It may be a common definition, but that doesn't make it a correct one.

What has feel to do with it? You are describing the user interface.

It may or may not be, but I think if you step outside the small world of personal computers and into the wider world that computer science covers you will discover its a very clearly and exactly defined concept.

"For example, take multiple recordings (each at 2*Nyquist) at different time-shifts and reconstruct a higher-res signal."

And that differs from a higher sampling rate in what way?

The important fact you are missing is real world signal to noise ratio. No source has zero noise, so below a certain signal level there is no signal only noise, so as long as the bit depth is sufficient to cover the available S/N ratio, and as long as the sampling frequency is high enough to cover the frequency range of interest, and in the case of audio thats well defined, and we could up the limit a few times to be sure. So as long as thats all met, and the equipment is working as it should, then yes, loss-less recording of the world is entirely possibly.

While its conceivable that some labels did that in the past, I think you are describing something that just doesn't happen anymore. So its a bit of a vacuous argument to make now.

Oh, and BTW, I think EMI did just fine from Pink Floyd, that was EMI before they sold themself to venture capitalists.

Unless, as in the case of my car, its fitted with run flat tyres and the car does not have space in the boot for a spare.

No good, A Hogshead is already defined as 54 imperial gallons

http://en.wikipedia.org/wiki/Hogshead

And could you be so kind as to provide the "definition of a mathematical formula" which makes your statement true please?

Well, ignoring the fact that it can be proved through the normal Turing machine proof. It can be also be shown in practice that it can be reduced to a set of logic equations. Use VHDL to describe a processor and memory and any i/o required. Add the VHDL to describe the stored program containing the instructions to run on the processor to implement the algorithm. Then synthesise the resultant VHDL to a FPGA target, and you have reduced the algorithm and the machine to implement it to a set of pure matematical equations of the sort that even a lawyer could recognise as mathematics.

May be a age thing, it reminded me of the Tek vector graphics terminal I used to use. (and of course asteroids)