WRT backdoors, if the idea is to cripple encryption standards by weakening ciphers, then it's no better than obscuring data. As we've seen demonstrated time and time again, if there's a backdoor, it will be found by people other than those it's intended for and it will be exploited. I'm sure companies transmitting trade secrets and the banking industry to name a few would just love that. So in answer to the point about not caring about gov't access but caring about criminal access - how about both?
And, I really, really don't trust the UK government's competence in designing such a cipher (they're still a good few years off understanding what this thing called the Internet is, it would seem). Neither AES, nor any respected ciphers, have any key escrow capability, so such an addition would require either a massive rework of the algorithm, or a new one entirely. And designing secure ciphers is hard. It doesn't take much of a slip-up to completely ruin the security of a cipher.
Also, considering that in the real world, we need to know the source code of ciphers (and even in a theoretical situation without it, decompiling code or taking photographs of chips achieves the same thing), it would be pretty trivial to reverse engineer it and discover how to exploit the backdoor. The whole concept of a backdoor'd-for-one-group cipher is fundamentally broken. All previous attempts have failed quite spectacularly.
The NSA were smart enough to give up on the idea a long time ago. They also gave up on gimped ciphers like DES when it became apparent that, in order for a gov't to have access, it also destroys the credibility of the cipher for securing data from everyone else. The damage caused by using a fundamentally insecure cipher standard is immeasurable. That's why we now have strong ciphers.
And that's pretty much my original point; how absurd the whole thing is from a technical perspective.
@wasabi: Missed your last point. Oh I agree, I also object on principle. And neither am I against spying when necessary.
https://www.schneier.com/blog/archiv...ty_vs_pri.html