1. Some alien civilizations found wireless to be slow and inefficient, and have adopted dense mode optic fibres, waveguides and other physical networking for most of their stuff. The same route we're taking. You only need enough for the density of radio transmitters in the galaxy to be very low. We can still detect them, using the method proposed by James Lovelock when he was at NASA - observe the atmosphere instead and look for combinations that can't arise naturally.
2. Alien civilizations have an abundance of less than one technological race in every 25 light-years (which is the current extreme range we can detect them). We won't be able to go past even the 100 light-year range until 2024, when SKA comes online. In other words, we've wildly overestimated our abilities.
3. We're pointing telescopes at systems without habitable planets. Now that we actually have the beginnings of a list of exoplanets, let's look and see how many of those which are viable we've actually looked at. Probably very few.
4. Sir Fred Hoyle was right, life rarely evolves on planets, it evolves in deep space. In that case, we're looking in the wrong places. The galaxy could be buzzing with life and we're ignoring it.
5. Encrypted messages are indistinguishable from a random oracle. They're all using Intergalactic IPSec because PRIVACY MATTERS to any evolved species.
As Plato pointed out, ignorance is expensive and dangerous to society. Society chooses to elect leaders who underfund and undercut education. The leaders didn't get there on their own. Ultimately, we are all responsible for all the consequences of our actions, however far down the road they are, particularly in the case of a perfect storm. And privacy has been a perfect storm.
We've known this for decades. Public Squid servers were pilloried. Public anonymous remailers were seized and burned at the stake, with public approval.
ISPs are dangerous to your privacy and physically destroy the networks of rivals, so what did Tennessee do? It passed a Constitutional Amendment banning Chattanooga's municipal system from expanding, after failing to get it outlawed entirely. And the folks of Tennessee? They have no problem defending the loss of their privacy at all and actively support those who despoiled it.
Online advertising cannot personalize your ads if you're using a Squid proxy for everything (which we did back then, because it was faster than going direct, and the JANET backbone supplied Squid at just about every university).
Social media. Ummm, what do you call Slashdot? Anyways, social media is not a particular privacy concern, for the same reason online ads aren't. They can't track you if you use Squid. But the cache network in the US and much of the UK has been dismantled. Mostly at the public request. Because it supported privacy and they didn't want anyone to have any. Yes, that was a public decision.
Google. Just a glorified Harvest/Glimpse setup on cheap commodity hardware, using a version of the networked querying mechanism. You could, of course, have used Harvest/Glimpse yourself. Built your own personal search engine. The source was open. I did. At that time, I was setting up a university department's network. Harvest searched all the commonly-used sites, Glimpse then cached responses from the most frequently-used. Nice setup. Everyone used the setup as the proxy, so we had decent performance, strong personal privacy and no reliance on outside search engines. But the hostility towards privacy and the popularity of Yahoo!'s directory/portal and AltaVista made it redundant. Pity.
No, the problem isn't with any of these. The problem is the user. Always has been.
1. Almost. I can scale SMP up to 16, if I have expensive enough chipsets. I can scale a Transputer network to the thousands without any chipset at all. And with each processor having a decent amount of RAM - if we scale, it would be the equivalent of all CPUs having a gigabyte of on-chip RAM, bit more than the L2 on my setup - and with the system capable of farming the processes a-la MOSIX across the nodes, bus memory has limited value. Good for RDMA, though.
2. The idea of PIM is that you don't have to allocate bus channels and switch their direction, you don't need to run things past the MMU, you don't have the latency of a PCI switch. You don't want a general-purpose processor in memory, but it's a long way to ferry data if all you want is to right-shift or left-shift a memory address. I'd rather see the networking stacks in a card that could DMA independently of the OS, since the primary object of PIM is to take out context switching and the OS.
3. The idea of CAM is that you don't have the OS or the CPU involved. It's a limited-use scenario, but there are times when you want a key-value database and you want it to run very very fast in memory. ASICs can do a little bit of it, but they still have to go in and out of RAM and that's slow. An obvious market for this would be in network gear. A switch or a router where you didn't need software to search the tables, where the RAM did it for you using the value you had as a key, would have much lower latency than either a software solution (such as IPTables/Netfilter) or an ASIC solution (that still has to fetch each value from RAM).
4. Yes, they do, and they did a reasonably decent job of it. It was one reason Mac OS lasted as long as it did. However, these days, we can improve on the concept a bit. We have pretty muscular GPUs that are certainly beefy enough to put the entire client side of the operation on. One of the reasons Windows rules the desktop is that they have decent polygon-fill. Why they're using polyfilla as a GUI beats me, but there you go. Anyways, it occurs to me that if any OS is to displace Windows, it has to be far, far better at this. And, frankly, it's a waste having the CPU render text as pixels and pipe vast amounts of data over the bus when it could be sending Metapost/Asymptote data + Postscript to the GPUs at a fraction of the effort.
5. Yes, but the idea was that it was mandatory. Everyone used IPSec, all the time. It would have made the network neutrality debate null and void, as ISPs can't classify what they can't see. We, as a community, handed a lot of power to the Evil Ones.
6. Yeah, but we're still air-cooling them. Full immersion sub-zero cooling has been discussed (complete with HOWTO guides) since the late 90s. And supercomputer companies used naked silicon in such environments before then. There's presumably a way to build the case so that the gold connections remain protected but the silicon is directly exposed, without creating backwaters. As long as you don't use a mineral oil or fluorinert substitute that doesn't react with copper, you should be fine. And having 16 terabytes of RAM on a single 3" wafer in a 3.5" cartridge is going to involve shorter distances and fewer intermediate chips than scattering them across a bunch of memory sticks.
Like I said, professional photographers tend to prefer film because - they say - that when you blow the image up, you need 75 megapixels to compare with any decent-grain film. My own experience of scanning film is that even 1960s film stock only really starts to show grain when I scan it at 12000x12000x24 on an Epson, which is 144 megapixels. I can't afford a digital camera of that res!
If I take multiple b/w photos, using colour filters, I can get near-uniform colour response over as wide or as narrow a gap as I like. B/W digital cameras do not offer uniform response (and often have filters you have to physically remove if you want to get a decent image at the edges of the visible spectrum) and just the colour chip for near-uniform response at decen
Tell Comcast its services are no longer required and Vermont will follow the Chattanooga model and supply everyone 10 gigabits to the home. After all, if it's that expensive, Vermont can't do it for less and therefore can't compete with Comcast, right? But if Vermont can do it, why waste time on a bunch of losers?
We'd have privacy by now but consumers wanted IPv4 rather than IPv6. We lost Mandatory IPSec across the Internet as a result.
We'd have privacy, but American voters chose not to have EU privacy laws and EU data protection laws, thus defiling the globe.
We'd have privacy, but users chose AOL.
We'd have privacy, but users chose Microsoft's email clients over ones that supported PGP/GPG.
We'd have privacy, but users chose Microsoft. (Windows 95 stored passwords in plain text. And users felt this was much safer than the encrypted stuff Linux and BSD were using.)
Sorry, I have no sympathy for a society that feels deprived of privacy when they have actively chosen to throw it away.
The Transputer was a computer on a chip with four networking ports. You built clusters by linking one pin on one to one pin on another. That's it. You could have external memory to bring it up to whatever capacity you wanted. It ran a high-level language - Occam - at almost instruction-set level (your compiler was really an assembler). A modern version running at 3 GHz, with FPU, with multiple cores on each chip, would be incredibly powerful. No need for expensive SMP chips to run distinct CPUs, everything's on a local bus, your PC would be a lot cheaper and a lot more compact. Your smartphone would also be running at a decent speed. USB would be running at the same speed as PCI Express.
2. Processor In Memory
This is basically the Transputer turned inside-out. Instead of having your main memory on the CPU, have part of the CPU inside the memory. Reduced latency, increased performance, reduced chip count. Seymour Cray's ambition was to have MPI built into RAM. A glance at CiteseerX shows other efforts have tried to put the BLAST genetic search system into RAM. Not sure on the latter, but there are obvious benefits to putting very standard libraries there. I'd probably look at the Hoard malloc replacement (an obvious thing for RAM to take care of) and maybe something like the Oil library - very common functions that need to be very fast and everyone gets wrong.
3. Content Addressable Memory
There have been attempts to have RAM chips that could act as databases, where instead of giving a location, you gave it a key field and it would retrieve the contents regardless of where in RAM it was. CAM would be incredibly useful as an add-on to modern computers, NoSQL on a chip.
4. Postscript As A GUI
There was an attempt to build an X11 alternative, and then an X11 WM, around Postscript. If you're wanting to do vectors rather than pixels, it is a much better way to go. If you are wanting a WM for wordprocessing rather than web surfing or games, why pay the huge overhead involved in the current approach? Computers should always be about empowering choice.
5. True Mobile IP
When IPv6 was first developed, the early protocol (and so the early stacks) implemented a form of Mobile IP. This form allowed you to move from one network to another and remain connected to things. You temporarily had two IP addresses and upstream routers NATted the old one to the new one. (Which means IPv6 supported NAT, for those curious about such things.)
This was intended for car-to-car networks (which constantly shift topology), networks on trains or aircraft (since the vehicle changes hotspot) and other contexts that we've now had to invent thousands of new wheels to handle (poorly) because the technology was removed. It was removed not out of privacy concerns (we now know we were all being spied on anyway, and this might have actually increased privacy by destroying the associations we now know they were using) but because Microsoft lobbied against anything that might hurt their sales.
6. Wafer-Scale Integratrion
It is possible to place maybe 512 chips on a single wafer and disable the ones that don't work (as per Sir Clive Sinclair's idea for WSI in the 80s). That's a lot of chips. And, now we know how to cheaply make large quantities of ultra-pure Si-28, a lot of chips with a very low failure rate. You don't need to imagine a Beowulf Cluster of these, they ARE a Beowulf Cluster! A supercomputer not much larger than a DVD. Obviously, the Transputer idea would combine well with this. Or you can design it as Flash and put 11.1-channel 24-bit audio, UHDTV video onto it, have half your movie collection at a quality you can barely imagine or use on one cartridge.
7. Big, Properly-Sprung Keyboards
I hate touchscreens, I loathe the cheap plastic toys they use for computers and I totally despise laptop keyboards. Give me a well-spaced, big, keyboard where keys go thunk and Mean It. Something robust. Something that can handle my typing s
Most of those are designed by people who have never learned how to design UIs. Human-Computer Interface courses are available and I'd gladly run one for the GNOME team if I thought they'd pay attention.
Absolutely agreed.On the other hand, Gnome+Unity is so utterly messed up beyond redemption, maybe something that'll kill off defective thinking would be helpful in the long run.
In politics, you have to throw the occasional bone. No such improved security measure exists, Wyden knows this, but the stupider members of Congress don't. The stupider members are likely to side with the FBI, but now they are likely to wonder how vulnerable their own stuff is and whether the FBI can't just wait a year or two. In politics, that's as close to a win as you can get.
Based on the FBI claims, boasts, entrapment "stings" to boost catch numbers, etc, I have to conclude that the FBI is manned entirely from security staff of the B Ark.
If there are two independent keys that decrypt the message equally, then there are N independent keys that decrypt the message equally, where in many cases N approaches all the possible decryption keys. The pin is not something you compare X with, so simply having a second isn't simply comparing X with two values rather than one. The pin is a part (Byzantine Encryption) or all of the decryption key. Having two means your algorithm has got to have f(x, k) and f(x, y) produce identical values. For most functions, that means making k and y irrelevant. Which means your encryption isn't just weak, it doesn't exist.
Given how many keys would be generated if people continued using IPSec or TLS, it's obviously impossible to catalog them all along with the timestamps for their validity. That means everyone must have identical keys. That means exposure by one person exposes not just that one thing but absolutely everything.
Worse, because all messages would have to be encrypted with the same key, a Bletchley Park-style attack would quickly figure out what they key is. If you know the structure and enough of the plaintext content, then the remaining 1% should be easy to crack.
First, the keys tend to be on the devices and not central servers. That means you need a program on the devices that can transfer keys to arbitrary computers (since every jurisdiction has a different server and servers change their IP addresses from time to time). That means all you need to know is the packet used to request a key for one device and you can fashion the same packet and send it to any device. The key will then be sent back along that channel to the sender.
Second, router poisoning means that stuff destined for a proper server can take a shortcut. Router poisoning is a very common form of attack because most routers don't validate that messages come from proper routers. Even when they do, BGP4+ defines a minimal method of authentication using shared secrets.
Third, let's imagine they use names rather than IP addresses. Almost nobody uses DNSSec (because it's crap) and DNS is very easy to poison. Again, all your top secret keys go to Joe "Just Doin My Job" Mafia-IT-Dude Bloggs.
Fourth, I thought Sony and the recent scandal whereby ALL antivirus vendors sent US government information to Russia had kind of disabused people of the idea that intelligent people were running corporate databases.
Fifth, the DoD was bloody broken into by viruses on thumb drives that copied information across airwalls. Then there were Manning and Snowden - not criminals, but obviously accessing highly classified information that nobody had bothered to secure. The Chinese got hold of the current generation of stealth fighter plans because the DoD dumped hard drives without wiping them. And you trust the government not to do anything stupid. Sorry to disappoint, but you couldn't find stupider if you hunted through the WalMart greeters. These are not people I'd trust with the time of day.
Sixth, in order to have every key, they'd have to disable SSL and TLS. Completely. As these use ad-hoc keys. In fact, they'd have to get rid of IPSec as well. Every connection you made would have to use the same key. This would make international banking interesting. Particularly for fans of Bletchley Park. Perhaps you don't understand how cryptography works. When you use the same key for everything, it doesn't.
It wasn't Clipper that did it, it was Clippy. Microsoft's electronic assistant. "I see you are trying to open a goat." There was only so much of it a Satanist could take.
I wrote to him when Congress was planning on banning software-defined radios. I was impressed by his reply. It was intelligent. It was readable. It showed he'd actually thought about the issue and the technical points.
An intelligent person in Congress is a rarity. That he keeps being re-elected is astonishing - by the nature of a democracy, it is unfair to hold a higher opinion of voters than of those voted in. After the Redneck Trials in Portland last year, when criminals were set free on the dubious grounds that being white proved their innocence, my opinion of those in the city took a dive. Mind you, it was already low. I'd seen 5th street at night and shootings were becoming common near Powell's bookstore.
Anyways, Wyden is maybe no genius but he is technologically astute and deserves to be listened to. Which means the FBI won't.
Well, in a sense, it is. By downgrading the definition of broadband to two tin cans and a piece of string, he has ensured every American has access to it.
Chattanooga needs to be able to sell Internet to whomever they damn well please and any other town in the US should have the absolute authority to do exactly the same. If it is truly we the people, then all people have the right to compete.
But the speed is about half and the cost more than double that of either South Korea or Sweden. Anyone who has run cable knows a cable running machine can do quite a bit more than someone shinning up a mountain (and speeds in the Swedish countryside can reach 40 gigabits per second).
So the Value For Money is a quarter that of rival technological nations. Well, when you're nickel-and-diming your infrastructure, the odd quarter should be expected.
Slashdot Mirror
Other alternatives:
1. Some alien civilizations found wireless to be slow and inefficient, and have adopted dense mode optic fibres, waveguides and other physical networking for most of their stuff. The same route we're taking. You only need enough for the density of radio transmitters in the galaxy to be very low. We can still detect them, using the method proposed by James Lovelock when he was at NASA - observe the atmosphere instead and look for combinations that can't arise naturally.
2. Alien civilizations have an abundance of less than one technological race in every 25 light-years (which is the current extreme range we can detect them). We won't be able to go past even the 100 light-year range until 2024, when SKA comes online. In other words, we've wildly overestimated our abilities.
3. We're pointing telescopes at systems without habitable planets. Now that we actually have the beginnings of a list of exoplanets, let's look and see how many of those which are viable we've actually looked at. Probably very few.
4. Sir Fred Hoyle was right, life rarely evolves on planets, it evolves in deep space. In that case, we're looking in the wrong places. The galaxy could be buzzing with life and we're ignoring it.
5. Encrypted messages are indistinguishable from a random oracle. They're all using Intergalactic IPSec because PRIVACY MATTERS to any evolved species.
As Plato pointed out, ignorance is expensive and dangerous to society. Society chooses to elect leaders who underfund and undercut education. The leaders didn't get there on their own. Ultimately, we are all responsible for all the consequences of our actions, however far down the road they are, particularly in the case of a perfect storm. And privacy has been a perfect storm.
We've known this for decades. Public Squid servers were pilloried. Public anonymous remailers were seized and burned at the stake, with public approval.
ISPs are dangerous to your privacy and physically destroy the networks of rivals, so what did Tennessee do? It passed a Constitutional Amendment banning Chattanooga's municipal system from expanding, after failing to get it outlawed entirely. And the folks of Tennessee? They have no problem defending the loss of their privacy at all and actively support those who despoiled it.
Online advertising cannot personalize your ads if you're using a Squid proxy for everything (which we did back then, because it was faster than going direct, and the JANET backbone supplied Squid at just about every university).
Social media. Ummm, what do you call Slashdot? Anyways, social media is not a particular privacy concern, for the same reason online ads aren't. They can't track you if you use Squid. But the cache network in the US and much of the UK has been dismantled. Mostly at the public request. Because it supported privacy and they didn't want anyone to have any. Yes, that was a public decision.
Google. Just a glorified Harvest/Glimpse setup on cheap commodity hardware, using a version of the networked querying mechanism. You could, of course, have used Harvest/Glimpse yourself. Built your own personal search engine. The source was open. I did. At that time, I was setting up a university department's network. Harvest searched all the commonly-used sites, Glimpse then cached responses from the most frequently-used. Nice setup. Everyone used the setup as the proxy, so we had decent performance, strong personal privacy and no reliance on outside search engines. But the hostility towards privacy and the popularity of Yahoo!'s directory/portal and AltaVista made it redundant. Pity.
No, the problem isn't with any of these. The problem is the user. Always has been.
Originally, it was mandatory in use. All IPv6 connections were encrypted. If you didn't supply the key for a connection, then it defaulted to ad-hoc.
1. Almost. I can scale SMP up to 16, if I have expensive enough chipsets. I can scale a Transputer network to the thousands without any chipset at all. And with each processor having a decent amount of RAM - if we scale, it would be the equivalent of all CPUs having a gigabyte of on-chip RAM, bit more than the L2 on my setup - and with the system capable of farming the processes a-la MOSIX across the nodes, bus memory has limited value. Good for RDMA, though.
2. The idea of PIM is that you don't have to allocate bus channels and switch their direction, you don't need to run things past the MMU, you don't have the latency of a PCI switch. You don't want a general-purpose processor in memory, but it's a long way to ferry data if all you want is to right-shift or left-shift a memory address. I'd rather see the networking stacks in a card that could DMA independently of the OS, since the primary object of PIM is to take out context switching and the OS.
3. The idea of CAM is that you don't have the OS or the CPU involved. It's a limited-use scenario, but there are times when you want a key-value database and you want it to run very very fast in memory. ASICs can do a little bit of it, but they still have to go in and out of RAM and that's slow. An obvious market for this would be in network gear. A switch or a router where you didn't need software to search the tables, where the RAM did it for you using the value you had as a key, would have much lower latency than either a software solution (such as IPTables/Netfilter) or an ASIC solution (that still has to fetch each value from RAM).
4. Yes, they do, and they did a reasonably decent job of it. It was one reason Mac OS lasted as long as it did. However, these days, we can improve on the concept a bit. We have pretty muscular GPUs that are certainly beefy enough to put the entire client side of the operation on. One of the reasons Windows rules the desktop is that they have decent polygon-fill. Why they're using polyfilla as a GUI beats me, but there you go. Anyways, it occurs to me that if any OS is to displace Windows, it has to be far, far better at this. And, frankly, it's a waste having the CPU render text as pixels and pipe vast amounts of data over the bus when it could be sending Metapost/Asymptote data + Postscript to the GPUs at a fraction of the effort.
5. Yes, but the idea was that it was mandatory. Everyone used IPSec, all the time. It would have made the network neutrality debate null and void, as ISPs can't classify what they can't see. We, as a community, handed a lot of power to the Evil Ones.
6. Yeah, but we're still air-cooling them. Full immersion sub-zero cooling has been discussed (complete with HOWTO guides) since the late 90s. And supercomputer companies used naked silicon in such environments before then. There's presumably a way to build the case so that the gold connections remain protected but the silicon is directly exposed, without creating backwaters. As long as you don't use a mineral oil or fluorinert substitute that doesn't react with copper, you should be fine. And having 16 terabytes of RAM on a single 3" wafer in a 3.5" cartridge is going to involve shorter distances and fewer intermediate chips than scattering them across a bunch of memory sticks.
Like I said, professional photographers tend to prefer film because - they say - that when you blow the image up, you need 75 megapixels to compare with any decent-grain film. My own experience of scanning film is that even 1960s film stock only really starts to show grain when I scan it at 12000x12000x24 on an Epson, which is 144 megapixels. I can't afford a digital camera of that res!
If I take multiple b/w photos, using colour filters, I can get near-uniform colour response over as wide or as narrow a gap as I like. B/W digital cameras do not offer uniform response (and often have filters you have to physically remove if you want to get a decent image at the edges of the visible spectrum) and just the colour chip for near-uniform response at decen
https://arstechnica.com/tech-p...
I'm not sure we want Comcast to "deal" with competition. They seem to have a very... "don't call me scarface" approach to it.
Tell Comcast its services are no longer required and Vermont will follow the Chattanooga model and supply everyone 10 gigabits to the home. After all, if it's that expensive, Vermont can't do it for less and therefore can't compete with Comcast, right? But if Vermont can do it, why waste time on a bunch of losers?
Nothing wrong with government regulation when it's logical and rational, as is the case 95% of the time in Scandanavia.
If you choose Republicans (and the Democrats have just been moderate Republicans for a long time), expect stupid.
We'd have privacy by now but consumers wanted IPv4 rather than IPv6. We lost Mandatory IPSec across the Internet as a result.
We'd have privacy, but American voters chose not to have EU privacy laws and EU data protection laws, thus defiling the globe.
We'd have privacy, but users chose AOL.
We'd have privacy, but users chose Microsoft's email clients over ones that supported PGP/GPG.
We'd have privacy, but users chose Microsoft. (Windows 95 stored passwords in plain text. And users felt this was much safer than the encrypted stuff Linux and BSD were using.)
Sorry, I have no sympathy for a society that feels deprived of privacy when they have actively chosen to throw it away.
Abandoned Systems:
1. Transputer
The Transputer was a computer on a chip with four networking ports. You built clusters by linking one pin on one to one pin on another. That's it. You could have external memory to bring it up to whatever capacity you wanted. It ran a high-level language - Occam - at almost instruction-set level (your compiler was really an assembler). A modern version running at 3 GHz, with FPU, with multiple cores on each chip, would be incredibly powerful. No need for expensive SMP chips to run distinct CPUs, everything's on a local bus, your PC would be a lot cheaper and a lot more compact. Your smartphone would also be running at a decent speed. USB would be running at the same speed as PCI Express.
2. Processor In Memory
This is basically the Transputer turned inside-out. Instead of having your main memory on the CPU, have part of the CPU inside the memory. Reduced latency, increased performance, reduced chip count. Seymour Cray's ambition was to have MPI built into RAM. A glance at CiteseerX shows other efforts have tried to put the BLAST genetic search system into RAM. Not sure on the latter, but there are obvious benefits to putting very standard libraries there. I'd probably look at the Hoard malloc replacement (an obvious thing for RAM to take care of) and maybe something like the Oil library - very common functions that need to be very fast and everyone gets wrong.
3. Content Addressable Memory
There have been attempts to have RAM chips that could act as databases, where instead of giving a location, you gave it a key field and it would retrieve the contents regardless of where in RAM it was. CAM would be incredibly useful as an add-on to modern computers, NoSQL on a chip.
4. Postscript As A GUI
There was an attempt to build an X11 alternative, and then an X11 WM, around Postscript. If you're wanting to do vectors rather than pixels, it is a much better way to go. If you are wanting a WM for wordprocessing rather than web surfing or games, why pay the huge overhead involved in the current approach? Computers should always be about empowering choice.
5. True Mobile IP
When IPv6 was first developed, the early protocol (and so the early stacks) implemented a form of Mobile IP. This form allowed you to move from one network to another and remain connected to things. You temporarily had two IP addresses and upstream routers NATted the old one to the new one. (Which means IPv6 supported NAT, for those curious about such things.)
This was intended for car-to-car networks (which constantly shift topology), networks on trains or aircraft (since the vehicle changes hotspot) and other contexts that we've now had to invent thousands of new wheels to handle (poorly) because the technology was removed. It was removed not out of privacy concerns (we now know we were all being spied on anyway, and this might have actually increased privacy by destroying the associations we now know they were using) but because Microsoft lobbied against anything that might hurt their sales.
6. Wafer-Scale Integratrion
It is possible to place maybe 512 chips on a single wafer and disable the ones that don't work (as per Sir Clive Sinclair's idea for WSI in the 80s). That's a lot of chips. And, now we know how to cheaply make large quantities of ultra-pure Si-28, a lot of chips with a very low failure rate. You don't need to imagine a Beowulf Cluster of these, they ARE a Beowulf Cluster! A supercomputer not much larger than a DVD. Obviously, the Transputer idea would combine well with this. Or you can design it as Flash and put 11.1-channel 24-bit audio, UHDTV video onto it, have half your movie collection at a quality you can barely imagine or use on one cartridge.
7. Big, Properly-Sprung Keyboards
I hate touchscreens, I loathe the cheap plastic toys they use for computers and I totally despise laptop keyboards. Give me a well-spaced, big, keyboard where keys go thunk and Mean It. Something robust. Something that can handle my typing s
Most of those are designed by people who have never learned how to design UIs. Human-Computer Interface courses are available and I'd gladly run one for the GNOME team if I thought they'd pay attention.
Absolutely agreed.On the other hand, Gnome+Unity is so utterly messed up beyond redemption, maybe something that'll kill off defective thinking would be helpful in the long run.
On the one hand, maximum function in minimum real-estate is a good idea.
On the other hand, GNOME has become an ungodly mess and Linux' reputation for stability and speed has greatly suffered.
Get GNOME to conform to NASA's Power of Ten rules and then let's talk.
In politics, you have to throw the occasional bone. No such improved security measure exists, Wyden knows this, but the stupider members of Congress don't. The stupider members are likely to side with the FBI, but now they are likely to wonder how vulnerable their own stuff is and whether the FBI can't just wait a year or two. In politics, that's as close to a win as you can get.
Based on the FBI claims, boasts, entrapment "stings" to boost catch numbers, etc, I have to conclude that the FBI is manned entirely from security staff of the B Ark.
If there are two independent keys that decrypt the message equally, then there are N independent keys that decrypt the message equally, where in many cases N approaches all the possible decryption keys. The pin is not something you compare X with, so simply having a second isn't simply comparing X with two values rather than one. The pin is a part (Byzantine Encryption) or all of the decryption key. Having two means your algorithm has got to have f(x, k) and f(x, y) produce identical values. For most functions, that means making k and y irrelevant. Which means your encryption isn't just weak, it doesn't exist.
Given how many keys would be generated if people continued using IPSec or TLS, it's obviously impossible to catalog them all along with the timestamps for their validity. That means everyone must have identical keys. That means exposure by one person exposes not just that one thing but absolutely everything.
Worse, because all messages would have to be encrypted with the same key, a Bletchley Park-style attack would quickly figure out what they key is. If you know the structure and enough of the plaintext content, then the remaining 1% should be easy to crack.
First, the keys tend to be on the devices and not central servers. That means you need a program on the devices that can transfer keys to arbitrary computers (since every jurisdiction has a different server and servers change their IP addresses from time to time). That means all you need to know is the packet used to request a key for one device and you can fashion the same packet and send it to any device. The key will then be sent back along that channel to the sender.
Second, router poisoning means that stuff destined for a proper server can take a shortcut. Router poisoning is a very common form of attack because most routers don't validate that messages come from proper routers. Even when they do, BGP4+ defines a minimal method of authentication using shared secrets.
Third, let's imagine they use names rather than IP addresses. Almost nobody uses DNSSec (because it's crap) and DNS is very easy to poison. Again, all your top secret keys go to Joe "Just Doin My Job" Mafia-IT-Dude Bloggs.
Fourth, I thought Sony and the recent scandal whereby ALL antivirus vendors sent US government information to Russia had kind of disabused people of the idea that intelligent people were running corporate databases.
Fifth, the DoD was bloody broken into by viruses on thumb drives that copied information across airwalls. Then there were Manning and Snowden - not criminals, but obviously accessing highly classified information that nobody had bothered to secure. The Chinese got hold of the current generation of stealth fighter plans because the DoD dumped hard drives without wiping them. And you trust the government not to do anything stupid. Sorry to disappoint, but you couldn't find stupider if you hunted through the WalMart greeters. These are not people I'd trust with the time of day.
Sixth, in order to have every key, they'd have to disable SSL and TLS. Completely. As these use ad-hoc keys. In fact, they'd have to get rid of IPSec as well. Every connection you made would have to use the same key. This would make international banking interesting. Particularly for fans of Bletchley Park. Perhaps you don't understand how cryptography works. When you use the same key for everything, it doesn't.
I can't wait for these questions to appear on University Challenge.
It wasn't Clipper that did it, it was Clippy. Microsoft's electronic assistant. "I see you are trying to open a goat." There was only so much of it a Satanist could take.
I wrote to him when Congress was planning on banning software-defined radios. I was impressed by his reply. It was intelligent. It was readable. It showed he'd actually thought about the issue and the technical points.
An intelligent person in Congress is a rarity. That he keeps being re-elected is astonishing - by the nature of a democracy, it is unfair to hold a higher opinion of voters than of those voted in. After the Redneck Trials in Portland last year, when criminals were set free on the dubious grounds that being white proved their innocence, my opinion of those in the city took a dive. Mind you, it was already low. I'd seen 5th street at night and shootings were becoming common near Powell's bookstore.
Anyways, Wyden is maybe no genius but he is technologically astute and deserves to be listened to. Which means the FBI won't.
Well, in a sense, it is. By downgrading the definition of broadband to two tin cans and a piece of string, he has ensured every American has access to it.
That requires US State Law to change.
Chattanooga needs to be able to sell Internet to whomever they damn well please and any other town in the US should have the absolute authority to do exactly the same. If it is truly we the people, then all people have the right to compete.
https://www.numbeo.com/cost-of...
US is given as the 9th worst for cost out of the 90 countries measured.
https://www.forbes.com/sites/n...
Forbes measures 196 countries and puts the US at 114th, so the 82nd worst in the world.
But what do you get for this?
https://en.wikipedia.org/wiki/...
Wikipedia says it's the tenth fastest... out of a list of ten.
https://www.statista.com/stati...
Statistica says the same.
https://www.fastmetrics.com/in...
Fastmetrics gives a shade of puke.
https://www.forbes.com/sites/k...
Forbes says ninth.
But the speed is about half and the cost more than double that of either South Korea or Sweden. Anyone who has run cable knows a cable running machine can do quite a bit more than someone shinning up a mountain (and speeds in the Swedish countryside can reach 40 gigabits per second).
So the Value For Money is a quarter that of rival technological nations. Well, when you're nickel-and-diming your infrastructure, the odd quarter should be expected.
No, Title II was never repealed.