This isn't the sort of thing likely to bother Apple; but the major downside will be that the monitor will be stuck with whatever GPU was integrated for its entire life; and odds are that a nice 5k panel will be good enough for the job longer than the GPU that would fit within a suitably slim power budget.
Not a huge problem if you just want a lot of screen to do relatively undemanding things; but if Apple hasn't entirely finished chasing off their workstation customers they will probably find the performance rather painful.
Space-NIMBY would be different than earth-NIMBY; but could actually be a great deal uglier. Space is big; but some orbits are a lot more useful than others, and at the velocities involved one dumbass having a satellite break up rather than being safely retired can be a real risk to everyone else trying to use the orbit that now has a bunch of extra debris flying around.
Plus, anything suitably large and solid in orbit could potentially be de-orbited onto a deserving target below; and anything suitable for orbital launches is pretty close to being an ICBM in terms of technical capability. These sorts of considerations are quite likely to make various interested parties really, really, jumpy about the more ambitious orbital development proposals.
I assume that this factory-in-the-sky plan also involves mining asteroids or something; but even if that is the case the cost of shipping would seem to make earth orbit about the worst possible place for heavy industry.
Reentry is free; unless you actually want the product to survive, in which case it becomes nontrivial; and if either the ingredients, the process, or the product(or all of the above) are noxious enough that you don't want to produce them on earth, you also don't want transports full of them breaking up in the atmosphere when reentry goes bad or a launch fails.(Plus, just about anything durable enough to not break up in the atmosphere is also no longer harmless because if you are in earth orbit you've got enough gravitational potential energy to make a pretty nifty kinetic kill vehicle.)
There's also the sheer size of 'heavy industry' and the comparatively low value per unit volume or unit weight(usually whichever is less convenient). Even on earth's surface, the scale of the transportation and the relatively low value of the cargo place substantial restraints on where it makes sense to build processing facilities, what sort of access to seaports or rail lines is needed to make exploiting a mineral deposit worth the trouble, and so on.
This isn't to say that orbit is necessarily useless(indeed, given the relatively small, but quite lively, market for satellites, even ones not funded by nation states for their spooks, commercial space is already a reality); it may well be that certain high value processes benefit from microgravity, and shipping costs matter a lot less if you are talking high end microprocessors or exotic medical proteins or something; but the stuff you'd describe as 'heavy industry' seems like the absolute last thing you'd move into space(with the one exception for 'if asteroid mining can be made to work, we'd almost certainly do that).
Heavy industry just screams when you try to impose assorted pollution reduction/worker safety/etc, measures that are vastly cheaper than "please move your factory into space"; and unless someone gets that space elevator built, I'm not at all sure we even could move heavy industry into space, even if we blindly threw all our resources at the task.
It would appear that 'apps' are rather intrusive if they are phoning home enough that we can say how many are opened only once. It would also appear that users are substantially less harsh in their assessments than the miserable shovel ware of the mobile world deserves if so many are being opened at all.
When spying is getting closer to being free with every improvement in IT efficiency it doesn't take much 'interesting' to be interesting enough to spy on.
In fairness, a lot of the less competent analog sets spent most of their lives drifting vaguely around some approximation of the image they were supposed to be displaying, those little trim pots that you were only supposed to use a plastic screwdriver to adjust weren't just for show.
What they weren't, though, was smart enough to be actively adversarial; and that is vastly worse.
The nasty thing about upgradeable firmware is that it effectively means that hardware is governed by all the various nasty terms, conditions, EULAs, licensed-not-solds, and subject-to-change-without-notice that software is.
It's times like this when Stallman's vociferous demand for nothing less than fully free software as a necessary condition of user control looks more like lucid foresight and less like blinkered monomania. The issue was largely dormant back when firmware upgrades were hard and internet connections were the exception rather than the rule; but now it is eminently practical for a vendor to extend their control over something they supposedly sold to you more or less in perpetuity.
One should never bet against dishonesty; but math is one of those places where 'X holds except for value N' is likely to be considered way more interesting than quietly suppressing the N case and declaring 'X holds'. The follow-up research on "why value N?" is almost certain to be of greater interest, and keep you getting cited longer, than the "yup, X is proven, move along."
While it does seem reasonable to treat 'proofs that are essentially incomprehensible except in summary' differently from ones that humans can actually use; it seems uncomfortably likely that most proofs, likely the overwhelming majority, are of the incomprehensible flavor with just a tiny little island of math that a suitably intelligent person could actually plow through. Beyond that, it seems reasonable to suspect that only a vanishingly small slice of the proofs that could, in principle, be generated in finite time on finite hardware are actually accessible to the very, very, finite amount of time and hardware we have at our disposal.
This isn't to say that such monstrous proofs are any more palatable; but even among the bits of math well behaved enough to be provable at all it is likely that being inaccessibly vast is the rule rather than the exception.
The 'He did a public service; but how he did it was wrong and wicked and why didn't he go through channels?' position is insufferably smarmy.
Had he gone through channels, that public service would never have occurred. He wasn't the first person inside the intelligence apparatus to grow uneasy with what was going on; but 'politely ask your boss to reconsider his own malfeasance' just doesn't do very much.
If you agree that he did something valuable for the public and the country; you pretty much have to admit that his options were limited for doing so(there is room for quibbling about the finer points of his plan; but you can't very well deny that it was a 'leak it or let it stay in the dark indefinitely' situation). That's the whole point of 'whistleblower' as a distinct class of people working against corruption: they are the ones who bring in external scrutiny when an organization's internal governance has been co-opted by whatever malfeasance they are exposing. More or less by definition they have to use 'illicit' methods; because the problem extends to the people who control all the legitimate methods.
It's perfectly cogent to argue that someone leaked something without reason, and should therefor face the consequences; but 'his reasons were excellent and he did us all a service' is an admission that there were no 'legitimate' channels through which he could have worked.
The trouble in this case is that it is frequently the armchair quarterbacks who are pushing the curriculum, and the teachers trying to pick up the pieces within that context. Letting that sort of thing pass without comment or challenge is allowing the armchair quarterbacks to mess with the teachers.
There is obviously a case to be made that "so kids, let's do some proofs about computability!" may not exactly draw the middle schoolers in; but it's also the case that "everybody learns to code because the app entrepreneurs future!!!" creates a strong incentive toward 'CS' watered down until everyone can be shoved through it without too much hassle.
I think that it's a mixture of "preorder" and "hype mechanism"; along with a convenient way to test demand. Plus, if you can get suckers to loan you money for an indeterminate time without guarantee of payback or any expectation of return on investment, why not?
I find the process rather seedy; but It's hard to argue with the pragmatic-evil reasons for doing it.
I'd imagine that the big variable on 8-in floppy cost is whether or not any modern use requires a magnetic medium of coercivity close enough to be compatible: the actually-floppy floppies are pretty simple on the inside, just a casing made of die-cut plastic sheet, some anti-dust pads on the inside, and the 'donut' of magnetic medium. If you have to commission bespoke magnetic medium because all the modern stuff is too high coercivity to suit high density magnetic recording, that could get unpleasant. If there is some user of magnetic film of appropriate coercivity, getting appropriately shaped pieces of it punched out of sheetstock shouldn't be too difficult.
Even if (and it wouldn't be too surprising) the floppies or their drives are ultimately impractical to keep running, it still might be easier and less risky to build a floppy drive emulator that speaks the appropriate protocol but uses some more modern storage mechanism. That's an entire cottage industry in hobby retrocomputing, allowing you to replace scarce oddball HDDs and weirdo floppies with CF or SD cards; and given the relative simplicity of historical floppy drive interfaces I would strongly suspect that you could get an all-American engineering team to cook up a drop-in replacement without too much trouble.
That said, hardware is certainly the area where obsolescence is likely to become a real logistical problem first; so any attempts at modernization(incremental or wholesale forklift-replacement) should be aimed at trying to decouple the system from specific hardware as much as possible(the 'baseline' hardware profiles used by virtualization systems to accommodate guest OSes that aren't virtualization aware and capable of playing nice with virtualized devices, say, are already obsolete hardware; but will probably be just as available decades from now as they are today); but even where hardware is involved, the difficulty of replacing the system as a whole makes trying to incrementally replace the hardware(with well defined compartmentalization at various interfaces to make the future replacement of your replacements easier) a viable consideration.
I wouldn't necessarily take refuge in obscurity if running something important; a core IRS system or nuclear-related control systems would be the sort of targets where you'll get some fairly motivated attackers rather than just kiddies looking for soft targets. That said, it's not necessarily the case that old=insecure in a situation where you aren't dealing with software thrown together as fast as possible to secure a first mover advantage or win a feature race with competitors.
There have been a lot of advances over the years in the average state of low cost hardware and software, and in attempting to mitigate the results of running a hodgepodge of untrusted and mostly crap software exposed to a constant stream of hostile input from the internet; but that newer-is-mostly-less-awful trend is really most notable in the cheap seats, not in comparatively simple(if only because the hardware wasn't available for anything bigger) and very expensive systems built for justifiably paranoid customers.
I suspect that some of the now outdated 'COTS' based systems are truly horrifying: new and common enough that plenty of known vulnerabilities exist, old and dysfunctional enough that they probably aren't getting fixed; but the more unusual evolutionary dead ends, while not cheap to support, have at least a chance of being extremely good at what they do.
The tricky bit is whether anyone bidding for such a job would want the project managed as competently as possible, or whether it's one of those situations where having a risibly old(but functional enough that disasters aren't drawing attention to the slipping deadlines of the replacement) legacy system makes meandering in the vague direction of a solution for as long as you can as good or better than actually delivering.
If something like the 'CityTime' payroll system upgrade project can go as excitingly wrong as it did; I'd hate to see what a project of this magnitude would do.
It was all pretty monochrome; but some surprisingly early GUIs existed. SAGE had them(with lightguns rather than mice, since it predated those by a fair bit); among various other flavors of 'it's actually pretty impressive what you can do with vacuum tubes if you have a lot of smart people and nigh-unlimited money' style tech.
I'd be curious to know how many of these seriously outdated systems are egregious piles of failure; and how many are utterly contrary to any fad of the week from the last three decades; but where done right the first time and actually compare pretty favorably to the results of (the so often horribly doomed) 'upgrade' efforts.
Some flavors of outdated are fairly clearly bad; if you can't get replacement hardware without raiding a museum or reverse engineering and cloning/emulating quirky 80s gear all by yourself, keeping your systems running is going to be unpleasant and expensive. If you have a system whose security depends on an OS or other 3rd party components that have exciting known vulnerabilities and haven't had vendor support even under a thrillingly expensive special extended contract with the vendor in a decade, you have a problem.
If you have a legacy system that is merely retro; but well built and supported by hardware you can still get without much trouble, you will certainly get your share of snide comments about its dreadfully antique design; but you are taking a real risk in trying to modernize it. Those sorts of 'upgrades' don't always fail; but agonizing, wildly expensive, upgrade attempts that languish in development so long that the upgrade is obsolete before you've finished deploying it are hardly uncommon.
Sure, in an ideal world, we'd all get to implement from scratch with all the benefits of hindsight and absolutely no accrued technical debt; but we don't live in an ideal world. How many of these systems are old as in broken; and how many are old as in classic?
So, tell me, how exactly does 'the authority of a democracy' exist when dealing with a program so secret that even the bulk of the congress knew relatively little about it, never mind the electorate at large?
It is nice that his conclusion(and he doesn't think that he is being arrogant in assuming his carefully curated little field trip is sufficiently accurate and representative?) was that the NSA was mostly abiding by the rules they made up, rather than going mad with power; but it's simply smarmy nonsense to pretend that anything that clandestine has any meaningful relationship to democracy. On a good day, such an enterprise might be an unaccountable black box more or less attempting to do what they interpret a democratic society's mandate for them to be; but you could say the exact same thing about a hereditary despot who tries to govern more or less according to the interests of the population as he understands them: aligned with the objectives of a democracy only by their own preference, if at all.
Their whole thing is a software-level redundancy arrangement designed to provide adequate reliability through redundancy on top of utter shit hardware. That's the company's niche. It does mean that they massacre drives like crazy; but their cost/GB is pretty impressive, so long as you are doing fairly cold storage, not something IOPS intensive.
Probably price: Backblaze's thing is using some sort of software abstraction and redundancy layer to get away with providing storage on the cheapest drives that they can get their hands on.
Makes them a pretty good value among providers of offsite backup/cold-ish storage; but they have a very limited interest in paying for more reliability at the hardware level, since that would fairly quickly push them into the domain of traditional storage vendors who use more expensive hardware to provide fault tolerance for software that isn't designed to handle that itself.
They obviously have an interest in getting the best value for money, hence the gathering reliability data, and they'd presumably be willing to pay a nonzero premium if the reliability difference were large enough; but their whole approach is a 'paper over lousy hardware in software' strategy. It makes their storage designs a poor drop-in replacement for many applications(even if you are using a fairly clever filesystem like ZFS that has good tolerance for some drives dying, the sight of SATA port multipliers hanging off the cheapest HBAs they can find might make you a bit nervous); but it's pretty difficult to buy a storage system where a lower percentage of the total cost is non-disk hardware.
Depends on your use case: the Backblaze people are operating a system specifically designed for cheapo drives that are expected to have a fairly high chance of falling over and dying(pragmatically speaking, that's part of why they are so nice and friendly about drive reliability data and sharing the designs for their 'pods': their real asset as a company is the software sauce that allows them to offer cheap, reliable, storage through software-level redundancy on top of a pile of low-end drives packed tight and connected with really cheap HBAs and SATA port multipliers: no fancy hardware RAID, no redundant-controller SAS, etc.)
If you are buying drives to use as the boot volume for computers that only get a single HDD, or even systems with small RAID arrays, you are going to be seriously inconvenienced by drive models that drop dead atypically fast, even if you save a few bucks upfront. Re-imaging a replacement drive or swapping out a failed RAID disk and rebuilding the volume take time and trouble.
If your purposes are very similar to theirs, then your sensitivity to failure is lower and getting a slightly better deal per GB might start to make sense; but you have to be pretty failure insensitive(or the price of reliability really steep) to be in the same boat.
You misinterpreted: this is a billion drive-hours worth of data, not a drive operating for a billion hours(given that that's a bit over 110,000 years, we don't really have that sort of reliability data, even if anyone cared).
And, when it comes to reliability analysis, that 'ridiculous amount of time' is enormously helpful. How else are you going to draw statistically significant conclusions about something with such an element of chance?
Slashdot Mirror
This isn't the sort of thing likely to bother Apple; but the major downside will be that the monitor will be stuck with whatever GPU was integrated for its entire life; and odds are that a nice 5k panel will be good enough for the job longer than the GPU that would fit within a suitably slim power budget.
Not a huge problem if you just want a lot of screen to do relatively undemanding things; but if Apple hasn't entirely finished chasing off their workstation customers they will probably find the performance rather painful.
Space-NIMBY would be different than earth-NIMBY; but could actually be a great deal uglier. Space is big; but some orbits are a lot more useful than others, and at the velocities involved one dumbass having a satellite break up rather than being safely retired can be a real risk to everyone else trying to use the orbit that now has a bunch of extra debris flying around.
Plus, anything suitably large and solid in orbit could potentially be de-orbited onto a deserving target below; and anything suitable for orbital launches is pretty close to being an ICBM in terms of technical capability. These sorts of considerations are quite likely to make various interested parties really, really, jumpy about the more ambitious orbital development proposals.
I assume that this factory-in-the-sky plan also involves mining asteroids or something; but even if that is the case the cost of shipping would seem to make earth orbit about the worst possible place for heavy industry.
Reentry is free; unless you actually want the product to survive, in which case it becomes nontrivial; and if either the ingredients, the process, or the product(or all of the above) are noxious enough that you don't want to produce them on earth, you also don't want transports full of them breaking up in the atmosphere when reentry goes bad or a launch fails.(Plus, just about anything durable enough to not break up in the atmosphere is also no longer harmless because if you are in earth orbit you've got enough gravitational potential energy to make a pretty nifty kinetic kill vehicle.)
There's also the sheer size of 'heavy industry' and the comparatively low value per unit volume or unit weight(usually whichever is less convenient). Even on earth's surface, the scale of the transportation and the relatively low value of the cargo place substantial restraints on where it makes sense to build processing facilities, what sort of access to seaports or rail lines is needed to make exploiting a mineral deposit worth the trouble, and so on.
This isn't to say that orbit is necessarily useless(indeed, given the relatively small, but quite lively, market for satellites, even ones not funded by nation states for their spooks, commercial space is already a reality); it may well be that certain high value processes benefit from microgravity, and shipping costs matter a lot less if you are talking high end microprocessors or exotic medical proteins or something; but the stuff you'd describe as 'heavy industry' seems like the absolute last thing you'd move into space(with the one exception for 'if asteroid mining can be made to work, we'd almost certainly do that).
Heavy industry just screams when you try to impose assorted pollution reduction/worker safety/etc, measures that are vastly cheaper than "please move your factory into space"; and unless someone gets that space elevator built, I'm not at all sure we even could move heavy industry into space, even if we blindly threw all our resources at the task.
It would appear that 'apps' are rather intrusive if they are phoning home enough that we can say how many are opened only once. It would also appear that users are substantially less harsh in their assessments than the miserable shovel ware of the mobile world deserves if so many are being opened at all.
When spying is getting closer to being free with every improvement in IT efficiency it doesn't take much 'interesting' to be interesting enough to spy on.
In fairness, a lot of the less competent analog sets spent most of their lives drifting vaguely around some approximation of the image they were supposed to be displaying, those little trim pots that you were only supposed to use a plastic screwdriver to adjust weren't just for show.
What they weren't, though, was smart enough to be actively adversarial; and that is vastly worse.
Hey, be fair. In 1984 The Party provided telescreens to the masses. We have to buy ours.
The nasty thing about upgradeable firmware is that it effectively means that hardware is governed by all the various nasty terms, conditions, EULAs, licensed-not-solds, and subject-to-change-without-notice that software is.
It's times like this when Stallman's vociferous demand for nothing less than fully free software as a necessary condition of user control looks more like lucid foresight and less like blinkered monomania. The issue was largely dormant back when firmware upgrades were hard and internet connections were the exception rather than the rule; but now it is eminently practical for a vendor to extend their control over something they supposedly sold to you more or less in perpetuity.
One should never bet against dishonesty; but math is one of those places where 'X holds except for value N' is likely to be considered way more interesting than quietly suppressing the N case and declaring 'X holds'. The follow-up research on "why value N?" is almost certain to be of greater interest, and keep you getting cited longer, than the "yup, X is proven, move along."
While it does seem reasonable to treat 'proofs that are essentially incomprehensible except in summary' differently from ones that humans can actually use; it seems uncomfortably likely that most proofs, likely the overwhelming majority, are of the incomprehensible flavor with just a tiny little island of math that a suitably intelligent person could actually plow through. Beyond that, it seems reasonable to suspect that only a vanishingly small slice of the proofs that could, in principle, be generated in finite time on finite hardware are actually accessible to the very, very, finite amount of time and hardware we have at our disposal.
This isn't to say that such monstrous proofs are any more palatable; but even among the bits of math well behaved enough to be provable at all it is likely that being inaccessibly vast is the rule rather than the exception.
The 'He did a public service; but how he did it was wrong and wicked and why didn't he go through channels?' position is insufferably smarmy.
Had he gone through channels, that public service would never have occurred. He wasn't the first person inside the intelligence apparatus to grow uneasy with what was going on; but 'politely ask your boss to reconsider his own malfeasance' just doesn't do very much.
If you agree that he did something valuable for the public and the country; you pretty much have to admit that his options were limited for doing so(there is room for quibbling about the finer points of his plan; but you can't very well deny that it was a 'leak it or let it stay in the dark indefinitely' situation). That's the whole point of 'whistleblower' as a distinct class of people working against corruption: they are the ones who bring in external scrutiny when an organization's internal governance has been co-opted by whatever malfeasance they are exposing. More or less by definition they have to use 'illicit' methods; because the problem extends to the people who control all the legitimate methods.
It's perfectly cogent to argue that someone leaked something without reason, and should therefor face the consequences; but 'his reasons were excellent and he did us all a service' is an admission that there were no 'legitimate' channels through which he could have worked.
The trouble in this case is that it is frequently the armchair quarterbacks who are pushing the curriculum, and the teachers trying to pick up the pieces within that context. Letting that sort of thing pass without comment or challenge is allowing the armchair quarterbacks to mess with the teachers. There is obviously a case to be made that "so kids, let's do some proofs about computability!" may not exactly draw the middle schoolers in; but it's also the case that "everybody learns to code because the app entrepreneurs future!!!" creates a strong incentive toward 'CS' watered down until everyone can be shoved through it without too much hassle.
I think that it's a mixture of "preorder" and "hype mechanism"; along with a convenient way to test demand. Plus, if you can get suckers to loan you money for an indeterminate time without guarantee of payback or any expectation of return on investment, why not?
I find the process rather seedy; but It's hard to argue with the pragmatic-evil reasons for doing it.
I'd imagine that the big variable on 8-in floppy cost is whether or not any modern use requires a magnetic medium of coercivity close enough to be compatible: the actually-floppy floppies are pretty simple on the inside, just a casing made of die-cut plastic sheet, some anti-dust pads on the inside, and the 'donut' of magnetic medium. If you have to commission bespoke magnetic medium because all the modern stuff is too high coercivity to suit high density magnetic recording, that could get unpleasant. If there is some user of magnetic film of appropriate coercivity, getting appropriately shaped pieces of it punched out of sheetstock shouldn't be too difficult.
Even if (and it wouldn't be too surprising) the floppies or their drives are ultimately impractical to keep running, it still might be easier and less risky to build a floppy drive emulator that speaks the appropriate protocol but uses some more modern storage mechanism. That's an entire cottage industry in hobby retrocomputing, allowing you to replace scarce oddball HDDs and weirdo floppies with CF or SD cards; and given the relative simplicity of historical floppy drive interfaces I would strongly suspect that you could get an all-American engineering team to cook up a drop-in replacement without too much trouble.
That said, hardware is certainly the area where obsolescence is likely to become a real logistical problem first; so any attempts at modernization(incremental or wholesale forklift-replacement) should be aimed at trying to decouple the system from specific hardware as much as possible(the 'baseline' hardware profiles used by virtualization systems to accommodate guest OSes that aren't virtualization aware and capable of playing nice with virtualized devices, say, are already obsolete hardware; but will probably be just as available decades from now as they are today); but even where hardware is involved, the difficulty of replacing the system as a whole makes trying to incrementally replace the hardware(with well defined compartmentalization at various interfaces to make the future replacement of your replacements easier) a viable consideration.
I wouldn't necessarily take refuge in obscurity if running something important; a core IRS system or nuclear-related control systems would be the sort of targets where you'll get some fairly motivated attackers rather than just kiddies looking for soft targets. That said, it's not necessarily the case that old=insecure in a situation where you aren't dealing with software thrown together as fast as possible to secure a first mover advantage or win a feature race with competitors.
There have been a lot of advances over the years in the average state of low cost hardware and software, and in attempting to mitigate the results of running a hodgepodge of untrusted and mostly crap software exposed to a constant stream of hostile input from the internet; but that newer-is-mostly-less-awful trend is really most notable in the cheap seats, not in comparatively simple(if only because the hardware wasn't available for anything bigger) and very expensive systems built for justifiably paranoid customers.
I suspect that some of the now outdated 'COTS' based systems are truly horrifying: new and common enough that plenty of known vulnerabilities exist, old and dysfunctional enough that they probably aren't getting fixed; but the more unusual evolutionary dead ends, while not cheap to support, have at least a chance of being extremely good at what they do.
The tricky bit is whether anyone bidding for such a job would want the project managed as competently as possible, or whether it's one of those situations where having a risibly old(but functional enough that disasters aren't drawing attention to the slipping deadlines of the replacement) legacy system makes meandering in the vague direction of a solution for as long as you can as good or better than actually delivering.
If something like the 'CityTime' payroll system upgrade project can go as excitingly wrong as it did; I'd hate to see what a project of this magnitude would do.
It was all pretty monochrome; but some surprisingly early GUIs existed. SAGE had them(with lightguns rather than mice, since it predated those by a fair bit); among various other flavors of 'it's actually pretty impressive what you can do with vacuum tubes if you have a lot of smart people and nigh-unlimited money' style tech.
I'd be curious to know how many of these seriously outdated systems are egregious piles of failure; and how many are utterly contrary to any fad of the week from the last three decades; but where done right the first time and actually compare pretty favorably to the results of (the so often horribly doomed) 'upgrade' efforts.
Some flavors of outdated are fairly clearly bad; if you can't get replacement hardware without raiding a museum or reverse engineering and cloning/emulating quirky 80s gear all by yourself, keeping your systems running is going to be unpleasant and expensive. If you have a system whose security depends on an OS or other 3rd party components that have exciting known vulnerabilities and haven't had vendor support even under a thrillingly expensive special extended contract with the vendor in a decade, you have a problem.
If you have a legacy system that is merely retro; but well built and supported by hardware you can still get without much trouble, you will certainly get your share of snide comments about its dreadfully antique design; but you are taking a real risk in trying to modernize it. Those sorts of 'upgrades' don't always fail; but agonizing, wildly expensive, upgrade attempts that languish in development so long that the upgrade is obsolete before you've finished deploying it are hardly uncommon.
Sure, in an ideal world, we'd all get to implement from scratch with all the benefits of hindsight and absolutely no accrued technical debt; but we don't live in an ideal world. How many of these systems are old as in broken; and how many are old as in classic?
So, tell me, how exactly does 'the authority of a democracy' exist when dealing with a program so secret that even the bulk of the congress knew relatively little about it, never mind the electorate at large?
It is nice that his conclusion(and he doesn't think that he is being arrogant in assuming his carefully curated little field trip is sufficiently accurate and representative?) was that the NSA was mostly abiding by the rules they made up, rather than going mad with power; but it's simply smarmy nonsense to pretend that anything that clandestine has any meaningful relationship to democracy. On a good day, such an enterprise might be an unaccountable black box more or less attempting to do what they interpret a democratic society's mandate for them to be; but you could say the exact same thing about a hereditary despot who tries to govern more or less according to the interests of the population as he understands them: aligned with the objectives of a democracy only by their own preference, if at all.
You know it's a bad sign when the weather report for the day is "sous-vide".
"Consult a doctor if execution lasts more than 4 hours."
Their whole thing is a software-level redundancy arrangement designed to provide adequate reliability through redundancy on top of utter shit hardware. That's the company's niche. It does mean that they massacre drives like crazy; but their cost/GB is pretty impressive, so long as you are doing fairly cold storage, not something IOPS intensive.
Probably price: Backblaze's thing is using some sort of software abstraction and redundancy layer to get away with providing storage on the cheapest drives that they can get their hands on.
Makes them a pretty good value among providers of offsite backup/cold-ish storage; but they have a very limited interest in paying for more reliability at the hardware level, since that would fairly quickly push them into the domain of traditional storage vendors who use more expensive hardware to provide fault tolerance for software that isn't designed to handle that itself.
They obviously have an interest in getting the best value for money, hence the gathering reliability data, and they'd presumably be willing to pay a nonzero premium if the reliability difference were large enough; but their whole approach is a 'paper over lousy hardware in software' strategy. It makes their storage designs a poor drop-in replacement for many applications(even if you are using a fairly clever filesystem like ZFS that has good tolerance for some drives dying, the sight of SATA port multipliers hanging off the cheapest HBAs they can find might make you a bit nervous); but it's pretty difficult to buy a storage system where a lower percentage of the total cost is non-disk hardware.
Depends on your use case: the Backblaze people are operating a system specifically designed for cheapo drives that are expected to have a fairly high chance of falling over and dying(pragmatically speaking, that's part of why they are so nice and friendly about drive reliability data and sharing the designs for their 'pods': their real asset as a company is the software sauce that allows them to offer cheap, reliable, storage through software-level redundancy on top of a pile of low-end drives packed tight and connected with really cheap HBAs and SATA port multipliers: no fancy hardware RAID, no redundant-controller SAS, etc.)
If you are buying drives to use as the boot volume for computers that only get a single HDD, or even systems with small RAID arrays, you are going to be seriously inconvenienced by drive models that drop dead atypically fast, even if you save a few bucks upfront. Re-imaging a replacement drive or swapping out a failed RAID disk and rebuilding the volume take time and trouble.
If your purposes are very similar to theirs, then your sensitivity to failure is lower and getting a slightly better deal per GB might start to make sense; but you have to be pretty failure insensitive(or the price of reliability really steep) to be in the same boat.
You misinterpreted: this is a billion drive-hours worth of data, not a drive operating for a billion hours(given that that's a bit over 110,000 years, we don't really have that sort of reliability data, even if anyone cared).
And, when it comes to reliability analysis, that 'ridiculous amount of time' is enormously helpful. How else are you going to draw statistically significant conclusions about something with such an element of chance?