Fundamental Compressionist Philosophy.

[Detlef Morgenstern]

Hi,

Reinhold Messner says:
'But where there is a way cleared, the banal takes root, and the
mysterious gets lost.' *)

Same thing with symbols.
Symbols are patterns. Someone selected them from the huge space of
possible patterns to be a representation of an abstraction of some
regularity he has found and which he wants to communicate. This gives
each symbol a 'meaning'. 'But where there is a symbol established,
... ' it gets caught in the context it represents. It loses freedom.
Any attempt of finding the abstract of a set of symbols, will bind
you to the context in which the set of symbols was introduced.

You can no longer 'abstractly abstract' after having pre-selected a
set of symbols. You will 'contextually abstract'. You must have
knowledge on the symbol context (meaning) to be able to perform some
useful abstraction.

Shannon's Information Theory, which guides most of today's thinking
about information, is such a pre-selection. It assumes, that there is
a channel, over which symbols are communicated. The symbols 'fill'
the channel in an sense, that each possible pattern in the channel is
a 'symbol'. There can't be anything in the channel, which is
'no_symbol'.

In real life, 'the channel' is filled by 99,999999...% with patterns
which are (still) no symbols. There might be some regularity behind
them, but we still don't know. If we want to find that regularity
out, we must be willing to 'abstractly abstract' - assuming nothing
but 'there might be some unknown regularity'.

As soon as we reflect the stream from the channel in what we already
know, we will lose the opportunity of learning more about what we
still don't know. If it won't fit, we will ignore it: 'It must not
be!' That's our tribute to mental comfort.

Andrew, 'life is so hard in this group' because you must be willing
to throw overboard all the established abstractions (and their
symbolic representations) of what abstraction is - if you want to get
to a model (theory, algorithm, machine) which can 'abstractly
abstract'. Which means, it can abstract 'knowledge' from a world of
patterns, initially knowing nothing about the context which produced
the patterns. Advancing in its abstraction, the abstractor builds a
model of how the patterns might be generated. And I predict that if
it is a strong regularity, such an abstractor will be able to
correctly forecast patterns which **it still has not observed**.

Gerry, I now know what is our common point:
Among all possible ICMAUS frameworks, there is one which operates
solely on two symbols, '0' and '1'. This is the only one which I
consider being acceptably context free to make it a tool for 'broad
band abstraction'.

Thanks for asking such a lot of strange questions. A good question is
half of the answer. And it makes one think all that over and over
again.


Detlef
PS.
When I speak about 'my machine', this means a model for context free
abstraction. But that's the difficult fate of models: People won't
believe you until they can press a button and see how the machine
abstracts. And they will tear to pieces each isolatedly published
module - which, naturally, can't function on its own.

You might remember the film 'Amadeus', where the Black Ghost
'customer' asks the composer whether the composition is ready.

'It is ready. It remains to write it down.'

Which might take several more months. There seem to be no donations
for crazy ideas.

'If only we had known earlier! Why didn't you tell us?'
People, I told you.

*)Reinhold Messner: 'Wo aber ein Weg gebahnt ist, setzt sich das
Banale fest, und das Geheimnisvolle kommt abhanden.'


__________________________________________________________________
Do You Yahoo!?
Gesendet von Yahoo! Mail - http://mail.yahoo.de