cooky560 wrote:Sorry to burst your bubble here, but SSL is the very technology that makes this protection available.
Think of it like a safe, a safe allows a person to safety protect real world items within the realm of the law, if I had confidential papers about you, but no safe, claiming "the statute" protects you just as the wind blows them all out the window and over the garden for any-one to pick up, you'd be unhappy and the statute has provided no protection whatsoever. SSL is very similar, SSL is a mechanism that provides the privacy needed to conform not only to statute law, but also to create a bond of trust between the customer and the business, it's like an online safe. Further, might I suggest you read the protections the act actually provides, it's largely against fraud rather than secure storage of details, which is what's being discussed here. Data Storage and the usage of things like SSL for what it's worth are more involved with the Data Protection Act, a statute that is well known for being open to abuse.
The reason I referred to the CCA, is that the most common use for SSL/TLS at the moment is online purchasing made with a credit card.
There people seem to believe (or vendors claim) that SSL makes such transactions safe. However, it does not.
Safety comes from ones ability to repudiate the transaction, and the CCA then places obligations upon the card issuer which are generally to the advantage of the holder if there was actual fraud. So basically conducting the transaction over an encrypted link is irrelevant. Further discussion of CCA applicability to this thread is a distraction.
For other uses of SSL/TLS, i.e. the context of this thread; the way browsers generally operate also makes it moot when the adversary is the government.
Especially with CCDP (which is Labour's IMP rehashed). I've not checked the details of CCDP, but what of what was necessary for IMP was actually the installation of middle boxes, i.e. inherently a man-in-the-middle. Just what those middle boxes would do is the subject of some current interest.
However, given that it is quite easy to defeat common browser use of SSL/TLS by use of fake CA certificates and DNS poisoning, any supposed protection against an adversary with greater capabilities (mandating that ISPs install middle boxes) is rather hopeful.
Key negotiation can be an issue but I highly doubt that the govt boxes will perform such an attack, the result would be too damaging to the online economy, which right now is one of the few where the UK is exporting more than it imports.
Easily handled. The mechanism is already there for selective blocking of locations, instead of blocking this can be used to redirect selective traffic to a middle box.
So all that is necessary is to detect such SSL/TLS utilising TCP connections, and shove via a transparent proxy. This can be avoided for certain well known sites
(addressing the above commercial concern), and is made even easier by the fact that most people have NATed connections already.
However, I do agree that it is (still) unlikely here.
But one needs to be aware that SSL/TSL without having previous knowledge of the expected public key is not providing real privacy.
A similar situation pertains with how SSH is commonly used, but at least there is caches the key and complains when it is seen to change.
Simply put, SSL/TLS is not a privacy panacea, it has to be used properly. I would also suggest that use via a proxy is improper.
Finally, on a technical note, SSL/TSL does not operate as you suggest. While it is block oriented, this does not necessarily result in a data expansion.
It is possible for compression to be negotiated along with encryption, such that less data is sent than would be over a plain TCP connection.
Even in the absence of compression, the expansion will at most be a few percent (each block can have up to 16384 bytes, with up to around 20 bytes of overhead).
The wikipedia article gives a good overview, and has references to the appropriate RFCs with all of the details.