Cory Petkovsek wrote:

> errors?  Errors are correct on a raid system because each of the drives
> contains information about the other drives.  In a holographic system,
> each "atom" or unit (not an elemental atom) contains information about
> the whole.  The more atoms there are, the more accurate information is
> available and the less viable errors, inconsistencies and missing
> information become.

You're almost describing forward error correction (FEC).  FEC adds
redundant parity information to a bitstream so that the entire
original bitstream can be recovered even if some pieces are lost.
Information theory says that if the original bitstream has N bits,
you need at least N bits to reconstruct it perfectly.

As for reducing resolution when pieces are missing, yes that works for
datatypes that can be compressed lossily (e.g., audio, images).  There
is enough redundant information in the original to reconstruct a
reasonable facsimile of some missing pieces.  The problem is that most
things that *can be* lossily compressed *are* lossily compressed, so
we don't have (very much) redundant info to work with.

It's a better engineering tradeoff (given current technology) to
compress using a lossy algorithm, then expand using FEC, than to use
inherent redundancy as the error correction mechanism.  Lossy
compression techniques can make a file smaller by 10X or better.  FEC
makes it bigger by 2x to 4x.  Net savings: at least 2.5x compression,
and resistant to data loss.

<half-serious>
We always assume that machine code can't be lossily compressed.  But I
wonder...  I'm sure there are thousands of basic blocks in the Mozilla
executable that I've never used, and I'm sure there are thousands of
identical or nearly identical basic blocks.  How to find the
unimportant ones, and how to compress them down?
</half-serious>

-- 
Bob Miller                              K<bob>
kbobsoft software consulting
http://kbobsoft.com                     [EMAIL PROTECTED]
_______________________________________________
EuG-LUG mailing list
[EMAIL PROTECTED]
http://mailman.efn.org/cgi-bin/listinfo/eug-lug

Reply via email to