"Pairing" does work (on my newly purchased Toyota Prius), but does have some small quirks (such as the A2DP audio issue described in another post, or lack of access to text messages).
Another annoying little thing: Once you've paired a phone with the car (because you want to use the hands-free (HFP) features), then the music (from CD, radio or USB stick) goes quiet a minute or so after getting into the car, because it automatically switches to the phone's A2DP output, "just in case".
A2DP should be selected only if explicitly selected by the user!
Why not do it the same way as for PC motherboard BIOS upgrades? Have the hardware recognize whether any given update is compatible with it, and refuse the upgrade if not.
Then, on the download site, users can chose between Toyota Prius Model year 2008-2010, or 2012-2013 for example.
Simple no?
Apparently not, because Toyota is not even able to make sure that the owner's manual matches the installed navigator unit...
The SN is public. So a person telling his SN to a third party has no way of proving to that third party that he didn't just pick it from the list.
However, you would probably be getting some kind of written receipt with the number on... care should be taken that this should be trivial to forge, or the "official" receipt might be proof enough...
Actually, as somebody else here has pointed out, they did submit one single dodgy paper to a reputable journal as well, and it got accepted! (meaning 100% acceptance rate in "reputable" journal, versus only 51% in open-access...) So the study's conclusion should not really be what it looks like at first glance... Ok, admitted, the sample size of control group is way too small, but that's needed to make the paper dodgy.
... but in any case, it's an interesting twist of the liar's paradox... I say, a twist, rather than an example, because no control group at all is even worse than one with a sample size of 1.
I'm convinced that if the right buzzwords and writing style is used, a sizeable percentage of so called "reputable" journals will fall into the same trap!
There's plenty of other agencies around who are interested in such silly concerns, unfortunately. And they do have the power and willingness to wreck people's life over such trivialities.
You need the Certificate Patrol plugin, which warns you when a site's certificate changes unexpectedly, even when the new certificate has a "valid" signature.
Unfortunately, this doesn't work with Google's servers, who rotate among a gazillion certificates "legitimately", and thus drown the user in false positives. But given Google's cooperation with Prism, maybe this effect is wanted?
People also forget that SSL is wholly dependent on Certificate Authorities
Well, technically, you could always very "certificate" fingerprints manually...
The problem is that data anyone that owns the network can conduct an MITM attack against your key.
Make that "... anyone that owns the network and the CA can conduct an attack...". The purpose of SSL is exactly to prevent attacks by people who "only" control the network between client and server.
SSL is fundamentally broken because it presents a perception of trust when it is incapable of providing that level of trust.
SSL doesn't supply trust, instead it relies on trust. Namely on the trust that CA's are doing their job properly (... which unfortunately, they don't always do...)
Not when you hold the same keys the real CA does. The NSA may well have their own copies of these keys.
The CA doesn't hold any private keys, at least not usually. Even the Mossad allows you to skip giving away your private key.
So, all a malicious CA can do is issue a second certificate with the same info, but for a different private/public key pair. But that means that the fingerprint will be different (this is a hash over the entire certificate, including the public key, which won't match the public key of the original).
So, an observing user can indeed spot this. Only the browser's automatic check (based solely on the CA's signature) will be fooled by this.
You don't even need to be actually flirting. Just keeping pictures of nice ladies on your computer can be enough. Or just helpfully repairing the computer of a friend who happens to keep such pictures is enough.
If the NSA want to feel like idiots, they're free to do so.
A similar thing happened to a friend in Germany. And not, the German police didn't feel like idiots, and quite happily wrecked the guys life. If you have a gun, you never feel like an idiot. Instead you just pull the trigger on anybody who dares to snicker...
Do you really think "mysecretdomain.com" certificate from shitty ass low cost certificate provider doesn't have a duplicate key on file at Comodo, Network Solutions, GoDaddy or TwoCows or whatever?
Only if you gave it to them. Which a competent webmaster would not do. But it's still a mystery to my why some certification authorities such as StartSSL attempt to ask for this (but, to their credit, you may skip...)
They don't have to brute force or hack anything if they have an appliance in the middle that automatically grabs the certificate from the certificate issuer and
The certificate alone is worthless to an eavesdropper without a matching private key.
spoofs both sides of the connection.
If an eavesdropper spoofed both sides, the client would notice that it is not speaking to the real server, because the spoofer doesn't have access to the proper public/private key pair. And he cannot just use another public/private pair because then the signature wouldn't match. Unless the CA cooperated by certifying another public key, but then this might become clear evidence of spoofing if the user was paranoid enough to manually compare.
If you want your traffic encrypted, you need to generate your own certificates using software you compiled after you reviewed the code.
The standard operating mode is indeed to generate your own CSR (which will only contains your public key for the certificate, but never you private key). You then give this CSR to your CA, which signs it, which makes it a certificate. The CA never sees your private key.
However, what this thread is about is that a subverted CA may help an eavesdropper by issuing him an extra certificate matching the victim's domain, but the eavesdropper's private key. This would evade standard checks in browsers (who blindly trust their CA's signature), but would still be obvious to a manual check (fingerprints would be different). So, an alert user might spot this, and save the fake certificate as evidence of such tampering. So far, no such certificate has been posted, which leads us to believe that either the NSA don't have done this yet, or only have done it very seldomly, against unsophisticated users.
In events where CAs were compromised by crackers (DigiNotar), such fake certificates did surface pretty quickly, proving that enough users actually do care to do such manual checks.
Good point about Assange. But there's probably multiple copies encoded in such a way that no single person can trigger early publishing, which means that we're probably looking at about a dozen people having some. Do you really think there's a dozen persons out of the reach of NSA, and trusted enough by Snowden?
If they do, do you think it would be used for the people? No, it would be used as leverage to further their own agenda by pulling the congress critter's strings./quote.
Not that far-fetched, after all, it happened that way in Luxembourg...
Also, those "life insurance" documents are certainly with people trusted by Snowden.
Which means that they must be very close friends, with whom he certainly interacted a lot. Which means that with a little bit of detective work, the NSA would have not trouble finding out who they are.
Slashdot Mirror
"Pairing" does work (on my newly purchased Toyota Prius), but does have some small quirks (such as the A2DP audio issue described in another post, or lack of access to text messages).
A2DP should be selected only if explicitly selected by the user!
Then, on the download site, users can chose between Toyota Prius Model year 2008-2010, or 2012-2013 for example.
Simple no?
Apparently not, because Toyota is not even able to make sure that the owner's manual matches the installed navigator unit...
So make that "get them back in one piece".
Even if it's your employer? That might put your manager behind bars, but yourself out of a job...
However, you would probably be getting some kind of written receipt with the number on... care should be taken that this should be trivial to forge, or the "official" receipt might be proof enough...
Their private key would allow them to prove to a third party how they voted... making them subject to pressure or bribery.
I'm convinced that if the right buzzwords and writing style is used, a sizeable percentage of so called "reputable" journals will fall into the same trap!
There's plenty of other agencies around who are interested in such silly concerns, unfortunately. And they do have the power and willingness to wreck people's life over such trivialities.
However, I somewhat doubt that the SWAT team would go after those police who don't like to feel like idiots...
Maybe I need to upgrade my browser.
You need the Certificate Patrol plugin, which warns you when a site's certificate changes unexpectedly, even when the new certificate has a "valid" signature.
Unfortunately, this doesn't work with Google's servers, who rotate among a gazillion certificates "legitimately", and thus drown the user in false positives. But given Google's cooperation with Prism, maybe this effect is wanted?
People also forget that SSL is wholly dependent on Certificate Authorities
Well, technically, you could always very "certificate" fingerprints manually...
The problem is that data anyone that owns the network can conduct an MITM attack against your key.
Make that "... anyone that owns the network and the CA can conduct an attack...". The purpose of SSL is exactly to prevent attacks by people who "only" control the network between client and server.
SSL is fundamentally broken because it presents a perception of trust when it is incapable of providing that level of trust.
SSL doesn't supply trust, instead it relies on trust. Namely on the trust that CA's are doing their job properly (... which unfortunately, they don't always do...)
Not when you hold the same keys the real CA does. The NSA may well have their own copies of these keys.
The CA doesn't hold any private keys, at least not usually. Even the Mossad allows you to skip giving away your private key.
So, all a malicious CA can do is issue a second certificate with the same info, but for a different private/public key pair. But that means that the fingerprint will be different (this is a hash over the entire certificate, including the public key, which won't match the public key of the original).
So, an observing user can indeed spot this. Only the browser's automatic check (based solely on the CA's signature) will be fooled by this.
You don't even need to be actually flirting. Just keeping pictures of nice ladies on your computer can be enough. Or just helpfully repairing the computer of a friend who happens to keep such pictures is enough.
If the NSA want to feel like idiots, they're free to do so.
A similar thing happened to a friend in Germany. And not, the German police didn't feel like idiots, and quite happily wrecked the guys life. If you have a gun, you never feel like an idiot. Instead you just pull the trigger on anybody who dares to snicker...
Do you really think "mysecretdomain.com" certificate from shitty ass low cost certificate provider doesn't have a duplicate key on file at Comodo, Network Solutions, GoDaddy or TwoCows or whatever?
Only if you gave it to them. Which a competent webmaster would not do. But it's still a mystery to my why some certification authorities such as StartSSL attempt to ask for this (but, to their credit, you may skip...)
They don't have to brute force or hack anything if they have an appliance in the middle that automatically grabs the certificate from the certificate issuer and
The certificate alone is worthless to an eavesdropper without a matching private key.
spoofs both sides of the connection.
If an eavesdropper spoofed both sides, the client would notice that it is not speaking to the real server, because the spoofer doesn't have access to the proper public/private key pair. And he cannot just use another public/private pair because then the signature wouldn't match. Unless the CA cooperated by certifying another public key, but then this might become clear evidence of spoofing if the user was paranoid enough to manually compare.
If you want your traffic encrypted, you need to generate your own certificates using software you compiled after you reviewed the code.
The standard operating mode is indeed to generate your own CSR (which will only contains your public key for the certificate, but never you private key). You then give this CSR to your CA, which signs it, which makes it a certificate. The CA never sees your private key.
However, what this thread is about is that a subverted CA may help an eavesdropper by issuing him an extra certificate matching the victim's domain, but the eavesdropper's private key. This would evade standard checks in browsers (who blindly trust their CA's signature), but would still be obvious to a manual check (fingerprints would be different). So, an alert user might spot this, and save the fake certificate as evidence of such tampering. So far, no such certificate has been posted, which leads us to believe that either the NSA don't have done this yet, or only have done it very seldomly, against unsophisticated users.
In events where CAs were compromised by crackers (DigiNotar), such fake certificates did surface pretty quickly, proving that enough users actually do care to do such manual checks.
Some airlines name their planes after cities, castles or other landmarks.
, but they don't refuse to go on missions because their plane might get scratched.
Yes, that's why it is called a plane and not a flying car...
Except that charter flight passengers (or even "general aviation") need to go through checkpoints too...
Don't forget, gmail.com is part of Prism!
But certainly, it stopped working after you threw it out of the window...
Good point about Assange. But there's probably multiple copies encoded in such a way that no single person can trigger early publishing, which means that we're probably looking at about a dozen people having some. Do you really think there's a dozen persons out of the reach of NSA, and trusted enough by Snowden?
If they do, do you think it would be used for the people? No, it would be used as leverage to further their own agenda by pulling the congress critter's strings./quote. Not that far-fetched, after all, it happened that way in Luxembourg...
Which means that they must be very close friends, with whom he certainly interacted a lot. Which means that with a little bit of detective work, the NSA would have not trouble finding out who they are.