Heh, yes, I know what you mean! This one's tricky, and I agree with you in principle: If you want to play video on Windows, just use DirectShow (and on Linux, standardize on... something. Gstreamer?). But in practice, for some reason, people have a much easier time with players like MPlayer and VLC which contain all their own codecs (or in the case of MPlayer have their own modular codecs separate from e.g., on Windows, DirectShow's), and by contrast have all sorts of difficulties with DirectShow codec packs like Klite. More generally, people talk about 'DLL hell' on Windows and 'dependency hell' on Linux and elsewhere. So, somewhere, it seems that modularity starts to fail and you get more reliability by making individual pieces of software as self-sufficient as possible -- even if this results in a bunch of code duplication. Arguably, this is what one does by running Firefox instead of Internet Explorer or Konqueror in the first place.
Also--doesn't this reverse 40 years of feminism by enforcing the idea that to get ahead "with business and social lives," woman must wear "fuck me pumps?"
There are many feminisms... One of them, so-called 'lipstick feminism' (and the stronger 'slut feminism') actually seeks to embrace things like high heels as symbols of female power. I don't like this particular variety of feminism (I think it is actively harmful) but it exists, and was relatively popular when I was in college (mainly, cynical-me thinks, because it provided an excuse to sleep around. Personally I don't think I would feel empowered by herpes, but, hey, if you want to ignore public health statistics to satisfy your ideology go ahead...).
(In fact, there are so many feminisms that statements like "I am a feminist" seem pretty meaningless to me... but I digress even more...)
It lacks ability to do pointers and even structs (C-style structs that is) making any kind of advanced data structures cumbersome to implement.
JScript avoids these issues and feels much nicer to me (just personal preference). The downside is that you'll find more documentation for VBScript than JScript.
Yes.
This incident has a lot of visibility, and the government can not afford to let it go with a slap.
If what happened to Exxon after Valdez is any indication, then there will be an initial, very large and very public fine, which they will eventually find a way to avoid paying. See here. In short: They were told just after the disaster to pay $2.5 billion, but years later the Supreme Court reduced that number to just $500 million.
Yeah, I'm basically with you. He was good in Enemy of the State, but in too many other of his movies (e.g., I, Robot) he just plays... Will Smith. So if you need a good guy with a swagger, I guess you can typecast him, but that's not really acting...
Unfortunately, Gibson was pretty explicit about what hacking looked like. It looked like riding a white shark.
You can't not make that look ridiculous to a year-2010 audience.
But then, I'm in that minority who thinks that William Gibson is extremely overrated as a novelist. I read Neuromancer at the recommendation of a friend in college; it was just-OK. I read Pattern Recognition too, and remember being annoyed by the feeling that Gibson was one of those fools who tries way too hard to be hip without knowing what the hell is actually going on -- like these people. And his collaboration with Bruce Sterling on The Difference Engine was just disappointing.
Anything they do, Stephenson does better. Except the Baroque Cycle. My god is that tedious.
They could go with Ellen McClain instead, but the mood in the theater would probably be wrecked by the sound of "THE CAKE IS A LIE!" shouted from the back rows...
Really though, it is possible to connect to existing P2P networks through NAT without any extra configuration. Why should a P2P social network suddenly make the exist solutions to the problem infeasible or more difficult to support?
Don't these protocols actually rely on central servers (or lists of servers, or "SuperNodes," or trackers in the case of BitTorrent) known to everyone behind NAT? AFAIK every NAT hole-punching scheme relies on an intermediary, and I can't imagine how it could be otherwise.
Maybe this isn't as big a problem as I'm making it out to be, but in the end NAT does mean at the very least that the scheme can't be totally decentralized.
This wasn't a user page though; it was literally "facebook.com/companyName," and the company was actually a big one -- something like "Verizon" or the like. I sensed it was less a "we're too poor to have a website" move and more a "all the cool kids are on facebook so we should be there" move.
A big problem is simply NAT. Non-technical people are not going to set up port forwarding. This basically broke the Internet, and pushed its development in undemocratic directions.
UPnP partially fixes this, but opens up a whole bunch of other problems, which are even worse.
IPv6 is supposed to fix this for real, but I don't count on it because IPv4 is "good enough," and I bet that it'll be easier for people to keep throwing NAT and subdomains at the problem. E.g., companies don't need to bother maintaining their own webservers and having their own public IPs; the way things are going they'll just point people to "facebook.com/companyName" (I heard an ad do this on the radio yesterday, in fact).
building an expensive and permanently burdensome infrastructure that will likely be obsolete by the time it's in place.
I think that's a totally legit objection. At the very least, it's an argument for flexible, well-thought-out systems.
There is no particular reason self-contained systems in the car cannot accomplish this goal
I disagree with this however. The entire idea is to use nonlocal information about how many cars are queued in various places, and this is information that cars cannot reliably measure for themselves.
You mean by narrowing down their choices to really only what you want?
The idea behind correlated equilibrium is that you don't reduce the set of choices that players can make at all. All you do is also give them an instruction about which of those choices they should make, that they are free to ignore.
For instance, consider two cars stopped at an intersection, and suppose that each has two possible actions: GO and WAIT. If both cars WAIT, nobody goes anywhere and each gets a payoff of 0. If both cars GO, they collide and get a payoff of -10; this is something they each want to avoid. If one car GOes and the other WAITs, then the GOer gets a payoff of +2 and the WAITer gets a payoff of +1. Question: What happens?
Rather than answering this here, I'll just say,
1 - Adding a stoplight that tells players whether to GO or WAIT (which they are free to ignore) actually improves the situation dramatically for both players. The reason is that both players know that the other is receiving instructions, and they know the (probabilistic) rule by which the instructions are given out.
2 - Wikipedia, as usual, has a pretty good explanation; rather than calling the actions GO and WAIT they call them DARE and CHICKEN (and use slightly different numbers), but the game is essentially the same.
But maybe you understood this and were making a more profound point, that by being the third, "correlating" party, you can control what the others do, because it is in their interest to listen to you?
Why not just put the timer on the traffic lights themselves then and allow the human to manually decide what happens? Either fully automate our personal motorcoaches or gtfo.
I think your idea of putting a visible countdown timer on lights might not be a bad idea. However, I also don't see an automated system as taking unnecessary control from the driver any more than I see using an automatic transmission that way. It's just a way to reduce cognitive load, and presumably have the computer make decisions more efficiently than the human could. So long as you can disable the system when you need to, and the engine automatically restarts when you manually hit the gas (a la cruise control and the brake), I think the driver has retained enough control, no?
The problem is the damn headline, that makes the idea sound Orwellian. It isn't. It's not about disabling your engine, or some other DRM-style idea. It's about giving your car additional information that it can choose to use to increase fuel efficiency.
Are you familiar with the idea of correlated equilibrium from game theory? By giving players a common instruction, which they can choose either to follow or to disobey, you can often get better Nash equilibria than if you simply made the players decide what to do independently. That's what this is -- applied to engine management.
DES is sufficiently weak that it is possible to build a home-grown cluster that can break a DES key in minutes. Yes, DES is "strong" in the sense that the algorithm itself has no significant flaws that anyone can detect, but [...] the cost of smashing DES would be 0.1% of the money the criminals could walk off with. In short, as close to nothing as to make no odds.
I think your "consider the cost in dollars" point of view is very insightful, and DES is indeed not very strong, but I still don't think it's fair to compare it to a substitution cipher. It takes ~$200k of hardware and 24 hours to crack DES. It takes a retiree five minutes and the back of a napkin (or a laptop less than a second) to break a substitution cipher. Organized criminals can break DES, but Joe Average can't without taking out a mortgage just to buy the hardware (or having access to a pretty extensive botnet).
If the keyspace can be reduced further, though, then things get easier. E.g., one program I investigated that sent DES-encrypted passwords over the network ignored the least significant bit of every character in a password, only allowed 7 or 8 character passwords, and was IIRC case-insensitive; if you additionally assumed that most passwords would be alphanumeric you ended up with a keyspace that could be bruteforced in a few days by no more than half a dozen workstations in a university computer lab.
The practical difference between a Caesar cipher and DES is that the Caesar cipher is faster so more transactions can be performed.
Umm, no?
Caesar, and substitution-based ciphers in general, are so easy to break that they're given as puzzles in the daily newspapers (some aphorism is encrypted with a substitution cipher; you need to figure out what it is). Basically, you know the frequencies with which various letters, pairs of letters, etc. appear in English text; so you can compute the frequencies with which the characters in the ciphertext appear and determine a correspondence that way.
DES, on the other hand, mangles the statistics pretty thoroughly. There are attacks, but AFAIK they're not obvious, and still pretty computationally heavy.
Slashdot Mirror
Heh, yes, I know what you mean! This one's tricky, and I agree with you in principle: If you want to play video on Windows, just use DirectShow (and on Linux, standardize on... something. Gstreamer?). But in practice, for some reason, people have a much easier time with players like MPlayer and VLC which contain all their own codecs (or in the case of MPlayer have their own modular codecs separate from e.g., on Windows, DirectShow's), and by contrast have all sorts of difficulties with DirectShow codec packs like Klite. More generally, people talk about 'DLL hell' on Windows and 'dependency hell' on Linux and elsewhere. So, somewhere, it seems that modularity starts to fail and you get more reliability by making individual pieces of software as self-sufficient as possible -- even if this results in a bunch of code duplication. Arguably, this is what one does by running Firefox instead of Internet Explorer or Konqueror in the first place.
Am I the only one who gets "The connection was reset" 9 out of 10 times I try to go to abstrusegoose.com?
1 - Is this literally true?
2 - What does it mean?
OK tan
Yay, skin cancer!
Also--doesn't this reverse 40 years of feminism by enforcing the idea that to get ahead "with business and social lives," woman must wear "fuck me pumps?"
There are many feminisms... One of them, so-called 'lipstick feminism' (and the stronger 'slut feminism') actually seeks to embrace things like high heels as symbols of female power. I don't like this particular variety of feminism (I think it is actively harmful) but it exists, and was relatively popular when I was in college (mainly, cynical-me thinks, because it provided an excuse to sleep around. Personally I don't think I would feel empowered by herpes, but, hey, if you want to ignore public health statistics to satisfy your ideology go ahead...).
(In fact, there are so many feminisms that statements like "I am a feminist" seem pretty meaningless to me... but I digress even more...)
+5 Insightful? Really? Really?
It lacks ability to do pointers and even structs (C-style structs that is) making any kind of advanced data structures cumbersome to implement.
JScript avoids these issues and feels much nicer to me (just personal preference). The downside is that you'll find more documentation for VBScript than JScript.
Yes. This incident has a lot of visibility, and the government can not afford to let it go with a slap.
If what happened to Exxon after Valdez is any indication, then there will be an initial, very large and very public fine, which they will eventually find a way to avoid paying. See here. In short: They were told just after the disaster to pay $2.5 billion, but years later the Supreme Court reduced that number to just $500 million.
Yeah, I'm basically with you. He was good in Enemy of the State, but in too many other of his movies (e.g., I, Robot) he just plays... Will Smith. So if you need a good guy with a swagger, I guess you can typecast him, but that's not really acting...
Unfortunately, Gibson was pretty explicit about what hacking looked like. It looked like riding a white shark.
You can't not make that look ridiculous to a year-2010 audience.
But then, I'm in that minority who thinks that William Gibson is extremely overrated as a novelist. I read Neuromancer at the recommendation of a friend in college; it was just-OK. I read Pattern Recognition too, and remember being annoyed by the feeling that Gibson was one of those fools who tries way too hard to be hip without knowing what the hell is actually going on -- like these people. And his collaboration with Bruce Sterling on The Difference Engine was just disappointing.
Anything they do, Stephenson does better. Except the Baroque Cycle. My god is that tedious.
They could go with Ellen McClain instead, but the mood in the theater would probably be wrecked by the sound of "THE CAKE IS A LIE!" shouted from the back rows...
I haven't seen a premise that [...] unoriginal, in years...
<facebook>LIKE</facebook>
Why is it so hard to figure out that if you can't figure it out you shouldn't agree to it?
The same could be said of software EULAS, mobile phone contracts, or half a dozen other things...
Really though, it is possible to connect to existing P2P networks through NAT without any extra configuration. Why should a P2P social network suddenly make the exist solutions to the problem infeasible or more difficult to support?
Don't these protocols actually rely on central servers (or lists of servers, or "SuperNodes," or trackers in the case of BitTorrent) known to everyone behind NAT? AFAIK every NAT hole-punching scheme relies on an intermediary, and I can't imagine how it could be otherwise.
Maybe this isn't as big a problem as I'm making it out to be, but in the end NAT does mean at the very least that the scheme can't be totally decentralized.
This wasn't a user page though; it was literally "facebook.com/companyName," and the company was actually a big one -- something like "Verizon" or the like. I sensed it was less a "we're too poor to have a website" move and more a "all the cool kids are on facebook so we should be there" move.
A big problem is simply NAT. Non-technical people are not going to set up port forwarding. This basically broke the Internet, and pushed its development in undemocratic directions.
UPnP partially fixes this, but opens up a whole bunch of other problems, which are even worse.
IPv6 is supposed to fix this for real, but I don't count on it because IPv4 is "good enough," and I bet that it'll be easier for people to keep throwing NAT and subdomains at the problem. E.g., companies don't need to bother maintaining their own webservers and having their own public IPs; the way things are going they'll just point people to "facebook.com/companyName" (I heard an ad do this on the radio yesterday, in fact).
building an expensive and permanently burdensome infrastructure that will likely be obsolete by the time it's in place.
I think that's a totally legit objection. At the very least, it's an argument for flexible, well-thought-out systems.
There is no particular reason self-contained systems in the car cannot accomplish this goal
I disagree with this however. The entire idea is to use nonlocal information about how many cars are queued in various places, and this is information that cars cannot reliably measure for themselves.
You mean by narrowing down their choices to really only what you want?
The idea behind correlated equilibrium is that you don't reduce the set of choices that players can make at all. All you do is also give them an instruction about which of those choices they should make, that they are free to ignore.
For instance, consider two cars stopped at an intersection, and suppose that each has two possible actions: GO and WAIT. If both cars WAIT, nobody goes anywhere and each gets a payoff of 0. If both cars GO, they collide and get a payoff of -10; this is something they each want to avoid. If one car GOes and the other WAITs, then the GOer gets a payoff of +2 and the WAITer gets a payoff of +1. Question: What happens?
Rather than answering this here, I'll just say,
1 - Adding a stoplight that tells players whether to GO or WAIT (which they are free to ignore) actually improves the situation dramatically for both players. The reason is that both players know that the other is receiving instructions, and they know the (probabilistic) rule by which the instructions are given out.
2 - Wikipedia, as usual, has a pretty good explanation; rather than calling the actions GO and WAIT they call them DARE and CHICKEN (and use slightly different numbers), but the game is essentially the same.
But maybe you understood this and were making a more profound point, that by being the third, "correlating" party, you can control what the others do, because it is in their interest to listen to you?
Why not just put the timer on the traffic lights themselves then and allow the human to manually decide what happens? Either fully automate our personal motorcoaches or gtfo.
I think your idea of putting a visible countdown timer on lights might not be a bad idea. However, I also don't see an automated system as taking unnecessary control from the driver any more than I see using an automatic transmission that way. It's just a way to reduce cognitive load, and presumably have the computer make decisions more efficiently than the human could. So long as you can disable the system when you need to, and the engine automatically restarts when you manually hit the gas (a la cruise control and the brake), I think the driver has retained enough control, no?
Totally agree.
The problem is the damn headline, that makes the idea sound Orwellian. It isn't. It's not about disabling your engine, or some other DRM-style idea. It's about giving your car additional information that it can choose to use to increase fuel efficiency.
Are you familiar with the idea of correlated equilibrium from game theory? By giving players a common instruction, which they can choose either to follow or to disobey, you can often get better Nash equilibria than if you simply made the players decide what to do independently. That's what this is -- applied to engine management.
For prior art we can claim a paper I did in elementary school about 30 years ago, if anyone can find it.
You wrote about networking traffic lights and cars, and applying queuing theory to increase fuel efficiency? Really? In elementary school?
"Flamebait?" Really?
DES is sufficiently weak that it is possible to build a home-grown cluster that can break a DES key in minutes. Yes, DES is "strong" in the sense that the algorithm itself has no significant flaws that anyone can detect, but [...] the cost of smashing DES would be 0.1% of the money the criminals could walk off with. In short, as close to nothing as to make no odds.
I think your "consider the cost in dollars" point of view is very insightful, and DES is indeed not very strong, but I still don't think it's fair to compare it to a substitution cipher. It takes ~$200k of hardware and 24 hours to crack DES. It takes a retiree five minutes and the back of a napkin (or a laptop less than a second) to break a substitution cipher. Organized criminals can break DES, but Joe Average can't without taking out a mortgage just to buy the hardware (or having access to a pretty extensive botnet).
If the keyspace can be reduced further, though, then things get easier. E.g., one program I investigated that sent DES-encrypted passwords over the network ignored the least significant bit of every character in a password, only allowed 7 or 8 character passwords, and was IIRC case-insensitive; if you additionally assumed that most passwords would be alphanumeric you ended up with a keyspace that could be bruteforced in a few days by no more than half a dozen workstations in a university computer lab.
The practical difference between a Caesar cipher and DES is that the Caesar cipher is faster so more transactions can be performed.
Umm, no?
Caesar, and substitution-based ciphers in general, are so easy to break that they're given as puzzles in the daily newspapers (some aphorism is encrypted with a substitution cipher; you need to figure out what it is). Basically, you know the frequencies with which various letters, pairs of letters, etc. appear in English text; so you can compute the frequencies with which the characters in the ciphertext appear and determine a correspondence that way.
DES, on the other hand, mangles the statistics pretty thoroughly. There are attacks, but AFAIK they're not obvious, and still pretty computationally heavy.