing knowledge into their sponge-like brains by demanding interaction with nearby group members, are on the cusp of successfully being tutored by expert systems with some animated cartoon characters tossed on top.
If the trouble with teaching were a matter of text scarcity, Gutenberg would have mostly kicked its ass. That's not exactly what happened.
It certainly appears that the cheap gear is crap; but it isn't actually obvious why such hardware should be expensive, save that the assorted rushed-to-market new entrants are immature and terrible.
These aren't machine tools or mechanical watches or something, where high quality materials and precision workmanship are Just Plain Expensive, take it or leave it. It's all dirt-cheap-and-even-cheaper-tomorrow commodity silicon running bad software. With maturity, the floor price of implementing a basic control SoC competently enough that it isn't usually hardware faults that cause the software to crash should be relatively trivial. Inconveniently, maturity is approximately as far away as fusion power when it comes to the companies in the cheap seats.
Given that 'dumb' thermostats are very good at what they do, just not very flexible, the savings of going 'smart' for heating/cooling depend pretty substantially on how much you do or don't bother to adjust thermostats manually and how erratic your comings and goings are(as well as how much control the climate in your area requires, obviously).
A good old mercury-and-bimetallic strip device can keep the temperature stable about as well as anything else; but you have to adjust it. A cheapy digital unit will have basic support for 'morning/day/evening/weekday/weekend' time programs, a 'smart' unit will hopefully be more elegant and accurate in building a schedule. If you made frequent manual adjustments to a conventional thermostat, or have such a predictable schedule that a rudimentary digital unit can capture it, the change will be relatively minimal. If less sophisticated gear was doing a bad job of capturing your actual demand schedule, you might save a bundle.
It's not as though the current offerings leave you with much of a choice; but 'smart' is never going to be worth it if it is merely a high tech re-implementation of what you can already do with a few bucks worth of mechanical switches. Even if it works flawlessly, it's still going to be expensive and unexciting.
The only real shot for 'smart' is to do things that conventional systems cannot or do not. Exactly what those things are is a bit vague(lighting and drapes that automatically adjust to available sunlight? automatic dimming when you fire up the TV? subtle color temperature modifications to facilitate greater alertness or easier sleep depending on time of day?); but unless they figure those out, there simply isn't any any way that 'smart' could possibly be worth the trouble. If they do, then we can talk; but 're-implementation of a light switch by dragging an entire wireless LAN and more computing power than existed on earth in the early 80s' is just dumb, even if you polish it properly.
There is no direct causal connection, as you say, embedded security has been pretty much crap for ages, particularly in the cheap seats; but it is the case that 'IoT' manages to combine a disturbing enthusiasm for giving anything and everything firmware and an IP address with a security record at least as slapdash and atrocious, if not more, as other low-end embedded vendors, which makes them a particularly messy case.
Why would they shut it down? Clearly this 'feature' is just there to help more things connect themselves to the IoT without inconveniencing the consumer by bothering them for a password!
This seems like a rather pointless question, since 'reviews' and 'review scores' serve somewhat different purposes.
If you want a comparatively deep examination of a game, strengths, weaknesses, what is it trying to do?, does it succeed?, who is it aimed at?, etc. an answer like "65" or "8" is practically useless. If you want to do a metacritic-style survey(or decide what long-form reviews to read when faced with 2,000 games), though, 3 pages of prose and musing, each, from two dozen sources isn't going to cut it.
Anyone pretending that a hundred-point score is actually that precise is likely fooling themselves; but there's a much stronger argument that you can get at least a 1-10 or so scoring system unless you are a pure, handwaving "It's all, like, intersubjective, man..." type.
I'd be very interested to see how such a design would stack up in practice(and, one assumes, with practice).
With an ordinary keyboard, your fingers do have to move; but each keystroke provides multiple bits worth of input(exactly how many varying with input language and a bunch of other fiddly stuff).
Would you be able to cycle all ten fingers fast enough to achieve the same number of effective bits per minute, or would the advantage of having multiple-bit input in a single keypress(and keys chosen to be the most commonly used multiple bit sequences) outweigh the need to move a bit to reach them all?
A test involving lots of freaky unicode that doesn't have dedicated keys would allow the bit-based layout to utterly crush an ordinary board; but a test weighted toward the probabilities of various characters in the language the keyboard was designed for might well go the other way.
Do Not Blaspheme. The Model M is the closest thing to the platonic essence of 'Keyboard' that mankind has ever managed to produce.
All unbelievers shall spend eternity doing data-entry using T9 text input on an aging dumbphone with intermittent key response.
Obviously, some details of a keyboard's operation must be mechanical(ergonomics, key travel, the Absolute and Unquestionable Superiority of Buckling Spring Designs, etc.); but other details can be addressed at the hardware level, the firmware level, or the OS/userland level. Tons of function keys, say: you can physically add the additional buttons for all sorts of functions(volume control, start/stop/play, application specific shortcuts and macros), you can have some sort of firmware-level capability for assigning unusual keycodes or keycode sequences to specific keys or modifier key+key combinations, or you can have the keyboard remain relatively dumb, with most of the work happening at the level of software on the host machine(as with IMEs that use English keyboards to handle input of East Asian languages, or programs that execute complex macros on demand).
In your ideal world, how would you divide the work up? Eat the extra space and cost, and break out a ton of dedicated buttons? Endure the rather limited user interface that keyboard firmware can allow, in order to be able to do complex macros, custom keycodes, and other fun stuff even in the BIOS, on another computer, in an OS without software support? Go for a fairly minimal-complexity keyboard; but use software that provides a great deal of power from the keycodes you can generate?
There's also the concern for legitimate 3rd-party firmware. There likely are ways to build fairly robust kill switches without destroying the ability to replace (most of) the firmware(you'd presumably need to have the small amount that stores the kill bit and validates the unlock code be hard to replace; but everything on top of that could still be replaceable); but they aren't necessarily the easiest ways, or the ways that manufacturers who don't necessarily much care about 3rd party firmware, or even actively dislike it, will choose.
It's analogous to 'secure boot' in the PC space. In principle, an OEM that cares can build a 'secure boot' implementation that gives the user full control over keying, root of trust, etc. (and, if memory serves, some corporate-focused models do offer the ability to slave the 'secure boot' stuff to your own in-house PKI setup); but it's a whole hell of a lot easier just to burn Microsoft's keys into the firmware and call it a day. With phones, the main risk is probably all the OEMs who have historically not given a damn, and implemented weak or nonexistent bootloader lockdown; but, if mandated, will respond with stronger, equally apathetic, lockdown instead.
Cryptographic security mechanisms are a good thing; but it's usually easier to build ones that are also control/lockdown mechanisms than it is to ensure that the user retains control, especially if you are targeting unsophisticated users.
Given the proportion of the value that is represented in the(almost invariably fused together in newer and higher end models) LCD/OLED and capacitive sensor, parting out seems like a fairly pragmatic strategy, especially when that's the part that everybody breaks, so the demand is there. Not quite as trivial as wipe 'n flip; but not substantially greater sophistication, and the skills and tools are common and not in themselves illicit or suspicious.
Any word on whether more sophisticated 'parting'/remanufacturing is going on as well? It's not something you'd want to do as a one-off, or without proper tools; but for a 400-600 dollar device, it seems easily conceivable that actual board-level rework(especially for phone models where the silicon where the 'kill switch' data actually live is available on the open market) to produce working and unlocked units from a mixture of stolen and locked, licit but damaged, and new or salvaged ICs would be economically viable.
That, and rats are better at trench warfare than humans will ever be, or would ever want to be. Plus, they are alarmingly good(by mammalian standards, nobody has anything on the unicellular guys in this area) at resisting poisoning attempts. Wolves, by all accounts, are quite vulnerable to the M44 and conceptually similar chemical traps.
Deer, on the other hand, are successful as a pest species almost entirely because of people getting jumpy about gunnmen running around the suburbs and shooting them.
With the larger mammals, we've actually done so well that they are either extinct, domesticated, or have to be legally protected as endangered species. It's the smaller ones that are pretty much unstoppable. Compare wolves to rats, say.
In the case of advertisers, their metabolic and phenotypic similarity to humans would afford them a degree of concealment in settled areas and make mass poisoning impractical; but they don't reproduce particularly quickly or have large litter sizes and are large enough to hunt from aircraft, so I suspect control would be possible.