Depends on the optics and the weather, I would suspect.
On a clear day, with excellent optics, probably surprisingly little(In Lunar laser rangefinding experiments, the laser spreads from being a near-point-source to a mere 4 mile diameter spot across ~240,000 miles). Your not-at-all-pricey 250-500mW DPSS greens would probably do just fine, if you could keep them stable and on target.
If your optics are shit, or there is fog/dust/substantial thermal shimmer, requirements would go up markedly...
The other issue, with electronic attacks specifically, is that effective "self defense" would require absurdly broad authorization.
In physical terms, you have states like Texas, where shooting trespassers is largely legal, and states like Massachusetts where you pretty much have to have run out of other options before you can use lethal force in self defense. When it comes to electronic attacks, everybody already enjoys greater-than-Texas level of self-defense capability. I can tell my routers and switches to drop whatever packets I want them to. I can terminate whatever processes I care to on my hardware. I can delete whatever files, etc. My network, my rules.
Given that everyone, in basically every jurisdiction anywhere, can already do that any call for expanded powers of self defense is a call to be allowed to just start shooting up the neighborhood with wild abandon. Not going to end well.
Unless the window has been replaced within about the last 20 minutes, it will likely have enough tiny scratches and little bits of crud and whatnot on it that just hitting the window will cause a large portion of it to light up a delightful bright green. Still not trivial; but you don't need to hit the eyeball, just the window.
That said, if people are that freaked out about it, they could just have the co-pilot wear a $30 pair of laser-safety glasses on approach, and take over if the pilot gets dazzled.
Longer term, it would be relatively inexpensive(in the grand scheme of passenger aircraft avionics) to equip the pilots with multiple screens showing feeds from cameras scattered strategically around the outside of the aircraft. Good luck blinding 6-8 camera lenses scattered around a 747's body without equipment so costly that you could save money by using a black-market Stinger(or equivalent) instead...
If I am firing hijacked passenger airliners at you, are the criminal homicide charges and the civil wrongful death suits that you would accrue by shooting them down worth it?
That's the problem: there is basically no such thing as a pure weapon on the internet. Most "stolen missiles" are simultaneously poorly secured home or business computers that have never left the ownership(and, in general, since the botnet guys don't want their hosts getting wiped) are still being actively used by their owners for whatever their intended purpose is.
Crippling them would, indeed, end the attack; but it would constitute committing dozens or hundreds of what(at least in the US) would be federal felonies and invitations to expensive civil suit. And, to be quite blunt about it, you would deserve to have your ass handed to you for doing so.
While the game theory behind spite(hurting others even at a cost to oneself, the under-appreciated counterpart to altruism, helping others even at a cost to oneself) is interesting, and suggests that it can actually be vital in maintaining some mutually beneficial equilibria, all that breaks down if the assumption that retailiation can be accurately allocated is violated. It also breaks down if the assumption that all agents are indivisible is violated.
Risk is a game of essentially perfect information(aside from the interpersonal alliance metadata). Everything on the board was in the open. Ye Olde Intertubes are often not so obliging. If I am hitting you through a botnet of compromised home users that I am renting from some botnet herder, do I fear your retaliation? Only if I think you are good enough to allocate it to me, despite multiple levels of indirection, and essentially innocent targets standing in your way.
Even if I am attacking directly, say from a colo owned by a shady shell corporation, the second assumption is violated. The shell corporation is an expendable appendage, nuke it into the ground for all I care, I've already extracted the value and moved on.
In the US, and in the sorts of theoretically-rule-of-law-y jurisdictions that corporations generally have substantial operations and assets in, most flavors of "cyberattack" are de jure Pretty. Seriously. Not. Legal.
This does approximately jack shit against gangs operating offshore in who-knows-where controlling botnets of enslaved Joe User XP home boxes; but it is the state of the law. Now, let's think about this for a second: Any "cyber-counterattack", unless unbelievably flawless, is probably going to have some amount of collateral damage: ISPs getting parts of their networks DDOSed, innocent-if-clueless home users getting their botnetted boxes taken down, etc. Even the direct damage will be illegal(though criminal gangs probably won't press charges); but the collateral damage will, in not a few cases, fall directly on people and businesses, in western jurisdictions, who had nothing to do with the original attack(other than, perhaps, not updating their AV often enough).
Now, when it comes to light that Foocorp LLC, a division of Deeppockets Industries, and their officers and employees have been guilty of numerous violations of federal cybercrime violations, most felonies, and a variety of civilly actionable property damage, where do you think the lawyers are going to go looking for blood? Yuri Shadymov and John Does 1-N, the mysterious perpetrators of the attack on Foocorp, or the conveniently-located-right-at-home Deeppockets Industries?
There would be a nonzero risk(and they would deserve every bit of it) that Deeppockets industries could find itself up to its eyeballs in civil suits, and the Foocorp IT team and every exec who knew of and authorized their actions could be looking at serious fines and some quality time in FPMITA...
I don't know of any implementation of a a mouse + touchscreen specifically; but keyboards with LCDs for dynamic reconfiguration are shipping products, wacom digitizer tablet screens are shipping products, mice with dynamic settings(sensitivity, scroll speed, etc.) indicated by LEDs on the mouse are shipping products, and the general notion that a touchscreen can present a reconfigurable set of buttons is basically the foundation of the contemporary smartphone industry.
Novel product? As best I can tell. Patent worthy? Srlsy?
Unless Sony really fucks up their crypto(again), no level of network cleverness will help. SSL/TLS verification of a remote host isn't rocket surgery...
I don't have much experience with Dell's cheap seats, only their corporate stuff, for which serviceability is clearly among the top priorities.
It really uglies up the chassis exterior; but I've seen reasonably experienced techs working on modestly recent designs(say Optiplex 520s and later) remove and replace every FRU in the unit with one screwdriver in under ten minutes. I'm not in that kind of practice; but even someone who has never done it before can usually manage the same, without a service guide, in 45 or so.
Some of the older designs are pure junk(SFF GX110 anybody? I've never seen the top cover of one of those go back on correctly within the first 3 tries); but the newer ones are all business.
True. Were they to go that route, I would assume that they would bring out a few pins, connected to a teeny dedicated bus controller, so that someone with the correct cryptographic handshake could open even a device whose primary CPU is dead or hard locked...
You can sometimes get good results with weird-ass screws by the following trick:
Obtain a cheap screwdriver of the sort sold with "modular" sets that contain one handle with a socket and a bunch of bits. Should run a few bucks retail or online.
Obtain some silicone spray lubricant.
Obtain some high-strength epoxy(putty is easiest) "J-B STIK" or similar.
Coat the inside of the screwdriver socket and the head of the exotic screw with silicone spray. This will keep the epoxy from sticking and act as a sort of mold release agent.
Prepare the epoxy according to instructions and fill the head of the screwdriver with it. Leave some excess sticking out. When the epoxy is of putty-like consistency, press firmly into the head of the exotic screw, to create a "positive" in the epoxy corresponding to the void in the screw head. As soon as it has firmed up enough to do so, remove the epoxy slug from the screwdriver, you don't want it sticking there.
After the manufacturer's recommended cure time for the epoxy, you now have a custom screwdriver bit that is compatible with modular handle systems and is a perfect fit for whatever weirdo screw you have encountered. Don't expect it to last as long as a proper, well manufactured bit; but it should do the job for low-torque stuff.
For 'one-way' screws, or if you don't care about the survival of the exotic screws, there is an easier way: Obtain a quantity of standard screws, for which you have a driver, and cut the heads off. Epoxy a standard head onto the head of each exotic screw. Allow to cure and remove.
I can understand switching away from Phillips, since those suckers strip pretty badly even in larger sizes, and strip like it's their job in smaller ones; but switching to some totally oddball screw type, as opposed to one of the various fairly standardized strip-resistant heads already in use in electronics and elsewhere seems like a dick move.
Once so called "smart screws" hit the market. The idea has been in the theory/laboratory stage for some years now: basically fasteners that, under electrical control, can move between their fastened and unfastened or extended/retracted states(assorted pizieo, MEMS, tiny motor, etc. principles of operation have been tried).
Cool thing is, since you no longer have to be able to reach the head of the fastener with a driver, it becomes possible to do case and assembly designs that would be impossible with conventional fasteners. On the minus side, if the fasteners are no longer exposed, and under electrical control(via a simple bus in the chassis) you'll have to gain software control of the device just to open it(without extreme violence to the case. Obviously, nothing resists a good power tool for long...)
They are, admittedly, ugly fuckers; but every desktop of theirs that I've dealt with in the past 4 or 5 years has been held together with a mixture of screwless plastic pieces(they've standardized on green as a visual code for "this plastic piece is an FRU) and hex-head phillips screws that can be removed with either a phillips or hex tool. Usually all the same length, too.
Laptops tend to have some variation in length, and don't feature the convenient dual hex/phillips; but you can take the entire laptop to bits with a single phillips screwdriver, and each screw hole is labelled with the length of the screw that goes into it(ugly, yes, convenient, also yes...)
While I agree that obsessing about minute risks is vastly overrated and overindulged in, you seem to be missing an important point that really makes your position that of an utter prick.
Unless the airline really fucked up on bookings, you are going to be sharing that metal tube with 100+ people. Also, that tube belongs to somebody, and costs some millions of dollars. It's flight plan is such that a crash at quite a few points along the way will involve landing on some people and/or property. While I fully support your right to take up your own plane, drop more LSD than Hunter S. Thompson, and treat the avionics interference from your onboard tesla coil as a substitute co-pilot(so long as you can pick a suitably out-of-the-way crash site); imposing your risk assessment preferences on all your fellow passengers and the owner of the aircraft (while unlikely to have any effect) is deep in "narcissistic prick" territory.
IIRC, aircraft, at least the reasonably high altitude ones, have to be designed to cope with the possibility of lightning strike(not really as bad as it sounds, even badass voltages are relatively harmless when you are inside an aluminum tube). Lighting, of course, is basically the biggest, meanest spark-gap in the entire terrestrial context(compared to, say, Jovian lightning, it isn't much at all, but that isn't really relevant to any aircraft except Xenu's space-DC9s...).
Spark gaps tend to put out some seriously gross, broad band, RF noise. A spark gap with the energy of a lighting bolt should be quite the RF emitter.
Unless the designers depend exclusively on the aircraft's outer skin for RF protection(which seems unlikely, given the systems that need to communicate and/or scan the outside world, which obviously can't be faraday-caged inside the outer skin...) they have presumably had to deal with RF of the sort that would make your weedy little powered-by-batteries-and-FCC-regulated widget wet itself.
Also one would sincerely hope, given what the higher level of cosmic ray exposure can(with low but nonzero frequency) do in terms of flipping bits in any circuitry that isn't rad-hard, critical systems would be redundant, watchdogged and quick to reboot, or both.
Given that their scheme is 3rd-party to facebook(if this were a facebook scheme, the crypto nonsense wouldn't be necessary, they could just have privacy controls that actually work, rather than being broken by design), I don't see why a keyserver run by you would be any less acceptable than a keyserver run by them(though you might have to resort to some horrid trick like using a placeholder image with the ciphertext embedded as EXIF data or something).
I'm just inferring, based on the fact that this is aimed at the demographic that has a serious problem with posting embarrasing images to facebook, that their plans do not rely on users running their own keyservers. Nothing stopping you from implementing the same idea with your own keyserver(other than the idea being stupid); but my understanding of their plan was that they keyserver would not be something that the user is expected to manage(which, realistically, is probably a very sensible assumption on their part, though it certainly doesn't give the privacy wonks who will never need their services anyway the warm and fuzzies...)
Do you think that those submarine systems designs that they are dusting off escaped marine cost expansion when they were written up, back when marine reactors were being designed and refined based on the initial terrestrial hardware?
Obviously, using a variant of systems you have already designed is cheaper than grabbing the plans for a terrestrial reactor and cutting it down until it fits; but those submarine systems presumably include all the costs of hardening the reactor against salt, pressure, and other aquatic nastiness. By using the same design, you skip the design costs; but you still incur those construction costs each time you kick out another one.
That was the core of my question: pretty much every part of nuclear submarine design comes down to military necessity and doesn't come cheap. What advantages are compelling enough to do that for a civilian reactor, rather than just grabbing an off-the-shelf big boat and/or drilling platform design and just dropping a reactor on top of it? Still more expensive than land; but cheaper than a nuclear submarine derivative.
Not that I consider attack scenarios all that likely(getting a commercial reactor to do anything terribly dramatic is actually pretty hard, if you just want a dirty bomb there are plenty of industrial and medical sources that are less carefully watched, etc.); but there is one point of vulnerability worth noting...
If this widget is to be useful, it will need a power/control conduit connecting it to land somewhere nearby. To carry the amounts of power we are talking about, that is going to be a big, obvious cable, with beefy conductors, probably an armored shell, and a nice fat magnetic signature.
Deep diving and underseas demolition work is a fairly rare and specialized skill. Building a drone with, say, magnetic wheels allowing it to clip onto the conduit on the beach or in shallow water and just keep rolling until it finds the sub would be a college level engineering project. It is also quite likely that somebody with knowledge of the local marine biology could detect the huge thermal plume just by observing the response of local sea life. Within a year or two of installation, you would probably see a pretty marked sorting of warmer water and colder water species by preference...
As I said, a nuclear reactor, land, floating, or undersea, is a target with a lousy difficulty/value ratio, so hitting one wouldn't make too much sense; but cabling something into the electrical grid and hiding it effectively are not complementary goals. Not impossible; but sufficiently tricky and expensive that you would probably be better off just making it more durable with the money you save by not bothering.
My impression(not speaking as an expert shipwright or anything) is that if you want to take a land-based system and get it going for reliable marine use, you'll be lucky if the cost doubles(Boat. Noun. A hole in the water into which one pours money). That, though, I I can see the benefits of. The art and science of building large floating objects is pretty well established, and then you pretty much plunk the reactor on top of that. Nice and portable, coolant all around, and sure beats trying to make your nuclear reactor a helicopter or something. Float it where you need it, run a glorified extension cable to shore, and away you go.
Underwater, though, just seems like a recipe for making the whole thing even more expensive than on the water, along with harder to monitor and maintain, and likely to be much more exciting if there is a steam leak or something. Is there some advantage that I am not seeing, or is this a case of "when you are a post-cold-war-nuclear-submarine-designer everything looks like it needs an underwater nuclear reactor"?
I certainly have nothing against local backups(as the NAS humming merrily a few feet from my head and keeping copies of home directories from machines on the LAN could attest...); I just find them to be a dubious archival solution. They are an excellent preservative against drive failures or needing to rebuild a hosed machine.
For the long term, though, since the consumer market has been relentlessly optimizing in favor of cheap and voluminous, with reliability only a consideration to 5 years or so, I have trouble coming up with a backup scheme that doesn't involve either trusting something that shouldn't be trusted for longer than it should be trusted or a substantial amount of manual drive-swapping and replacement. At that point, the economies of scale of having one screwdriver jockey who can drive swap for 10,000+ customers start to look pretty attractive, even given the slow restore times over my fairly pathetic connection.
I don't like having something nibbling at my credit card either; but the going rate for general-purpose-backup of a PC seems to be about $50-60 a year(something photo-only like Flickr pro about half that). So, I can pretty much purchase a single backup HDD, which I then have to manually swap around, for a year's worth of backup costs. If I want something fancier(or something that will resist my house being burgled or catching fire), like a RAIDed NAS, we are easily talking 3+ years worth of just letting somebody else do it. Unless my memory is excellent and my time close to worthless, I just can't make the case for it...
Slashdot Mirror
Depends on the optics and the weather, I would suspect.
On a clear day, with excellent optics, probably surprisingly little(In Lunar laser rangefinding experiments, the laser spreads from being a near-point-source to a mere 4 mile diameter spot across ~240,000 miles). Your not-at-all-pricey 250-500mW DPSS greens would probably do just fine, if you could keep them stable and on target.
If your optics are shit, or there is fog/dust/substantial thermal shimmer, requirements would go up markedly...
The other issue, with electronic attacks specifically, is that effective "self defense" would require absurdly broad authorization.
In physical terms, you have states like Texas, where shooting trespassers is largely legal, and states like Massachusetts where you pretty much have to have run out of other options before you can use lethal force in self defense. When it comes to electronic attacks, everybody already enjoys greater-than-Texas level of self-defense capability. I can tell my routers and switches to drop whatever packets I want them to. I can terminate whatever processes I care to on my hardware. I can delete whatever files, etc. My network, my rules.
Given that everyone, in basically every jurisdiction anywhere, can already do that any call for expanded powers of self defense is a call to be allowed to just start shooting up the neighborhood with wild abandon. Not going to end well.
Unless the window has been replaced within about the last 20 minutes, it will likely have enough tiny scratches and little bits of crud and whatnot on it that just hitting the window will cause a large portion of it to light up a delightful bright green. Still not trivial; but you don't need to hit the eyeball, just the window.
That said, if people are that freaked out about it, they could just have the co-pilot wear a $30 pair of laser-safety glasses on approach, and take over if the pilot gets dazzled.
Longer term, it would be relatively inexpensive(in the grand scheme of passenger aircraft avionics) to equip the pilots with multiple screens showing feeds from cameras scattered strategically around the outside of the aircraft. Good luck blinding 6-8 camera lenses scattered around a 747's body without equipment so costly that you could save money by using a black-market Stinger(or equivalent) instead...
If I am firing hijacked passenger airliners at you, are the criminal homicide charges and the civil wrongful death suits that you would accrue by shooting them down worth it?
That's the problem: there is basically no such thing as a pure weapon on the internet. Most "stolen missiles" are simultaneously poorly secured home or business computers that have never left the ownership(and, in general, since the botnet guys don't want their hosts getting wiped) are still being actively used by their owners for whatever their intended purpose is.
Crippling them would, indeed, end the attack; but it would constitute committing dozens or hundreds of what(at least in the US) would be federal felonies and invitations to expensive civil suit. And, to be quite blunt about it, you would deserve to have your ass handed to you for doing so.
While the game theory behind spite(hurting others even at a cost to oneself, the under-appreciated counterpart to altruism, helping others even at a cost to oneself) is interesting, and suggests that it can actually be vital in maintaining some mutually beneficial equilibria, all that breaks down if the assumption that retailiation can be accurately allocated is violated. It also breaks down if the assumption that all agents are indivisible is violated.
Risk is a game of essentially perfect information(aside from the interpersonal alliance metadata). Everything on the board was in the open. Ye Olde Intertubes are often not so obliging. If I am hitting you through a botnet of compromised home users that I am renting from some botnet herder, do I fear your retaliation? Only if I think you are good enough to allocate it to me, despite multiple levels of indirection, and essentially innocent targets standing in your way.
Even if I am attacking directly, say from a colo owned by a shady shell corporation, the second assumption is violated. The shell corporation is an expendable appendage, nuke it into the ground for all I care, I've already extracted the value and moved on.
Novel and non-obvious = Patent Worthy.
In the US, and in the sorts of theoretically-rule-of-law-y jurisdictions that corporations generally have substantial operations and assets in, most flavors of "cyberattack" are de jure Pretty. Seriously. Not. Legal.
This does approximately jack shit against gangs operating offshore in who-knows-where controlling botnets of enslaved Joe User XP home boxes; but it is the state of the law. Now, let's think about this for a second: Any "cyber-counterattack", unless unbelievably flawless, is probably going to have some amount of collateral damage: ISPs getting parts of their networks DDOSed, innocent-if-clueless home users getting their botnetted boxes taken down, etc. Even the direct damage will be illegal(though criminal gangs probably won't press charges); but the collateral damage will, in not a few cases, fall directly on people and businesses, in western jurisdictions, who had nothing to do with the original attack(other than, perhaps, not updating their AV often enough).
Now, when it comes to light that Foocorp LLC, a division of Deeppockets Industries, and their officers and employees have been guilty of numerous violations of federal cybercrime violations, most felonies, and a variety of civilly actionable property damage, where do you think the lawyers are going to go looking for blood? Yuri Shadymov and John Does 1-N, the mysterious perpetrators of the attack on Foocorp, or the conveniently-located-right-at-home Deeppockets Industries?
There would be a nonzero risk(and they would deserve every bit of it) that Deeppockets industries could find itself up to its eyeballs in civil suits, and the Foocorp IT team and every exec who knew of and authorized their actions could be looking at serious fines and some quality time in FPMITA...
You are holding it wrong.
-S. Jobs
I don't know of any implementation of a a mouse + touchscreen specifically; but keyboards with LCDs for dynamic reconfiguration are shipping products, wacom digitizer tablet screens are shipping products, mice with dynamic settings(sensitivity, scroll speed, etc.) indicated by LEDs on the mouse are shipping products, and the general notion that a touchscreen can present a reconfigurable set of buttons is basically the foundation of the contemporary smartphone industry.
Novel product? As best I can tell. Patent worthy? Srlsy?
Unless Sony really fucks up their crypto(again), no level of network cleverness will help. SSL/TLS verification of a remote host isn't rocket surgery...
I don't have much experience with Dell's cheap seats, only their corporate stuff, for which serviceability is clearly among the top priorities.
It really uglies up the chassis exterior; but I've seen reasonably experienced techs working on modestly recent designs(say Optiplex 520s and later) remove and replace every FRU in the unit with one screwdriver in under ten minutes. I'm not in that kind of practice; but even someone who has never done it before can usually manage the same, without a service guide, in 45 or so.
Some of the older designs are pure junk(SFF GX110 anybody? I've never seen the top cover of one of those go back on correctly within the first 3 tries); but the newer ones are all business.
True. Were they to go that route, I would assume that they would bring out a few pins, connected to a teeny dedicated bus controller, so that someone with the correct cryptographic handshake could open even a device whose primary CPU is dead or hard locked...
I have no idea what the time horizon is on this, just that it is being worked on.
This stuff might also allow some interesting 'sealed but serviceable' designs...
You can sometimes get good results with weird-ass screws by the following trick:
Obtain a cheap screwdriver of the sort sold with "modular" sets that contain one handle with a socket and a bunch of bits. Should run a few bucks retail or online.
Obtain some silicone spray lubricant.
Obtain some high-strength epoxy(putty is easiest) "J-B STIK" or similar.
Coat the inside of the screwdriver socket and the head of the exotic screw with silicone spray. This will keep the epoxy from sticking and act as a sort of mold release agent.
Prepare the epoxy according to instructions and fill the head of the screwdriver with it. Leave some excess sticking out. When the epoxy is of putty-like consistency, press firmly into the head of the exotic screw, to create a "positive" in the epoxy corresponding to the void in the screw head. As soon as it has firmed up enough to do so, remove the epoxy slug from the screwdriver, you don't want it sticking there.
After the manufacturer's recommended cure time for the epoxy, you now have a custom screwdriver bit that is compatible with modular handle systems and is a perfect fit for whatever weirdo screw you have encountered. Don't expect it to last as long as a proper, well manufactured bit; but it should do the job for low-torque stuff.
For 'one-way' screws, or if you don't care about the survival of the exotic screws, there is an easier way: Obtain a quantity of standard screws, for which you have a driver, and cut the heads off. Epoxy a standard head onto the head of each exotic screw. Allow to cure and remove.
I can understand switching away from Phillips, since those suckers strip pretty badly even in larger sizes, and strip like it's their job in smaller ones; but switching to some totally oddball screw type, as opposed to one of the various fairly standardized strip-resistant heads already in use in electronics and elsewhere seems like a dick move.
Once so called "smart screws" hit the market. The idea has been in the theory/laboratory stage for some years now: basically fasteners that, under electrical control, can move between their fastened and unfastened or extended/retracted states(assorted pizieo, MEMS, tiny motor, etc. principles of operation have been tried).
Cool thing is, since you no longer have to be able to reach the head of the fastener with a driver, it becomes possible to do case and assembly designs that would be impossible with conventional fasteners. On the minus side, if the fasteners are no longer exposed, and under electrical control(via a simple bus in the chassis) you'll have to gain software control of the device just to open it(without extreme violence to the case. Obviously, nothing resists a good power tool for long...)
When did you last open a Dell?
They are, admittedly, ugly fuckers; but every desktop of theirs that I've dealt with in the past 4 or 5 years has been held together with a mixture of screwless plastic pieces(they've standardized on green as a visual code for "this plastic piece is an FRU) and hex-head phillips screws that can be removed with either a phillips or hex tool. Usually all the same length, too.
Laptops tend to have some variation in length, and don't feature the convenient dual hex/phillips; but you can take the entire laptop to bits with a single phillips screwdriver, and each screw hole is labelled with the length of the screw that goes into it(ugly, yes, convenient, also yes...)
Toshiba, on the other hand...
While I agree that obsessing about minute risks is vastly overrated and overindulged in, you seem to be missing an important point that really makes your position that of an utter prick.
Unless the airline really fucked up on bookings, you are going to be sharing that metal tube with 100+ people. Also, that tube belongs to somebody, and costs some millions of dollars. It's flight plan is such that a crash at quite a few points along the way will involve landing on some people and/or property. While I fully support your right to take up your own plane, drop more LSD than Hunter S. Thompson, and treat the avionics interference from your onboard tesla coil as a substitute co-pilot(so long as you can pick a suitably out-of-the-way crash site); imposing your risk assessment preferences on all your fellow passengers and the owner of the aircraft (while unlikely to have any effect) is deep in "narcissistic prick" territory.
IIRC, aircraft, at least the reasonably high altitude ones, have to be designed to cope with the possibility of lightning strike(not really as bad as it sounds, even badass voltages are relatively harmless when you are inside an aluminum tube). Lighting, of course, is basically the biggest, meanest spark-gap in the entire terrestrial context(compared to, say, Jovian lightning, it isn't much at all, but that isn't really relevant to any aircraft except Xenu's space-DC9s...).
Spark gaps tend to put out some seriously gross, broad band, RF noise. A spark gap with the energy of a lighting bolt should be quite the RF emitter.
Unless the designers depend exclusively on the aircraft's outer skin for RF protection(which seems unlikely, given the systems that need to communicate and/or scan the outside world, which obviously can't be faraday-caged inside the outer skin...) they have presumably had to deal with RF of the sort that would make your weedy little powered-by-batteries-and-FCC-regulated widget wet itself.
Also one would sincerely hope, given what the higher level of cosmic ray exposure can(with low but nonzero frequency) do in terms of flipping bits in any circuitry that isn't rad-hard, critical systems would be redundant, watchdogged and quick to reboot, or both.
Never. Ever. Ever. fuck with an angry thermite colony...
Even fire ants won't burn their way through an engine block just to get at you.
Given that their scheme is 3rd-party to facebook(if this were a facebook scheme, the crypto nonsense wouldn't be necessary, they could just have privacy controls that actually work, rather than being broken by design), I don't see why a keyserver run by you would be any less acceptable than a keyserver run by them(though you might have to resort to some horrid trick like using a placeholder image with the ciphertext embedded as EXIF data or something).
I'm just inferring, based on the fact that this is aimed at the demographic that has a serious problem with posting embarrasing images to facebook, that their plans do not rely on users running their own keyservers. Nothing stopping you from implementing the same idea with your own keyserver(other than the idea being stupid); but my understanding of their plan was that they keyserver would not be something that the user is expected to manage(which, realistically, is probably a very sensible assumption on their part, though it certainly doesn't give the privacy wonks who will never need their services anyway the warm and fuzzies...)
Do you think that those submarine systems designs that they are dusting off escaped marine cost expansion when they were written up, back when marine reactors were being designed and refined based on the initial terrestrial hardware?
Obviously, using a variant of systems you have already designed is cheaper than grabbing the plans for a terrestrial reactor and cutting it down until it fits; but those submarine systems presumably include all the costs of hardening the reactor against salt, pressure, and other aquatic nastiness. By using the same design, you skip the design costs; but you still incur those construction costs each time you kick out another one.
That was the core of my question: pretty much every part of nuclear submarine design comes down to military necessity and doesn't come cheap. What advantages are compelling enough to do that for a civilian reactor, rather than just grabbing an off-the-shelf big boat and/or drilling platform design and just dropping a reactor on top of it? Still more expensive than land; but cheaper than a nuclear submarine derivative.
Not that I consider attack scenarios all that likely(getting a commercial reactor to do anything terribly dramatic is actually pretty hard, if you just want a dirty bomb there are plenty of industrial and medical sources that are less carefully watched, etc.); but there is one point of vulnerability worth noting...
If this widget is to be useful, it will need a power/control conduit connecting it to land somewhere nearby. To carry the amounts of power we are talking about, that is going to be a big, obvious cable, with beefy conductors, probably an armored shell, and a nice fat magnetic signature.
Deep diving and underseas demolition work is a fairly rare and specialized skill. Building a drone with, say, magnetic wheels allowing it to clip onto the conduit on the beach or in shallow water and just keep rolling until it finds the sub would be a college level engineering project. It is also quite likely that somebody with knowledge of the local marine biology could detect the huge thermal plume just by observing the response of local sea life. Within a year or two of installation, you would probably see a pretty marked sorting of warmer water and colder water species by preference...
As I said, a nuclear reactor, land, floating, or undersea, is a target with a lousy difficulty/value ratio, so hitting one wouldn't make too much sense; but cabling something into the electrical grid and hiding it effectively are not complementary goals. Not impossible; but sufficiently tricky and expensive that you would probably be better off just making it more durable with the money you save by not bothering.
My impression(not speaking as an expert shipwright or anything) is that if you want to take a land-based system and get it going for reliable marine use, you'll be lucky if the cost doubles(Boat. Noun. A hole in the water into which one pours money). That, though, I I can see the benefits of. The art and science of building large floating objects is pretty well established, and then you pretty much plunk the reactor on top of that. Nice and portable, coolant all around, and sure beats trying to make your nuclear reactor a helicopter or something. Float it where you need it, run a glorified extension cable to shore, and away you go.
Underwater, though, just seems like a recipe for making the whole thing even more expensive than on the water, along with harder to monitor and maintain, and likely to be much more exciting if there is a steam leak or something. Is there some advantage that I am not seeing, or is this a case of "when you are a post-cold-war-nuclear-submarine-designer everything looks like it needs an underwater nuclear reactor"?
I certainly have nothing against local backups(as the NAS humming merrily a few feet from my head and keeping copies of home directories from machines on the LAN could attest...); I just find them to be a dubious archival solution. They are an excellent preservative against drive failures or needing to rebuild a hosed machine.
For the long term, though, since the consumer market has been relentlessly optimizing in favor of cheap and voluminous, with reliability only a consideration to 5 years or so, I have trouble coming up with a backup scheme that doesn't involve either trusting something that shouldn't be trusted for longer than it should be trusted or a substantial amount of manual drive-swapping and replacement. At that point, the economies of scale of having one screwdriver jockey who can drive swap for 10,000+ customers start to look pretty attractive, even given the slow restore times over my fairly pathetic connection.
I don't like having something nibbling at my credit card either; but the going rate for general-purpose-backup of a PC seems to be about $50-60 a year(something photo-only like Flickr pro about half that). So, I can pretty much purchase a single backup HDD, which I then have to manually swap around, for a year's worth of backup costs. If I want something fancier(or something that will resist my house being burgled or catching fire), like a RAIDed NAS, we are easily talking 3+ years worth of just letting somebody else do it. Unless my memory is excellent and my time close to worthless, I just can't make the case for it...