|
Authored by: Anonymous on Wednesday, December 05 2012 @ 04:52 PM EST |
The thing about sign-vehicles is that it is OK if they change states. We've all
seen light arrays that say things like "Slow Down" which changes,
after an interval, to something like "Congestion Ahead".
I don't think anyone would dispute that the device that does this is a
sign-vehicle.
I mention this because classic Stop signs, for example, don't change their
state. If you think of them and then of a computer you might be misled into
believing that changes of state render it something other than a sign-vehicle.
I'm just trying to cut short that train of thought.[ Reply to This | Parent | # ]
|
|
Authored by: Anonymous on Wednesday, December 12 2012 @ 08:33 AM EST |
If you give me such a short string of bits, Kolmogorov complexity won't help me.
If you give me several megabytes of data, it will help me. There are many entire
fields built on the premise of understanding signs with missing context.
One is WW2-style cryptoanalysis. In this case Turing et al get a mountain of
enrypted message- string of symbols, where part of the context (which ciphertext
letter corresponds to which plaintext letter) is intentionally missing and the
relation is intentionally complicated.
The second one, which springs to mind, is reverse engineering. You get some
data, which has -often intenionally- lots of missing context. Then you
reconstruct the missing context.
A third example is the understanding of Cuneiform script. Again, we have a lot
of missing context (we do not know the language!), which we can infer. (This was
one of the most amazing feats of archeological / linguistic science)
Another nice example is genome analysis of sequencing data. Often, we are
lacking the context necessary to execute / fully understand the data, e.g. we
are missing the understanding of the relevant binding sites for proteins, which
regulate gene expression. The living cell uses this context in order to decide
which parts of the genome codes for proteins and which parts to express.
Game over? No! For example,we can use statistical methods like
wikipedia:"molecular clock" in order to identify sites which feel
evolutionary pressure and infer many interesting things from sequencing data!
(even if sequencing data alone does not contain enough context to reconstruct a
living cell)
This is to say: If you claim that some data is useless without context, you need
to argue/prove this.
For example there are very good reasons to assume that AES-encrypted data
without the context of the encryption key is meaningless in our physical world
(which has limits on the amount of computation).
Note that this statement does indeed quantify under which context (no crypto
key), which general assumptions (current understanding of the laws of physics,
standard assumptions of crypto [current mathematical state-of-the-art is unable
to prove these assumptions]) the data is meaningless in which sense (plaintext
and no "easy function" except for length of plaintext can be
recovered).
Also note that "no context" is not possible. There are several
implicit bits of context like "data has been considered important by a
human beeing", which are very powerful.
[ Reply to This | Parent | # ]
|
|
|
|
|