The reason that crypto can never really be made 100% convenient is Zooko's triangle: you want the name by which you refer to your correspondant to be memorable, globally unique, and free from centralized control, but you can't have all three (see also Clay Shirky's restatement of this idea). So if you want to use email addresses, someone has to be the centralized authority from which is ultimately derived your right to state that you are the legitimate recipient of a certain email.
If we had DNSSEC - if domain authorities routinely certified DNSSEC public keys with the same authority by which they allow name server records to change - then this would mean the central authority was at least doing their job properly and we could use it to build an email infrastructure. But then people wouldn't pay Verisign for certificates, so that would never do.
He's correct about parameterization of cryptographic algorithms. In general, that results in less secure protocols, not more.
The comparison with DJB is way overblown. It's a shame he was so rude about Greg Stein, but I think you overstate greatly the extent of his obnoxiousness.
No - that's not how the forgeability is achieved. Actually there's no magic to it at all - you just use a symmetric authentication primitive instead of an asymmetric one. In other words, you use the same key to verify the signature as you do to generate it. If Alice sends Bob an authenticated message, Bob can't prove to Eve that Alice sent it, because Eve knows that Bob might have just generated it and signed it himself.
In fact, PFS means that even if you record the entire conversation and later also grab Alice and Bob's private keys, you still can't decrypt the conversation.
I agree. I only wish that they'd ended the movie much earlier, in that case; everything after the final reunion of the surviving main characters is pretty pointless without the Scouring.
(1) All the cars in front go to their destinations. The maintenance car travels to the broken down car by going "the wrong way" down the line, and drags the failed car away.
(2) Popular stops need enough stopping positions to handle likely peak need.
(3) Cars wander the lines to where they're most needed.
(4) Yes, there's such a button. But teenagers who press them for fun will get caught...
You might as well count and record the touchscreen count, but treat it like a "super-exit-poll", and the scan as the final result. There will be small discrepancies due to people making ballot cards they don't drop into the ballot box...
No, bits of paper or card are currently the key technology in every way of doing a full, secure election that we know of. In a securely designed system, the paper trail is the true, authorititative source of the election results, and the DRE counts are just an early indicator to satisfy the "must-know-result-immediately" crowd. This is because the voter can verify that the paper count says what they mean it to say, and once it's dropped in the ballot box, reliable low-tech means are sufficient to ensure that no-one can tamper with it.
One good solution prints the voter's intent on each ballot as a barcode and in human-readable form. These would be compared for a large number of ballots, to detect any attempt at hacking the vote by generating "lying" ballots.
People talk about this in a misleading way, it's true. See Open Voting Consortium for a demonstration of how this should actually work - each vote should be recorded on paper, in a human-readable form and as a barcode. Then
(a) votes can be counted quickly and cheaply (b) each stage of the process can be checked by observers using sampling techniques
If we find an efficient way to solve NP-style problems, then conventional keyed encryption (eg AES in EAX mode) falls too, because we can efficiently search for the key that makes the plaintext make sense (and MAC correctly).
However, P?=NP doesn't directly bear on these problems, because they are not organised in families of increasing problem size, such that there are always larger and more difficult problems.
And if there's a way of factoring large numbers whose compute time grows on the fiftieth power on the length of the number, then it wouldn't in practice make any problem for those using RSA, even though it brings the problem into P.
In this context, "advocacy site" is a euphemism for "nakedly partisan site". A site whose purpose is to advocate Linux for those who haven't made a decision is a different thing entirely.
We're never going to disguise the fact that there are a lot of Linux users who fucking hate Microsoft, and we would be wrong to try. Sure, don't use these icons in the slideshow you make for your boss to convince them to switch, but don't insist we all pretend we're still undecided about whether Microsoft are a bunch of bastards who we hate.
Slashdot Mirror
I totally agree. It's just the sort of rotten stunt that Micro$hit would pull.
The reason that crypto can never really be made 100% convenient is Zooko's triangle: you want the name by which you refer to your correspondant to be memorable, globally unique, and free from centralized control, but you can't have all three (see also Clay Shirky's restatement of this idea). So if you want to use email addresses, someone has to be the centralized authority from which is ultimately derived your right to state that you are the legitimate recipient of a certain email.
If we had DNSSEC - if domain authorities routinely certified DNSSEC public keys with the same authority by which they allow name server records to change - then this would mean the central authority was at least doing their job properly and we could use it to build an email infrastructure. But then people wouldn't pay Verisign for certificates, so that would never do.
Yours probably does too - it's a common legal device.
Already done - XORing the source text with itself is a provably perfectly secure form of encryption!
I'd assumed you were greatly exaggerating for dramatic effect, but benchmarks show a range from almost no improvement to a factor of 5.
He's correct about parameterization of cryptographic algorithms. In general, that results in less secure protocols, not more.
The comparison with DJB is way overblown. It's a shame he was so rude about Greg Stein, but I think you overstate greatly the extent of his obnoxiousness.
It would have been even better if they'd put the text of a URL in the image explaining the fraud and what they'd done...
IBM will want costs. And blood.
No - that's not how the forgeability is achieved. Actually there's no magic to it at all - you just use a symmetric authentication primitive instead of an asymmetric one. In other words, you use the same key to verify the signature as you do to generate it. If Alice sends Bob an authenticated message, Bob can't prove to Eve that Alice sent it, because Eve knows that Bob might have just generated it and signed it himself.
In fact, PFS means that even if you record the entire conversation and later also grab Alice and Bob's private keys, you still can't decrypt the conversation.
You mean those flourescent strips are powered by fusion?
I agree. I only wish that they'd ended the movie much earlier, in that case; everything after the final reunion of the surviving main characters is pretty pointless without the Scouring.
Does anyone know if the EE covers all the stuff with "Mr Sharkey" (have I remembered correctly?) and the Shire? Did they even film that stuff?
No, GMT is the same as UTC. You're thinking of UT1.
No, GMT is the same as UTC.
(1) All the cars in front go to their destinations. The maintenance car travels to the broken down car by going "the wrong way" down the line, and drags the failed car away.
(2) Popular stops need enough stopping positions to handle likely peak need.
(3) Cars wander the lines to where they're most needed.
(4) Yes, there's such a button. But teenagers who press them for fun will get caught...
The urine is detected automatically, the doors lock, and they get transported direct to the pokey...
You might as well count and record the touchscreen count, but treat it like a "super-exit-poll", and the scan as the final result. There will be small discrepancies due to people making ballot cards they don't drop into the ballot box...
No, bits of paper or card are currently the key technology in every way of doing a full, secure election that we know of. In a securely designed system, the paper trail is the true, authorititative source of the election results, and the DRE counts are just an early indicator to satisfy the "must-know-result-immediately" crowd. This is because the voter can verify that the paper count says what they mean it to say, and once it's dropped in the ballot box, reliable low-tech means are sufficient to ensure that no-one can tamper with it.
One good solution prints the voter's intent on each ballot as a barcode and in human-readable form. These would be compared for a large number of ballots, to detect any attempt at hacking the vote by generating "lying" ballots.
People talk about this in a misleading way, it's true. See Open Voting Consortium for a demonstration of how this should actually work - each vote should be recorded on paper, in a human-readable form and as a barcode. Then
(a) votes can be counted quickly and cheaply
(b) each stage of the process can be checked by observers using sampling techniques
I'm generally a nice guy and I, ah, get around a lot. Liking sex does not equal being a bad guy.
The OTP is a special case.
If we find an efficient way to solve NP-style problems, then conventional keyed encryption (eg AES in EAX mode) falls too, because we can efficiently search for the key that makes the plaintext make sense (and MAC correctly).
However, P?=NP doesn't directly bear on these problems, because they are not organised in families of increasing problem size, such that there are always larger and more difficult problems.
And if there's a way of factoring large numbers whose compute time grows on the fiftieth power on the length of the number, then it wouldn't in practice make any problem for those using RSA, even though it brings the problem into P.
Some sort of "NAT punch-through"?
l
http://www.mindcontrol.org/~hplus/nat-punch.htm
http://midcom-p2p.sourceforge.net/
The tokens it hands out aren't *that* valuable, so if the machine goes down you don't lose all that much.
Certainly you'd never do something as sad as read /., or post to it.
In this context, "advocacy site" is a euphemism for "nakedly partisan site". A site whose purpose is to advocate Linux for those who haven't made a decision is a different thing entirely.
We're never going to disguise the fact that there are a lot of Linux users who fucking hate Microsoft, and we would be wrong to try. Sure, don't use these icons in the slideshow you make for your boss to convince them to switch, but don't insist we all pretend we're still undecided about whether Microsoft are a bunch of bastards who we hate.