These accusations of coercion are blatantly defamatory. We are simply offering incentives, which the interns, as free and rational agents, are choosing to accept or decline. It's practically a libertarian utopia, trade among men, as equals, free from the dead hand of state power. Anyone who says otherwise is probably some sort of commie, who thinks that labor and capital negotiate from positions of unequal strength or some bullshit like that.
I suspect that it's partially inertia/penny-pinching and partially because crypto only solves certain (quite specific) problems within the larger problem of 'run an election'.
For instance, in those countries that have smartcard-or-equivalent national IDs, cryptographically signed votes would be trivial; but you'd be reusing keys explicitly designed to not be anonymous, indeed, designed to be identifying. That is an issue. Beats some 'SSN+Mother's maiden name' bullshit; because at least it verifies something; but it isn't what you are looking for.
If you anonymously issue keys, now you've got a weak spot there that crypto can't help you with: the crypto makes it quite possible to ensure that Anon_Key_X was responsible for Vote_X, and only Vote_X; but you still need to devise a system by which an eligible voter can obtain (without some absurd hassle) one and only one anonymous key, without it being covertly linked back to them, or them being able to sign up for ten, or the people running the system being able to generate 250,000(or simply keep a copy of the keys as they are issued, and 'win the race' to get a signed ballot into the pot with that key).
If you have such a system, you also have a system that could trivially just hand the voter a ballot, since you have already satisfied anonymity, uniqueness, resistance to plural voting, etc. No need for the crypto at all.
(Also, aside from that, a country with vote rigging tendencies is presumably going to use hierarchical PKI, not some web-of-trust cypherpunk wet dream, so what exactly will an election whose ballots are signed with keys that all descend from the 'Glorious Cryptographic Key for Make Benefit of People's Republic Motherland' prove? Hierarchical PKI schemes, as SSL has taught us, work OK if you are primarily concerned with criminals and frauds; but if the CA is the enemy, you are fucked. If you are the root, you can generate mathematically pristine child keys as fast as your little ASICs can carry you without the slightest trouble.)
Oh, I'm not saying that the correct answer is Java, just that choosing a language that appears to be actively dying (rather than just choosing a pedagogical language without listening to the 'But, but, practical skills!' idiots) seems like it would add an extra layer of logistical hassle.
Given the state of South African secondary education, 'extra layer of logistical hassle' is not high on the list of things I would want.
If memory serves, both the US NRC (presumably because GE was involved, or because somebody asked) and their Japanese equivalent pointed out that there were... issues... with various aspects of the facility(including flooding-related ones that ended up being a problem, in addition to retro reactor design problems that they didn't know about); but nobody likes that guy who can't be a team player, and so such issues were not addressed.
Not bad by fire alarm standards ( you see quoted lifespans from 5-10 years usually, with shorter ones for very safety critical carbon monoxide sensors), it just seems kind of churlish not to have the $10-$15ish commodity parts in a swappable FRU, just like the batteries...
It's entirely possible that much of 'they' weren't really asked. If the decision-making around Fukishima (when it was first installed) was anything like that around Minamata, stonewalling and general contempt for the 'opinions' of whiny little people was the official policy of both business and state.
"Energy" is an awfully persuasive argument. Without it, your civilization will crumble to little more than medieval standards of living surprisingly fast.
Given that Japan has ~0 coal, oil, gas (not even much wood, per capita), it was either nukes or imports. Much of the rest of the world hasn't had to face up to the problem as dramatically because they've got a big stash of cheap 'n nasty coal somewhere convenient. That has its own downsides; but those have proven easy to ignore (unless you make the mistake of living in our under-construction Appalacian Lunar Theme Park or something).
It does seem increasingly clear that TEPCO couldn't be trusted to take care of your fishtank for a weekend(not even the freshwater one that's really relaxed about sampling and balance adjustments); but it should be noted, in fairness, that wacky piping accidents do get easier the more thickly built (ideally ad-hoc, and in poorly labelled stages, with evolutionary growth here and there) the piping rat's nest gets.
The essence of true competence is to avoid getting into situations where continuous high levels of competence are needed; by not backing yourself into a clusterfuck of a system that is always one false move away from doing something dangerous; but if you've fucked up and done that, it's really just a matter of time until somebody gets tapped as the fall guy by the pitiless gods of blind chance.
Let us hope that the work they do there comes to nothing, and is overwhelmingly forgotten. It's a more charming approach than simple brute force and lobbying; but this is also an extension of Verizon's work to stave off becoming a dumb pipe, and reap the rents that doing so allows. That isn't good news for anybody except them, and possibly their favored app and/or device buddies.
Not only are they surprisingly bad at it (hands up, everyone who ever had the pleasure of a phone with a fully Verizoned ROM, or a Comcast-rented cable box UI); but the conflicts of interest inherent between offering a product or service and controlling the infrastructure over which that offering is delivered are irreconcilably dangerous.
I think that they get enough enjoyment out of enforcing those that they don't count as 'burdensome' (except for the executee; but nobody cares about him).
If not connectivity through a phone, then connectivity through a replacable module. Decoupling the input and output make this easy and easily upgradable. It also lets people pick their own services. I like using Google services myself, but someone else may not. There may be a new high speed comminications technology out in the near future. Decoupling aall of these thimgs makes it much more upgradable, and if they use open connectors/protocols, then you can use your phone to replace or supplement this functionality as well.
Like poor old ISO 7736, which doesn't seem to have been doing all that well in recent models, alas?
Which makes it particularly odd to choose some obscure oddball with a tiny handful of necrotic vendors and a few under-maintained OSS projects as options for tools, rather than an otherwise-equivalent one for which ongoing interest exists.
Quite possibly. 55 inches, especially if ostensibly shared, seems too large to benefit significantly. I'd put the sweet spot at around 30"... small enough that they sell ones you are supposed to sit close to, large enough that the corners are really significantly further from your eyes than the center.
Very little as apparently the article thinks the NRC is responsible for foiling terrorist plots to go after nuclear reactors.
Personally, I'm more worried about increased negligence from operators without somebody breathing down their necks than I am about terrorists.
(The most recent example, luckily nonnuclear, being the juxtaposition between the marathon bombers and the West Fertilizer company. Kill three people with a backpack full of explosives and all of greater Boston goes full tactical on you. Blow up 500,000 pounds of ammonium nitrate, killing 15 and leveling a good portion of the nearby town? Eh, we try to avoid burdensome regulations here in Texas...)
I have to imagine that any significant curve on a screen that small would end up looking sort of weird, though.
At somewhere in the 24-30 inch range (for monitors, larger for TVs that you are expected to sit further away from) I can see curvature as being a good thing, so that all parts of the screen are roughly equidistant from your face, rather than the edges requiring refocusing as you move your eyes around; but that won't be at play on a phone. If they keep it subtle you'll probably adjust pretty quickly, though.
WTF! At least a thermostat actually does something worth making it programmable for... this is just a ridiculously overengineered implementation of "if fire, make noise."
The fact that both photoelectric smoke detectors and carbon monoxide detectors (especially the latter) have finite lifespans (as little as five years, less if certain sensor poisons are present) and neither component is separately replaceable makes the $130 price tag extra hard to swallow...
In practice, 'Version XYZ of Open Standard Whatever, with certain bugs and idiosyncrasies' is 'your API' if you go the open standards route. Over time, the really nasty stuff gets ironed out; but as long as it's "Well, we needed an API and there wasn't one, so we made one, here it is." there isn't necessarily anything wrong with that (in fact, more than a few now-open standards exist because somebody was first to need an API, so they Just Did It and then let(or couldn't stop) standardization from happening around that. Like Hayes-command-set modems.
Now, if Nest means 'their' API in the sense that you have to sign a EULA in blood to write a program that talks to your own hardware (obviously, if an API is a mechanism for accessing resources on somebody else's system, they can set whatever access rules they like; because they own that computer; but if they claim to own the API per se and control all uses of it, We Have A Problem.) they can go suck it; but if they just mean 'Our API, the API that we use and implement in our products' no big deal.
Oh, I'd be the last to claim that most 'retro' computing projects make a whole lot of sense, except perhaps as pedagogical exercises (I have a personal fondness for 'retro' projects involving actual old-school hardware, also a totally irrational hobby; but I find the enthusiasm for using pitifully-weak-but-modern gear kind of baffling), just that if you do happen to swing that way, this FTDI device looks like the graphics and sound synthesis device for you, in a way that contemporary hardware could certainly emulate, probably trivially; but is fundamentally different from. People seem to find that the sport is spoiled by emulation for some reason.
It's not really a 'GPU' at all in the current sense of term(Do modern 'GPU's even do hardware sprites anymore, or do they just treat them as special, particularly flat, cases of textured polygons?); but if you are into the retro aesthetic and design style/limitations, a chip that does high-speed sprite jockeying is probably going to make you a lot happier than any of the 'Yup, just another OpenGL ES GPU that your desktop would stomp on; but which is so powerful that you would have sold your soul for it back when GLQuake came out...' GPUs that get mated with ARM SoCs these days.
The notebook I bought in 2008 had 5 GHz support already. Then just recently upgraded it to one with a sandy bridge cpu and that did not have... Went to ebay and ordered the same old wifi card my old notebook had, cost $8, inc shipping from China. The manufacturer probably saved 10 cents on each notebook because he decided to choose the inferior type.
One quirk in the WiFi market (less noticeable in strictly consumer laptops; but more visible in business ones) was the wrinkle introduced by 802.11A more or less entirely dying a horrible death.
For a while, '802.11A/B/G' was sort of the standard boring-business-laptop wifi card to have; but (I presume) as the volume of consumer wireless increased, and with it shipments of B/G chipsets, A withered and N (while theoretically, if the right boxes are ticked, doing pretty much everything A did, and more) can also be a 2.4GHz-only, incrementally-better-than-G flavor and often is, and so with the decline in setups with explicit A support, you frequently lost features with B/G or B/G/N(lite edition) cards swapped in instead.
Because 802.11A-capable APs always cost excessive amounts it isn't a surprise that A died (thanks to progress, it still tends to cost less to buy a proper 802.11N router and client card to match than it would to do the same for A); but A did have the advantage of specifying a relatively high baseline, unlike N, which is...flexible.
I'd be more inclined to blame standards bodies(though, depending on how they are structured, that could go right back to hardware manufacturers, albeit probably not the really low-margin box-slinger ones, more the silicon guys).
When it comes to 'features' that customers can see, manufacturers are crazy responsive (sometimes to the point of just making them up, or lying about them; but so it goes...) You care about the sticker price? We shave every penny that doesn't remove a bullet point from the spec sheet. You want megapixels? Here, have so many damn megapixels that the less-quantifiable shitty plastic optics can't even steer photons on to some of them consistently! More megahertz is faster? How about a NetBurst architecture?
Whenever it is difficult to convey the benefits of a given feature, it tends to get chopped in favor of price. Unfortunately, then economies of scale kick in and help mop up the remaining few people who do care. (In specialty markets, like APs designed for campus WiFi deployment, you are dealing with a situation where buyers really do need all the power they can get, and are often even capable of understanding what to ask for; but that's also the land of $400 access points, so it doesn't trickle down to mass market laptops much...)
Since you can get the 802.11n! label by implementing a single stream at 2.4GHz, and most people don't even know that something else is available (the existence of 5GHz is starting to break through, if only because 2.4 is blitzed into total uselessness in denser areas), the manufacturers don't have much incentive to be 'that one that mysteriously costs $10 more than the other ones, and they are all 802.11n'.
Not really. It does require that there not be people unique in their abilities; but it works just as well if you consider that a 'shift in supply' may be a matter of trivial retraining (sending the guy who sweeps the floor out to shovel snow), nontrival retraining (a prolonged apprenticeship in some technical skill, say) or the time required for enough people to be born and tested such that the person with the necessary raw talent is identified (a good math PhD, say, is almost certainly born and made; so the time to deliver one is X years of school plus some additional amount of time because most students have to be discarded as unsuitable). Such people are, of course, very well placed indeed if demand for them goes up, because generating new supply could take decades, while skills requiring less unusual innate talents can be churned out in much shorter times.
Slashdot Mirror
These accusations of coercion are blatantly defamatory. We are simply offering incentives, which the interns, as free and rational agents, are choosing to accept or decline. It's practically a libertarian utopia, trade among men, as equals, free from the dead hand of state power. Anyone who says otherwise is probably some sort of commie, who thinks that labor and capital negotiate from positions of unequal strength or some bullshit like that.
I suspect that it's partially inertia/penny-pinching and partially because crypto only solves certain (quite specific) problems within the larger problem of 'run an election'.
For instance, in those countries that have smartcard-or-equivalent national IDs, cryptographically signed votes would be trivial; but you'd be reusing keys explicitly designed to not be anonymous, indeed, designed to be identifying. That is an issue. Beats some 'SSN+Mother's maiden name' bullshit; because at least it verifies something; but it isn't what you are looking for.
If you anonymously issue keys, now you've got a weak spot there that crypto can't help you with: the crypto makes it quite possible to ensure that Anon_Key_X was responsible for Vote_X, and only Vote_X; but you still need to devise a system by which an eligible voter can obtain (without some absurd hassle) one and only one anonymous key, without it being covertly linked back to them, or them being able to sign up for ten, or the people running the system being able to generate 250,000(or simply keep a copy of the keys as they are issued, and 'win the race' to get a signed ballot into the pot with that key).
If you have such a system, you also have a system that could trivially just hand the voter a ballot, since you have already satisfied anonymity, uniqueness, resistance to plural voting, etc. No need for the crypto at all.
(Also, aside from that, a country with vote rigging tendencies is presumably going to use hierarchical PKI, not some web-of-trust cypherpunk wet dream, so what exactly will an election whose ballots are signed with keys that all descend from the 'Glorious Cryptographic Key for Make Benefit of People's Republic Motherland' prove? Hierarchical PKI schemes, as SSL has taught us, work OK if you are primarily concerned with criminals and frauds; but if the CA is the enemy, you are fucked. If you are the root, you can generate mathematically pristine child keys as fast as your little ASICs can carry you without the slightest trouble.)
Oh, I'm not saying that the correct answer is Java, just that choosing a language that appears to be actively dying (rather than just choosing a pedagogical language without listening to the 'But, but, practical skills!' idiots) seems like it would add an extra layer of logistical hassle.
Given the state of South African secondary education, 'extra layer of logistical hassle' is not high on the list of things I would want.
If memory serves, both the US NRC (presumably because GE was involved, or because somebody asked) and their Japanese equivalent pointed out that there were... issues... with various aspects of the facility(including flooding-related ones that ended up being a problem, in addition to retro reactor design problems that they didn't know about); but nobody likes that guy who can't be a team player, and so such issues were not addressed.
Not bad by fire alarm standards ( you see quoted lifespans from 5-10 years usually, with shorter ones for very safety critical carbon monoxide sensors), it just seems kind of churlish not to have the $10-$15ish commodity parts in a swappable FRU, just like the batteries...
It's entirely possible that much of 'they' weren't really asked. If the decision-making around Fukishima (when it was first installed) was anything like that around Minamata, stonewalling and general contempt for the 'opinions' of whiny little people was the official policy of both business and state.
Its almost like the are in a culture where you can't call out people's mistakes and follow orders blindly.
A corporation?
"Energy" is an awfully persuasive argument. Without it, your civilization will crumble to little more than medieval standards of living surprisingly fast.
Given that Japan has ~0 coal, oil, gas (not even much wood, per capita), it was either nukes or imports. Much of the rest of the world hasn't had to face up to the problem as dramatically because they've got a big stash of cheap 'n nasty coal somewhere convenient. That has its own downsides; but those have proven easy to ignore (unless you make the mistake of living in our under-construction Appalacian Lunar Theme Park or something).
Anybody feel up to a rendition of 'Yackety Sax' on a geiger counter and a selection of alarm klaxons?
It does seem increasingly clear that TEPCO couldn't be trusted to take care of your fishtank for a weekend(not even the freshwater one that's really relaxed about sampling and balance adjustments); but it should be noted, in fairness, that wacky piping accidents do get easier the more thickly built (ideally ad-hoc, and in poorly labelled stages, with evolutionary growth here and there) the piping rat's nest gets.
The essence of true competence is to avoid getting into situations where continuous high levels of competence are needed; by not backing yourself into a clusterfuck of a system that is always one false move away from doing something dangerous; but if you've fucked up and done that, it's really just a matter of time until somebody gets tapped as the fall guy by the pitiless gods of blind chance.
Let us hope that the work they do there comes to nothing, and is overwhelmingly forgotten. It's a more charming approach than simple brute force and lobbying; but this is also an extension of Verizon's work to stave off becoming a dumb pipe, and reap the rents that doing so allows. That isn't good news for anybody except them, and possibly their favored app and/or device buddies.
Not only are they surprisingly bad at it (hands up, everyone who ever had the pleasure of a phone with a fully Verizoned ROM, or a Comcast-rented cable box UI); but the conflicts of interest inherent between offering a product or service and controlling the infrastructure over which that offering is delivered are irreconcilably dangerous.
"Record executions not withstanding."
I think that they get enough enjoyment out of enforcing those that they don't count as 'burdensome' (except for the executee; but nobody cares about him).
If not connectivity through a phone, then connectivity through a replacable module. Decoupling the input and output make this easy and easily upgradable. It also lets people pick their own services. I like using Google services myself, but someone else may not. There may be a new high speed comminications technology out in the near future. Decoupling aall of these thimgs makes it much more upgradable, and if they use open connectors/protocols, then you can use your phone to replace or supplement this functionality as well.
Like poor old ISO 7736, which doesn't seem to have been doing all that well in recent models, alas?
Which makes it particularly odd to choose some obscure oddball with a tiny handful of necrotic vendors and a few under-maintained OSS projects as options for tools, rather than an otherwise-equivalent one for which ongoing interest exists.
Quite possibly. 55 inches, especially if ostensibly shared, seems too large to benefit significantly. I'd put the sweet spot at around 30"... small enough that they sell ones you are supposed to sit close to, large enough that the corners are really significantly further from your eyes than the center.
Very little as apparently the article thinks the NRC is responsible for foiling terrorist plots to go after nuclear reactors.
Personally, I'm more worried about increased negligence from operators without somebody breathing down their necks than I am about terrorists.
(The most recent example, luckily nonnuclear, being the juxtaposition between the marathon bombers and the West Fertilizer company. Kill three people with a backpack full of explosives and all of greater Boston goes full tactical on you. Blow up 500,000 pounds of ammonium nitrate, killing 15 and leveling a good portion of the nearby town? Eh, we try to avoid burdensome regulations here in Texas...)
Neither is your leg, typically.
I have to imagine that any significant curve on a screen that small would end up looking sort of weird, though.
At somewhere in the 24-30 inch range (for monitors, larger for TVs that you are expected to sit further away from) I can see curvature as being a good thing, so that all parts of the screen are roughly equidistant from your face, rather than the edges requiring refocusing as you move your eyes around; but that won't be at play on a phone. If they keep it subtle you'll probably adjust pretty quickly, though.
WTF! At least a thermostat actually does something worth making it programmable for... this is just a ridiculously overengineered implementation of "if fire, make noise."
The fact that both photoelectric smoke detectors and carbon monoxide detectors (especially the latter) have finite lifespans (as little as five years, less if certain sensor poisons are present) and neither component is separately replaceable makes the $130 price tag extra hard to swallow...
In practice, 'Version XYZ of Open Standard Whatever, with certain bugs and idiosyncrasies' is 'your API' if you go the open standards route. Over time, the really nasty stuff gets ironed out; but as long as it's "Well, we needed an API and there wasn't one, so we made one, here it is." there isn't necessarily anything wrong with that (in fact, more than a few now-open standards exist because somebody was first to need an API, so they Just Did It and then let(or couldn't stop) standardization from happening around that. Like Hayes-command-set modems.
Now, if Nest means 'their' API in the sense that you have to sign a EULA in blood to write a program that talks to your own hardware (obviously, if an API is a mechanism for accessing resources on somebody else's system, they can set whatever access rules they like; because they own that computer; but if they claim to own the API per se and control all uses of it, We Have A Problem.) they can go suck it; but if they just mean 'Our API, the API that we use and implement in our products' no big deal.
Oh, I'd be the last to claim that most 'retro' computing projects make a whole lot of sense, except perhaps as pedagogical exercises (I have a personal fondness for 'retro' projects involving actual old-school hardware, also a totally irrational hobby; but I find the enthusiasm for using pitifully-weak-but-modern gear kind of baffling), just that if you do happen to swing that way, this FTDI device looks like the graphics and sound synthesis device for you, in a way that contemporary hardware could certainly emulate, probably trivially; but is fundamentally different from. People seem to find that the sport is spoiled by emulation for some reason.
Anybody interesting and hilariously anti-drug in public life on the list yet, or do those get filtered out before they send in the jackboots?
It's not really a 'GPU' at all in the current sense of term(Do modern 'GPU's even do hardware sprites anymore, or do they just treat them as special, particularly flat, cases of textured polygons?); but if you are into the retro aesthetic and design style/limitations, a chip that does high-speed sprite jockeying is probably going to make you a lot happier than any of the 'Yup, just another OpenGL ES GPU that your desktop would stomp on; but which is so powerful that you would have sold your soul for it back when GLQuake came out...' GPUs that get mated with ARM SoCs these days.
The notebook I bought in 2008 had 5 GHz support already. Then just recently upgraded it to one with a sandy bridge cpu and that did not have... Went to ebay and ordered the same old wifi card my old notebook had, cost $8, inc shipping from China. The manufacturer probably saved 10 cents on each notebook because he decided to choose the inferior type.
One quirk in the WiFi market (less noticeable in strictly consumer laptops; but more visible in business ones) was the wrinkle introduced by 802.11A more or less entirely dying a horrible death.
For a while, '802.11A/B/G' was sort of the standard boring-business-laptop wifi card to have; but (I presume) as the volume of consumer wireless increased, and with it shipments of B/G chipsets, A withered and N (while theoretically, if the right boxes are ticked, doing pretty much everything A did, and more) can also be a 2.4GHz-only, incrementally-better-than-G flavor and often is, and so with the decline in setups with explicit A support, you frequently lost features with B/G or B/G/N(lite edition) cards swapped in instead.
Because 802.11A-capable APs always cost excessive amounts it isn't a surprise that A died (thanks to progress, it still tends to cost less to buy a proper 802.11N router and client card to match than it would to do the same for A); but A did have the advantage of specifying a relatively high baseline, unlike N, which is...flexible.
I'd be more inclined to blame standards bodies(though, depending on how they are structured, that could go right back to hardware manufacturers, albeit probably not the really low-margin box-slinger ones, more the silicon guys).
When it comes to 'features' that customers can see, manufacturers are crazy responsive (sometimes to the point of just making them up, or lying about them; but so it goes...) You care about the sticker price? We shave every penny that doesn't remove a bullet point from the spec sheet. You want megapixels? Here, have so many damn megapixels that the less-quantifiable shitty plastic optics can't even steer photons on to some of them consistently! More megahertz is faster? How about a NetBurst architecture?
Whenever it is difficult to convey the benefits of a given feature, it tends to get chopped in favor of price. Unfortunately, then economies of scale kick in and help mop up the remaining few people who do care. (In specialty markets, like APs designed for campus WiFi deployment, you are dealing with a situation where buyers really do need all the power they can get, and are often even capable of understanding what to ask for; but that's also the land of $400 access points, so it doesn't trickle down to mass market laptops much...)
Since you can get the 802.11n! label by implementing a single stream at 2.4GHz, and most people don't even know that something else is available (the existence of 5GHz is starting to break through, if only because 2.4 is blitzed into total uselessness in denser areas), the manufacturers don't have much incentive to be 'that one that mysteriously costs $10 more than the other ones, and they are all 802.11n'.
Not really. It does require that there not be people unique in their abilities; but it works just as well if you consider that a 'shift in supply' may be a matter of trivial retraining (sending the guy who sweeps the floor out to shovel snow), nontrival retraining (a prolonged apprenticeship in some technical skill, say) or the time required for enough people to be born and tested such that the person with the necessary raw talent is identified (a good math PhD, say, is almost certainly born and made; so the time to deliver one is X years of school plus some additional amount of time because most students have to be discarded as unsuitable). Such people are, of course, very well placed indeed if demand for them goes up, because generating new supply could take decades, while skills requiring less unusual innate talents can be churned out in much shorter times.