I challenge you to wipe your Windows HD, re-install everything from just a windows CD and see how much of that hardware works.
That's a straw man - if a piece of hardware requires extra drivers for Windows, it comes with a driver CD. If you lose the CD you can easily download and install the drivers without ever seeing a command prompt, let alone recompiling your kernel or googling to find out where your distro stores its firmware images or a thousand other annoying little jobs. I love Linux and I use it exclusively, but I'm not blind to its weaknesses, and hardware support (especially for wireless cards, as the GP pointed out) is one of those weaknesses.
I imagine that the current research is on even better codes that wouldn't require as much feedback to work well.
Yup, rateless codes (e.g. LT, Raptor and online codes) produce an endless stream of output symbols, any sufficiently large subset of which can be used to reconstruct the input.
Not everybody who stores or carries the data will need to display it. If the data's in Sanskrit, nobody needs to have a Sanskrit font except the person entering the data and the person reading it, but everyone else can still cleanse, fold and manipulate it because it's in Unicode.
I was thinking the same thing so I've just been looking into it. Up to version 6, a Flash movie could establish a network connection to any machine in the same base domain as the machine from which it was downloaded - eg movies from www.foo.com could connect to downloads.foo.com. But starting with version 7, Flash has the same restrictions as Java: a Flash movie can only establish a network connection to the exact machine it was downloaded from. However, it's possible to circumvent this by installing a policy file on the machine you want to connect to. Flash will download the policy file, and if the policy allows the domain from which the movie was downloaded to connect, it will make the connection. (Yes, this is thoroughly broken because it relies on the client to enforce the security policy.)
That still leaves the problem of how to accept incoming connections - Flash doesn't support server sockets. TCP has a rarely used feature called simultaneous open that allows two clients to establish a connection if their SYN packets cross on the wire. This can also be used for NAT and firewall traversal, and the NUTTS group at Cornell has achieved an 85% success rate by combining TCP simultaneous open with port prediction. So it looks like BitTorrent in Flash might be feasible after all, if a little hacky.
My guess is that they've implemented BitTorrent inside the Flash movie player.
"But Flash XMLSockets can only make outgoing connections to servers in the same domain that supplied the movie," I hear you cry.
"Ah," I respond sagely, "who determines what is inside thevideobay.org's domain if not thevideobay.org itself?"
"So they're assigning single-use dynamic DNS names to all their clients, but that doesn't solve the outgoing connection limitation," you protest.
"Hush little one," I reply with irritating condescension, "have you never heard of TCP simultaneous open with port prediction? It has an 85% success rate through domestic NATs and doesn't require raw sockets."
Look closer. We aren't exactly sending in the B-52s to airdrop loads of McMuffins, LOTR DVDs, sneakers, and twinkies onto the Noble Primitive Peoples who are Honoring the Sacred Traditions of Their Ancestors. It's a pull situation much more than a push. Western culture, simply put, is addictive.
Try telling a Third World farmer who can't compete with subsidised US grain prices that you're not bombing them with Twinkies. The military, economic and cultural influences of the West are not easily separated. Western companies strongly promote their products in non-Western markets, with the economic and if necessary military backing of Western governments. "The national interest" includes business interests, and the national interest is defended and extended by force - look at what happened to Mossadegh and Allende when they expropriated the assets of Western companies.
Helping poor people in a foreign country is "aid". Helping poor people in your own country is "socialism". One is acceptable to middle-class voters, the other is not.
1. Single point of failure. Everyone's information is on one database, and lots of people need access to (parts of) that database.
2. Ubiquity. When there's one standard ID card, it's a lot easier to demand that everyone carries it - "you must show Real ID to enter this stadium" is easier to enforce than "you must show one of this long list of acceptable forms of ID, or two of this even longer list of marginally acceptable forms of ID, to enter this stadium". That means more opportunistic data collection, a greater sense of being under surveillance, and a harder life for people who prefer not to present an ID card every time they spend money or enter a building.
3. Network effects. Just as in the UK, the database is the goal, the card is just a distraction. If identity databases didn't make it easier to monitor people, why would governments be prepared to spend so much money on them? They've seen the kind of data mining and profiling that's possible with commercial databases, and they know they can do better because they can demand more information. But a database of 2 million people is more valuable than two databases of 1 million people each - data mining and profiling don't just work on individuals, they work on groups, but only if everyone's on the same database.
This discovery is just the tip of the iceberg. I shouldn't be telling you this, but the Chinese have actually built a full-scale replica of the entire United States in a remote corner of Hunan Province. The model is complete down to the smallest detail - it even includes a replica of you, sitting and reading this article right now. But for some reason the replica of you is dressed in stiff nylon clothes cut in a vaguely 1950s style. In fact the whole place has the feel of a faded Polaroid photograph; the children's smiles are slightly too wide, the food looks slightly too waxy, and of course everyone speaks slightly out of sync...
Findability is interesting, but speaking as a paranoid basket case I'm more interested in unfindability. It strikes me that there's a widening gap between findable and unfindable information, and the defining feature of that gap is names. Whether they're URLs or sets of googleable keywords, names have become vastly more useful - and more powerful - in the last ten years. If you know the name of something, it's now a trivial task to find information about it. But if you don't know the name of the concept you're searching for, you're stuck with 20th century methods of searching.
Here's an example: I came up with a statistical tool that's a simple modification of a standard statistical tool, and I wouldn't be surprised if someone has thought of it before. But if they did, I have no idea what they called it. So to find out whether it already exists I have to resort to 20th century search tools: experts and libraries. I have to find someone who's well-versed enough in the area to have come across the idea if it exists, but not so well-versed that my assinine beginner's question will just annoy them. I have to borrow textbooks from the university library and page through them looking for a description that matches my idea (or points me to another book where I might find out more).
These are both valuable exercises that can lead to unexpected and useful discoveries, and when I call them "20th century search tools" I don't mean it in some condescending Web 2.0 way, but there's an undeniable and growing difference between those methods of searching and the methods I would have used if I knew the name of the concept.
done properly, other people might be able to locate an object with a particular identifier, but not know that it is the keys to your car
If they could watch an object leaving my house, getting into my car, driving to work, going to my desk, going out for lunch, returning to work and later returning home, I wouldn't particularly care if they knew whether they were tracking my car keys or my socks.
Don't worry, the location of your keys will be protected by a state of the art security system. To find out where you left your car keys you'll have to enter a 64 character pass phrase and swipe your RFID-enabled driver's license. To find out where you left your driver's license you'll have to enter a 128 character pass phrase, submit to a retina scan, present an X.509 certificate signed by both Bruce Schneier and Phil Zimmerman, and swipe your RFID-enabled car keys.
Everyone accused of being a terrorist is a terrorist.
Good example of this yesterday: the headline of the daily rag was TERROR SUSPECTS FREE IN BRITAIN. My reaction was "of course they're free, that's why they're called 'suspects' rather than 'convicts'". They might as well have said EVERYONE INNOCENT UNTIL PROVEN GUILTY. But if most people had the same reaction it wouldn't sell papers...
Here's an academic paper by the designers of the system described in the article. Unfortunately the paper's only available to journal subscribers, but someone seems to have published it on Freenet, or you can find a preprint version here. From the paper:
The minimum energy required to capture
CO2 from the air at a partial pressure of 4×10^-4 atm and deliver it at one atmosphere
is therefore about 20 kJ/mol or 1.6 GJ/tC (gigajoules per ton carbon). If we add
the energy required for compressing the CO2 to the 100 atm pressure required
for geological storage (assuming a 50% efficiency for converting primary energy
to compressor work) the overall energy requirement for air capture with geologic
sequestration is about 4 GJ/tC.
The 4 GJ/tC minimum may be compared to the carbon-specific energy content
of fossil fuels: coal, oil, and natural gas have about 40, 50, and 70 GJ/tC respectively.
Thus if the energy for air capture is provided by fossil fuels then the amount of
carbon captured from the air can in principle be much larger than the carbon
content of the fuel used to capture it. The fuel carbon can, of course, be captured
as part of the process rather than being emitted to the air.
Digging huge holes and filling them with corn stalks would be rather energy-intensive. Burning the corn stalks or letting them rot, then capturing the carbon from the air and pumping it underground, is probably a more practical solution.
Ironically, one of the biggest markets for CO2 is oil extraction: you pump CO2 into a dying well to force out the last of the oil. (Air is unsafe for obvious reasons.) Afterwards you leave the CO2 underground in the same chambers that previously held the oil, so you get sequestration for free. From the press release:
For example, the CO2 originating from all those vehicles in Bangkok can be captured in an oil field in Alberta, Canada, where it could be used on-site for enhanced oil recovery (EOR) operations or it could be captured in South Africa to feed a growing demand in that country for feed stocks for petrochemical production. If the goal is to sequester a given quantity of CO2 in a specific geological formation, the air capture system could be located at that physical location.
Slashdot Mirror
Just right-click on the Recycle Bin and select Explore...
Yup, rateless codes (e.g. LT, Raptor and online codes) produce an endless stream of output symbols, any sufficiently large subset of which can be used to reconstruct the input.
Not everybody who stores or carries the data will need to display it. If the data's in Sanskrit, nobody needs to have a Sanskrit font except the person entering the data and the person reading it, but everyone else can still cleanse, fold and manipulate it because it's in Unicode.
That still leaves the problem of how to accept incoming connections - Flash doesn't support server sockets. TCP has a rarely used feature called simultaneous open that allows two clients to establish a connection if their SYN packets cross on the wire. This can also be used for NAT and firewall traversal, and the NUTTS group at Cornell has achieved an 85% success rate by combining TCP simultaneous open with port prediction. So it looks like BitTorrent in Flash might be feasible after all, if a little hacky.
My guess is that they've implemented BitTorrent inside the Flash movie player.
"But Flash XMLSockets can only make outgoing connections to servers in the same domain that supplied the movie," I hear you cry.
"Ah," I respond sagely, "who determines what is inside thevideobay.org's domain if not thevideobay.org itself?"
"So they're assigning single-use dynamic DNS names to all their clients, but that doesn't solve the outgoing connection limitation," you protest.
"Hush little one," I reply with irritating condescension, "have you never heard of TCP simultaneous open with port prediction? It has an 85% success rate through domestic NATs and doesn't require raw sockets."
Try telling a Third World farmer who can't compete with subsidised US grain prices that you're not bombing them with Twinkies. The military, economic and cultural influences of the West are not easily separated. Western companies strongly promote their products in non-Western markets, with the economic and if necessary military backing of Western governments. "The national interest" includes business interests, and the national interest is defended and extended by force - look at what happened to Mossadegh and Allende when they expropriated the assets of Western companies.
Yeah, but can you filibuster to prevent $SOFTWARE_MONOPOLY from releasing bad code?
Oh yeah? Prove it.
Helping poor people in a foreign country is "aid". Helping poor people in your own country is "socialism". One is acceptable to middle-class voters, the other is not.
The CanSecWest presentation that started all this is available here.
1. Single point of failure. Everyone's information is on one database, and lots of people need access to (parts of) that database.
2. Ubiquity. When there's one standard ID card, it's a lot easier to demand that everyone carries it - "you must show Real ID to enter this stadium" is easier to enforce than "you must show one of this long list of acceptable forms of ID, or two of this even longer list of marginally acceptable forms of ID, to enter this stadium". That means more opportunistic data collection, a greater sense of being under surveillance, and a harder life for people who prefer not to present an ID card every time they spend money or enter a building.
3. Network effects. Just as in the UK, the database is the goal, the card is just a distraction. If identity databases didn't make it easier to monitor people, why would governments be prepared to spend so much money on them? They've seen the kind of data mining and profiling that's possible with commercial databases, and they know they can do better because they can demand more information. But a database of 2 million people is more valuable than two databases of 1 million people each - data mining and profiling don't just work on individuals, they work on groups, but only if everyone's on the same database.
This discovery is just the tip of the iceberg. I shouldn't be telling you this, but the Chinese have actually built a full-scale replica of the entire United States in a remote corner of Hunan Province. The model is complete down to the smallest detail - it even includes a replica of you, sitting and reading this article right now. But for some reason the replica of you is dressed in stiff nylon clothes cut in a vaguely 1950s style. In fact the whole place has the feel of a faded Polaroid photograph; the children's smiles are slightly too wide, the food looks slightly too waxy, and of course everyone speaks slightly out of sync...
Here's an example: I came up with a statistical tool that's a simple modification of a standard statistical tool, and I wouldn't be surprised if someone has thought of it before. But if they did, I have no idea what they called it. So to find out whether it already exists I have to resort to 20th century search tools: experts and libraries. I have to find someone who's well-versed enough in the area to have come across the idea if it exists, but not so well-versed that my assinine beginner's question will just annoy them. I have to borrow textbooks from the university library and page through them looking for a description that matches my idea (or points me to another book where I might find out more).
These are both valuable exercises that can lead to unexpected and useful discoveries, and when I call them "20th century search tools" I don't mean it in some condescending Web 2.0 way, but there's an undeniable and growing difference between those methods of searching and the methods I would have used if I knew the name of the concept.
If they could watch an object leaving my house, getting into my car, driving to work, going to my desk, going out for lunch, returning to work and later returning home, I wouldn't particularly care if they knew whether they were tracking my car keys or my socks.
Don't worry, the location of your keys will be protected by a state of the art security system. To find out where you left your car keys you'll have to enter a 64 character pass phrase and swipe your RFID-enabled driver's license. To find out where you left your driver's license you'll have to enter a 128 character pass phrase, submit to a retina scan, present an X.509 certificate signed by both Bruce Schneier and Phil Zimmerman, and swipe your RFID-enabled car keys.
Whereas any sufficiently recent non-Microsoft OS contains an ad hoc, informally-specified, bug-ridden, slow implementation of 95% of Unix. ;-)
I agree with you about Foucault's Pendulum, but don't judge him solely by that - The Name of the Rose is fantastic (and not a Rosicrucian in sight).
Good example of this yesterday: the headline of the daily rag was TERROR SUSPECTS FREE IN BRITAIN. My reaction was "of course they're free, that's why they're called 'suspects' rather than 'convicts'". They might as well have said EVERYONE INNOCENT UNTIL PROVEN GUILTY. But if most people had the same reaction it wouldn't sell papers...
Incompetence is the charitable answer; there are others.
Damn, I was really hoping the process would in some way involve fracing lasers. ;-)
You're barking up the wrong tree.
Digging huge holes and filling them with corn stalks would be rather energy-intensive. Burning the corn stalks or letting them rot, then capturing the carbon from the air and pumping it underground, is probably a more practical solution.
Ironically, one of the biggest markets for CO2 is oil extraction: you pump CO2 into a dying well to force out the last of the oil. (Air is unsafe for obvious reasons.) Afterwards you leave the CO2 underground in the same chambers that previously held the oil, so you get sequestration for free. From the press release: For example, the CO2 originating from all those vehicles in Bangkok can be captured in an oil field in Alberta, Canada, where it could be used on-site for enhanced oil recovery (EOR) operations or it could be captured in South Africa to feed a growing demand in that country for feed stocks for petrochemical production. If the goal is to sequester a given quantity of CO2 in a specific geological formation, the air capture system could be located at that physical location.