I have to wonder if the 16GB "requirement" is more of a recommendation and/or a bunch of default settings that deliberately avoid the disk as much as possible, and keep as many cores as you can throw at the job busy by compiling every little bit and piece in parallel...
On the one hand, with 16GB of RAM in the desktop/light workstation 4x4GB only running around $100(with the more workstation-friendly 2x8GB with ECC only twice that), it seems rather pointless to burn any developer time on trying to optimize the RAM needs of building the entire OS. RAM is cheap.
On the other hand, I have to wonder what they could possibly be doing to the process of compiling what is basically a weird-but-not-unrecognizable linux distro to make it that RAM hungry.
The slightly-tangled-and-mostly-software history of "Hyperion Entertainment" doesn't fill me with worlds of confidence; but there doesn't seem to be anything architecturally implausible about shipping a genuine "AmigaOS netbook", except that volume makes slapping a sticker on an Atom based system cheaper.
Despite ARM having pretty aggressively filled the 'consumer-visible stuff that isn't x86' market(not quite sure why they cleaned up so hard; but they did), there are still plenty of PPC SoCs around that have adequately low power demands and support a reasonable number of modern peripheral interfaces. There are a few desktop-ish boards built around them, no reason you couldn't shove one in a notebook, other than the price efficiencies of volume...
Frankly, this only seems like an "important free speech issue" in one respect: the (quite likely) possibility that the park management are using selective enforcement of (possibly outdated) regulations against people who merely annoy them.
The notion that certain things that incidentally happen to be speech can be curtailed or limited because they are also hazardous or deeply disruptive has been more or less unproblematic as long as the notion of freedom of speech has been a matter of political possibility. However, such limitations do offer a potentially hazardous temptation for anti-speech selective enforcement(Is running around a dense residential district at 3am and shouting your head off legitimately "disturbance of the peace"? Yeah. Does that mean that it would be OK for police to ignore some disturbers of the peace and arrest those who say unpopular things? Not So Much.)
If this case turns out to be the park staff using a permitting system written back when cameras were barely man-portable and 'filming' implied a trail of havoc to selectively quash the weirdos while ignoring That Vacationing Camcorder Asshole, whose life only has meaning if they glimpse it continuously through a viewfinder, they need a smacking down.
If it turns out that the permit requirements are applied uniformly, then it becomes the much less weighty question of whether or not the decreasing size and disruptiveness of cameras makes them due for a rewrite or not...
Your nerd "technical reasons" leave us no room for endless inflammatory arguments and trite gender sterotyping! How will we fill the empty silence now?
Giving the user a choice would imply that they know better than Apple does about what is best.
Apple's entire business model is founded on the operational assumption that they know better than the customer. Their present success is founded on the fact that they are quite often correct about this....
Given that they charge $6month/user, you'd think that that would chill spammers out of the service pretty quickly. If you are quasi-legit(yeah, yeah, you're an 'opt-in marketing professional', right...), MS has your payment details and can always nuke your account if you don't heed warnings.
If you are an outright scammer, a major American corporation with a history of litigation against people like you seems like a very odd place to try to pay for spam delivery with your skimmed CC accounts, surely there are better dodgy operators who will be cheaper and more cooperative...
Being free, hotmail anti-spam is much more of a technical problem, since they have to fight robo-signups and account cracking(these days, guessing some idiot's password is probably easier than reading most captchas...)
I'm going to theorize "Moving less product at higher margins is certainly a possibility" on that one. There is, after all, absolutely no reason to suspect that anybody is going to lower prices any further than they absolutely have to.
My point was just that, architecturally, online retailers can modify prices within minutes according to demand, their costs, whatever they can infer about incoming customers' price sensitivity, etc. That doesn't mean that they'll lower prices as fast as they raised them; but it does make grandparent's "That's fine, they will sell exactly ZERO of them until the price goes back to normal." situation quite unlikely: a price at which there are no buyers would be adjusted within a very short time. If it so happens that you can get most of the buyers at twice the price, though, no real reason to move down...
Alpha emitters should be approached with considerable caution:
Because alpha radiation doesn't penetrate well, simple proximity to an alpha source isn't a big deal; but if you manage to absorb an alpha emitter, the alpha radiation no longer has to penetrate well to cause significant damage(just ask the late Mr. Litvinenko).
Effectively, alpha emitters have to be handled as though they possess pretty extreme chemical toxicity. Properly sealed sources are pretty much harmless. Dusts, dissolved compounds, aerosols, etc. are to be avoided.
Given how quickly online retailers can adjust prices(essentially limited only by the speed at which the factors that control pricing decisions change), I'm guessing that they are either selling enough drives to be satisfied with the situation, or simply cannot afford to sell them any more cheaply, because of upstream price increases.
Unless you have absolutely no overhead, and everybody has some, moving zero product isn't a situation you shoot for. Moving less product at higher margins is certainly a possibility, so that can't be ruled out on theoretical grounds; but it seems incredibly unlikely that the retailers have stopped moving stock entirely.
I can't speak for the collective consciousness of Slashdot; but the various 'trusted computing' stuff seems to have exactly the same set of trade-offs now that it did in the 90's: It does make malicious modification(by untrusted 3rd parties, malice by trusted parties actually becomes easier) more difficult; but there isn't an enormous amount of room for optimism about the percentage of devices that will accept the user as the root of trust, rather than whoever the vendor burned in. The number won't be zero, certainly; but it seems only reasonable to expect that the 'trusted' future will be dominated by hardware whose trust list does not include you.
This is picking a nit with the examples, rather than the concept; but both floppies and SD cards have a physical switch in only the loosest sense of the term:
Floppies have no internal logic capable of acting on the switch state, it is entirely up to the floppy drive to sense and obey. SD cards do have an internal controller, and could theoretically enforce write-blocking on themselves; but they don't. Their switch is also just a little plastic tab, and it is entirely up to the reader to sense and obey the tab position. The card's PCB has no connection at all to the switch, and has no way of sensing its position...
Android has really never been an open development model: the applications that make android "android"(from the consumer perspective) are closed, and the development of each successive release has been between Google and their Best Buddy of the moment until release.
The big question has been, since 'Ice Cream Sandwich', whether it would continue to be closed development/open source, or whether it would go closed entirely, except for a few GPL-obligated kernel bits...
I haven't really been able to get a good sense of Drobo reliability at the population level. There are certainly enough horror stories in the wild to incline one not to trust them more than other storage appliances as a class; but I've also had the (dis)pleasure of observing RAID cards that cost more than the low to midrange Drobos, never mind the chassis and redundant PSUs, murder-suicide with all the data downstream of them...
They certainly don't seem to be markedly better than the competition in their price brackets(and since "BeyondRAID" is a proprietary feature, there aren't multiple vendors to play the price/reliability tradeoff game with); but I've not seen evidence that they are markedly worse, in strict reliability terms, just much weirder than all but the horrid little NAS units whose vendors are too cheap to slap a halfway competent interface on top of stock Linux RAID.
While the electrical erasing is certainly a convenient feature, I'm pretty sure that if you can scrape enough of the case off to bombard the die with hard UV, an EEPROM will also be erased(not terribly useful, in practice, just as the epoxy-case windowless EPROMs were functionally WORM media). I'd be fascinated by the gory details if it turns out that EEPROMs are actually light-resistant at a die level, though...
Given that, in the context of ethernet, "Jumbo frame" usually implies a whole 9000 bytes, I'd say that the HDD-based system does have the clear upper hand in potential frame size...
Pity about the latency and the being half-duplex...
Unless you are absolutely phobic about exposure to harrowingly technical terms like "raid5", you should approach the drobo with notrivial caution.
They are quite pricey for their size and performance, which has historically been pretty tepid. Probably worse than that(which is a set of vices shared with quite a few other underpowered NAS units), is that their "BeyondRAID" system makes up for some powerful features by being Just Fucking Weird in some annoying ways.
Perhaps my least favorite is the ghastly hack that they use to make automatic array expansion 'easy'. To quote from their getting started guide:
"- Volume size does not represent how much actual storage space is available on
your Drobo/Pro/S. It represents virtual storage space. For example, your Drobo/
Pro/S may be loaded with 2TB of hard drive space, but you can create a volume
of 8TB. What this enables you to do is add more capacity to your Drobo/Pro/S (by
inserting an additional drive or replacing a smaller capacity drive with a larger
capacity one) without having to format an additional volume. The additional
capacity becomes part of the same volume you formatted originally.
- Your operating system may show the virtual space you have available on your
Drobo device, as defined by the volume size."
Yup. Unlike traditional RAID, you don't have to break and re-create the array to enlarge it, which is nice; but only if you initially create the volume to appear as large or larger than the expanded array. So, either you ignore the handy expansion feature, or you have a volume that your OS thinks is larger, possibly substantially so, than it physically is(just like a counterfeit fleabay flash drive...) Nothing bad can possibly come of having your OS and filesystem capacity numbers being based on lies, with the only true capacity report being through the proprietary dashboard application... Their oddity and heavy dependence on nonstandard host software is also an issue with their higher end iSCSI-supporting products. "- You can purchase an add-on gigabit Ethernet adapter card for your computer if
needed. Note, however, that a regular network adapter card is required, as Data
Robotics, Inc., does not support iSCSI-specific cards, or HBA (host bus adapter)
cards."
Hooray, pay the substantial premium for an iSCSI model vs. the equivalent NAS unit, and don't even get HBA support...(an 8 bay NAS, for example, starts at $2,500. 8 bays of SAN, $3,999...)
I'm not saying that they are a terrible product; but you really have to hate the limitations of normal RAID or make strong aesthetic demands of your storage arrays before it becomes worth looking.
The PCIe spec is flexible enough that, in theory, you could probably network with it(directly, that is, not just by hanging a gig-E chipset off each host, which would be the sane thing to do). PCIe switches are supposed to be used for fanout of a limited number of host lanes to support more peripherals; but you could likely put one in a separate box, with thunderbolt bridges for communication off-board.
It'd be damned expensive, and I'm sure all sorts of horrible things would happen, given that host-host links aren't a design consideration for most gear, so you'd probably need custom-blessed firmware and software for everything involved(or, alternately, you could attempt to use Thunderbolt as a freaky nonstandard interconnect for ATCA boards, if you are sick like that).
An utterly terrible plan, given that cheap and standard network attached storage is already here; but it could probably be done as a stunt(rather like firewire can actually be used for comparatively complex networking, not just glorified crossover-cable usage, only much less supported and probably a worse idea)...
It's true. Remember that the so called "speed of light" was imposed by an atheist jew intellectual, and has since been propped up by an academic elitist cabal supported by big government's tax-and-spend agenda.
In fact, the entire enterprise of physics is inherently statist. It spends essentially all its time and resources imposing as many universal laws as possible. If only physics were deregulated, and the behavior of matter and energy left to the free market, those particles whose behavior is best adapted to the demands of the marketplace would outcompete less efficient matter and create a utopia.
Is there a pool of developers out there, saying to themselves, "I'd totally develop for blackberry; but their kernel is t37 suxxor!"?
If, by some strange chance, the answer is yes, then yes, they should come flocking.
Otherwise, their fortunes will likely continue to depend on how pleasant their systems are to develop for, and how many devices capable of running applications are in the hands of users interested in buying them...
By all accounts, QNX is an accomplished OS; but it doesn't(in itself) solve the direst of problems with RIM's 3rd party dev efforts, which are not so much kernel limitations as user environment, dev tool, and API ones. If RIM can outperform its historical self in those areas, good for them. Otherwise, this "BBX" is going to offer the delightful choice of the same old blackberry crap, or Adobe Flash running like a wounded fainting goat on some flavor of ARM SoC; but with a rock-solid foundation...
It is absolutely false to say that my client is building "shadow profiles" on "non-users". They are doing no such thing.
Admittedly, a simple abundance of caution does require the prudent stewardship of shareholder value in the form of compiling dossiers on non-aligned-persons; but that is an entirely distinct matter.
That's the sort of tosh that sounds very poetic; but is really nonsense.
Moods don't "mean" things: they are physiological states, not symbols. Further, "happy" isn't something you infer by playing compare-and-contrast, it's the immediate introspective impression of a certain state(just as certain sensations on the skin are pleasant per se, not by contrast to being on fire.)
Our present knowledge of psycho-pharmacology and neurology is blunt enough that shooting for permanent happiness is not a particularly good move; but that's a technological problem, not some sort of issue in epistemology.
Slashdot Mirror
I have to wonder if the 16GB "requirement" is more of a recommendation and/or a bunch of default settings that deliberately avoid the disk as much as possible, and keep as many cores as you can throw at the job busy by compiling every little bit and piece in parallel...
On the one hand, with 16GB of RAM in the desktop/light workstation 4x4GB only running around $100(with the more workstation-friendly 2x8GB with ECC only twice that), it seems rather pointless to burn any developer time on trying to optimize the RAM needs of building the entire OS. RAM is cheap.
On the other hand, I have to wonder what they could possibly be doing to the process of compiling what is basically a weird-but-not-unrecognizable linux distro to make it that RAM hungry.
The slightly-tangled-and-mostly-software history of "Hyperion Entertainment" doesn't fill me with worlds of confidence; but there doesn't seem to be anything architecturally implausible about shipping a genuine "AmigaOS netbook", except that volume makes slapping a sticker on an Atom based system cheaper.
Despite ARM having pretty aggressively filled the 'consumer-visible stuff that isn't x86' market(not quite sure why they cleaned up so hard; but they did), there are still plenty of PPC SoCs around that have adequately low power demands and support a reasonable number of modern peripheral interfaces. There are a few desktop-ish boards built around them, no reason you couldn't shove one in a notebook, other than the price efficiencies of volume...
Frankly, this only seems like an "important free speech issue" in one respect: the (quite likely) possibility that the park management are using selective enforcement of (possibly outdated) regulations against people who merely annoy them.
The notion that certain things that incidentally happen to be speech can be curtailed or limited because they are also hazardous or deeply disruptive has been more or less unproblematic as long as the notion of freedom of speech has been a matter of political possibility. However, such limitations do offer a potentially hazardous temptation for anti-speech selective enforcement(Is running around a dense residential district at 3am and shouting your head off legitimately "disturbance of the peace"? Yeah. Does that mean that it would be OK for police to ignore some disturbers of the peace and arrest those who say unpopular things? Not So Much.)
If this case turns out to be the park staff using a permitting system written back when cameras were barely man-portable and 'filming' implied a trail of havoc to selectively quash the weirdos while ignoring That Vacationing Camcorder Asshole, whose life only has meaning if they glimpse it continuously through a viewfinder, they need a smacking down.
If it turns out that the permit requirements are applied uniformly, then it becomes the much less weighty question of whether or not the decreasing size and disruptiveness of cameras makes them due for a rewrite or not...
Your nerd "technical reasons" leave us no room for endless inflammatory arguments and trite gender sterotyping! How will we fill the empty silence now?
Giving the user a choice would imply that they know better than Apple does about what is best.
Apple's entire business model is founded on the operational assumption that they know better than the customer. Their present success is founded on the fact that they are quite often correct about this....
Female computer voices also include SHODAN, GladOS, and basically all vocalized self-destruct countdowns... Can't trust 'em.
Given that they charge $6month/user, you'd think that that would chill spammers out of the service pretty quickly. If you are quasi-legit(yeah, yeah, you're an 'opt-in marketing professional', right...), MS has your payment details and can always nuke your account if you don't heed warnings.
If you are an outright scammer, a major American corporation with a history of litigation against people like you seems like a very odd place to try to pay for spam delivery with your skimmed CC accounts, surely there are better dodgy operators who will be cheaper and more cooperative...
Being free, hotmail anti-spam is much more of a technical problem, since they have to fight robo-signups and account cracking(these days, guessing some idiot's password is probably easier than reading most captchas...)
I'm going to theorize "Moving less product at higher margins is certainly a possibility" on that one. There is, after all, absolutely no reason to suspect that anybody is going to lower prices any further than they absolutely have to.
My point was just that, architecturally, online retailers can modify prices within minutes according to demand, their costs, whatever they can infer about incoming customers' price sensitivity, etc. That doesn't mean that they'll lower prices as fast as they raised them; but it does make grandparent's "That's fine, they will sell exactly ZERO of them until the price goes back to normal." situation quite unlikely: a price at which there are no buyers would be adjusted within a very short time. If it so happens that you can get most of the buyers at twice the price, though, no real reason to move down...
Alpha emitters should be approached with considerable caution:
Because alpha radiation doesn't penetrate well, simple proximity to an alpha source isn't a big deal; but if you manage to absorb an alpha emitter, the alpha radiation no longer has to penetrate well to cause significant damage(just ask the late Mr. Litvinenko).
Effectively, alpha emitters have to be handled as though they possess pretty extreme chemical toxicity. Properly sealed sources are pretty much harmless. Dusts, dissolved compounds, aerosols, etc. are to be avoided.
Given how quickly online retailers can adjust prices(essentially limited only by the speed at which the factors that control pricing decisions change), I'm guessing that they are either selling enough drives to be satisfied with the situation, or simply cannot afford to sell them any more cheaply, because of upstream price increases.
Unless you have absolutely no overhead, and everybody has some, moving zero product isn't a situation you shoot for. Moving less product at higher margins is certainly a possibility, so that can't be ruled out on theoretical grounds; but it seems incredibly unlikely that the retailers have stopped moving stock entirely.
I can't speak for the collective consciousness of Slashdot; but the various 'trusted computing' stuff seems to have exactly the same set of trade-offs now that it did in the 90's: It does make malicious modification(by untrusted 3rd parties, malice by trusted parties actually becomes easier) more difficult; but there isn't an enormous amount of room for optimism about the percentage of devices that will accept the user as the root of trust, rather than whoever the vendor burned in. The number won't be zero, certainly; but it seems only reasonable to expect that the 'trusted' future will be dominated by hardware whose trust list does not include you.
This is picking a nit with the examples, rather than the concept; but both floppies and SD cards have a physical switch in only the loosest sense of the term:
Floppies have no internal logic capable of acting on the switch state, it is entirely up to the floppy drive to sense and obey. SD cards do have an internal controller, and could theoretically enforce write-blocking on themselves; but they don't. Their switch is also just a little plastic tab, and it is entirely up to the reader to sense and obey the tab position. The card's PCB has no connection at all to the switch, and has no way of sensing its position...
Android has really never been an open development model: the applications that make android "android"(from the consumer perspective) are closed, and the development of each successive release has been between Google and their Best Buddy of the moment until release.
The big question has been, since 'Ice Cream Sandwich', whether it would continue to be closed development/open source, or whether it would go closed entirely, except for a few GPL-obligated kernel bits...
I haven't really been able to get a good sense of Drobo reliability at the population level. There are certainly enough horror stories in the wild to incline one not to trust them more than other storage appliances as a class; but I've also had the (dis)pleasure of observing RAID cards that cost more than the low to midrange Drobos, never mind the chassis and redundant PSUs, murder-suicide with all the data downstream of them...
They certainly don't seem to be markedly better than the competition in their price brackets(and since "BeyondRAID" is a proprietary feature, there aren't multiple vendors to play the price/reliability tradeoff game with); but I've not seen evidence that they are markedly worse, in strict reliability terms, just much weirder than all but the horrid little NAS units whose vendors are too cheap to slap a halfway competent interface on top of stock Linux RAID.
While the electrical erasing is certainly a convenient feature, I'm pretty sure that if you can scrape enough of the case off to bombard the die with hard UV, an EEPROM will also be erased(not terribly useful, in practice, just as the epoxy-case windowless EPROMs were functionally WORM media). I'd be fascinated by the gory details if it turns out that EEPROMs are actually light-resistant at a die level, though...
Given that, in the context of ethernet, "Jumbo frame" usually implies a whole 9000 bytes, I'd say that the HDD-based system does have the clear upper hand in potential frame size...
Pity about the latency and the being half-duplex...
Unless you are absolutely phobic about exposure to harrowingly technical terms like "raid5", you should approach the drobo with notrivial caution.
They are quite pricey for their size and performance, which has historically been pretty tepid. Probably worse than that(which is a set of vices shared with quite a few other underpowered NAS units), is that their "BeyondRAID" system makes up for some powerful features by being Just Fucking Weird in some annoying ways.
Perhaps my least favorite is the ghastly hack that they use to make automatic array expansion 'easy'. To quote from their getting started guide:
"- Volume size does not represent how much actual storage space is available on your Drobo/Pro/S. It represents virtual storage space. For example, your Drobo/ Pro/S may be loaded with 2TB of hard drive space, but you can create a volume of 8TB. What this enables you to do is add more capacity to your Drobo/Pro/S (by inserting an additional drive or replacing a smaller capacity drive with a larger capacity one) without having to format an additional volume. The additional capacity becomes part of the same volume you formatted originally. - Your operating system may show the virtual space you have available on your Drobo device, as defined by the volume size."
Yup. Unlike traditional RAID, you don't have to break and re-create the array to enlarge it, which is nice; but only if you initially create the volume to appear as large or larger than the expanded array. So, either you ignore the handy expansion feature, or you have a volume that your OS thinks is larger, possibly substantially so, than it physically is(just like a counterfeit fleabay flash drive...) Nothing bad can possibly come of having your OS and filesystem capacity numbers being based on lies, with the only true capacity report being through the proprietary dashboard application... Their oddity and heavy dependence on nonstandard host software is also an issue with their higher end iSCSI-supporting products. "- You can purchase an add-on gigabit Ethernet adapter card for your computer if needed. Note, however, that a regular network adapter card is required, as Data Robotics, Inc., does not support iSCSI-specific cards, or HBA (host bus adapter) cards."
Hooray, pay the substantial premium for an iSCSI model vs. the equivalent NAS unit, and don't even get HBA support...(an 8 bay NAS, for example, starts at $2,500. 8 bays of SAN, $3,999...)
I'm not saying that they are a terrible product; but you really have to hate the limitations of normal RAID or make strong aesthetic demands of your storage arrays before it becomes worth looking.
The PCIe spec is flexible enough that, in theory, you could probably network with it(directly, that is, not just by hanging a gig-E chipset off each host, which would be the sane thing to do). PCIe switches are supposed to be used for fanout of a limited number of host lanes to support more peripherals; but you could likely put one in a separate box, with thunderbolt bridges for communication off-board.
It'd be damned expensive, and I'm sure all sorts of horrible things would happen, given that host-host links aren't a design consideration for most gear, so you'd probably need custom-blessed firmware and software for everything involved(or, alternately, you could attempt to use Thunderbolt as a freaky nonstandard interconnect for ATCA boards, if you are sick like that).
An utterly terrible plan, given that cheap and standard network attached storage is already here; but it could probably be done as a stunt(rather like firewire can actually be used for comparatively complex networking, not just glorified crossover-cable usage, only much less supported and probably a worse idea)...
It's true. Remember that the so called "speed of light" was imposed by an atheist jew intellectual, and has since been propped up by an academic elitist cabal supported by big government's tax-and-spend agenda.
In fact, the entire enterprise of physics is inherently statist. It spends essentially all its time and resources imposing as many universal laws as possible. If only physics were deregulated, and the behavior of matter and energy left to the free market, those particles whose behavior is best adapted to the demands of the marketplace would outcompete less efficient matter and create a utopia.
The sort of fail that can be totally erased by 30 minutes of hard UV is sadly rare these days...
Is there a pool of developers out there, saying to themselves, "I'd totally develop for blackberry; but their kernel is t37 suxxor!"?
If, by some strange chance, the answer is yes, then yes, they should come flocking.
Otherwise, their fortunes will likely continue to depend on how pleasant their systems are to develop for, and how many devices capable of running applications are in the hands of users interested in buying them...
By all accounts, QNX is an accomplished OS; but it doesn't(in itself) solve the direst of problems with RIM's 3rd party dev efforts, which are not so much kernel limitations as user environment, dev tool, and API ones. If RIM can outperform its historical self in those areas, good for them. Otherwise, this "BBX" is going to offer the delightful choice of the same old blackberry crap, or Adobe Flash running like a wounded fainting goat on some flavor of ARM SoC; but with a rock-solid foundation...
I've always found "Intravenous Diazepam" by the Swiss/international group 'Roche' to be more relaxing, personally.
It is absolutely false to say that my client is building "shadow profiles" on "non-users". They are doing no such thing.
Admittedly, a simple abundance of caution does require the prudent stewardship of shareholder value in the form of compiling dossiers on non-aligned-persons; but that is an entirely distinct matter.
If you can put 100% trust in a programmable device, and tell it to behave in a certain way, you can be sure that it will behave in a certain way!
It's Genius!
That's the sort of tosh that sounds very poetic; but is really nonsense.
Moods don't "mean" things: they are physiological states, not symbols. Further, "happy" isn't something you infer by playing compare-and-contrast, it's the immediate introspective impression of a certain state(just as certain sensations on the skin are pleasant per se, not by contrast to being on fire.)
Our present knowledge of psycho-pharmacology and neurology is blunt enough that shooting for permanent happiness is not a particularly good move; but that's a technological problem, not some sort of issue in epistemology.