[Paranoia] CD Ripping Uncertainty Principle?

[Monty]

On Tue, Sep 28, 2004 at 09:01:06PM +0200, Nils Chr. Framstad wrote:
> An (obviously) uneducated question from me, then: where does the
> _randomness_ occur? Is it
>  - the bits? E.g. the CD has bits with values so close to 1/2 that
> Heisenberg, God and Kolmogorov (please delete as appropriate)
> sometimes outputs 0, sometimes 1, for any reasonable CD-player?

yes. 

>  - the physics of the spinner? If software requests a re-read, and you
> basically don't know where you bounce into the groove, you get
> different sets of samples at each rip?

No, we're proofed against that.

> What puzzles me also is that cdparanoia does not report the error.

All cdparanoia knows is that something doesn't match; it doesn't try
to classify different kinds of matches to that extent.  It does report
an event, in this case, usually a '+'.

> Is
> it so that all those errors are re-read until they by coincidence pass
> the error corrections, and then cdparanoia is satisfied and cannot
> tell the difference between a correct sample and a erroneous which by
> coincidence looks right on the nth read?

Yes.  Once upon a time it was pointless to try to intuit more from
'unstable' bits; every drive behaved differently and what worked on
one didn't work on another.  Most drives do not tell you if there are
corrected/uncorrected errors in a returned data vector.  Once upon a
time, Plextor was the only brand that would do so, and it was vaguely
pointless to support the feature.  More drives will report that
information today (although many report it incorrectly).

Monty