Re: object system...


"Dmitry A. Kazakov" <mailbox@xxxxxxxxxxxxxxxxx> wrote in message
news:1qlpwnvjqzagv.1tk2ls4xkiveq.dlg@xxxxxxxxxxxxx
On Sat, 17 Jan 2009 00:10:17 +1000, cr88192 wrote:

"Dmitry A. Kazakov" <mailbox@xxxxxxxxxxxxxxxxx> wrote in message
news:6ik1oc0h4teb$.136i402exchjw$.dlg@xxxxxxxxxxxxx
On Fri, 16 Jan 2009 00:41:16 +1000, cr88192 wrote:

so, as I see it, the whole paradox is a misnomer, if anything, because
this
would not turn out this way in practice (and, if set up, the barber
would
invariably end up making an alternating set of decisions...).

he would evaluate "do I shave myself?" and would remember answer from
the
last time. then it would work like a binary negation, and he would
choose
the other way, and the next time, the opposite again.

granted, this assumes the barber used imperative logic, an actual human
being far more likely to realize that the whole thing is silly...

The paradox only says that the statement is contradictory within given
logical framework. No more, no less. You are trying to interpret it as
something absolute having its own merit. It is not. It only shows the
bounds of a given language, and is conditional to the logic. You have to
limit the language in order to avoid paradoxes. Otherwise, you have to
deal with the consequences.

You can deploy a logic that incorporates some paradoxes as proper
statements (Belnap logic, fuzzy logic etc), but then you automatically
lose the porperties these paradoxes violate, like the law of excluded
middle
etc. But even so, 1) each such logic is self-consistent, 2) in a
deterministic environment you are forced to make choices anyway, 3) you
still need classic logic where your logic is a model.

groan... I continue not really understanding what is being said...

To put it simple, all extended logics have to be based on a
self-consistent
"normal" logic, where barber does not exist. The way they work is to
consider "exists Barber that shaves ..." as a formula. At some point where
you wished to drop the quotation marks and *decide* if the barber indeed
does this or that, you land back into paradox. In fuzzy logic this is
called "defuzzification." Some things cannot be defuzzified, because they
are, well, fuzzy. (:-))

The bottom line, you can reason about contradiction (=contradictory data),
but you shall not reason contradictory. This the same story: data /=
program (behavior), subject /= object, set /= element, never reason about
yourself...


I guess the next question is, WRT a program, what exactly is "self"
anyways?...

like, there is some piece of code or data that is unable to be processed,
and everything else is ok?...


but, in any case, the computer is its own self-reinforcing pattern,
already
well beyond the stage of a formally-defined system operating purely on
input, output, and intermediate steps...

No. If all incomputable elements (clock, random generators, keyboard
input)
are considered as inputs, which they indeed are, then any computer
becomes
a finite state machine. It does not have either soul or free will.

Some (many) people belive that we too are finite state machines agitated
by
infinite inputs. So far nobody showed either.

well, issues of "soul" or "free will" are outside the scope of what I am
considering (first one has to define what they are exactly, after
all...).

Infinite, hypercomputable etc.


ok...

well, I guess that is a bit different to how I was seeing things:
soul: very possibly a continued manifestation of a physical body following
death, very possible wave-like and/or composed of some sort of "soul
matter"...

free will: ability to make decisions in such a way that one can be credited
with having made decisions (and responsibility for these decisions). I am
not sure what exactly could give free will, only that I am fairly sure it
can't be manifested within a purely deterministic process...


the issue though is that a computer remains deterministic, but has long
since ceased being definable purely in terms of input, process, and
output.
FWIW, this notion breaks down much past the level of an Altair or early
mainframe.

as I see it, computers operate under cybernetic principles, where pretty
much all of the computer-world operates in a sort of loop, and where
humans
have become partly an agent of this system. now, I am not saying this in
the
sense that humans have "physically" (or even really mentally) unified
with
the machines, only that, at this point, no machine is built or operated
purely independent of other machines (both hardware-wise and
software-wise),
activities by one person on one machine (or by the machines themselves)
influence later actions by other people using other machines, a kind of
machine-world "ecosystem" exists (even if in the form of continuous
activity
by humans, ...), ...

No, this is untrue, because we still design and program our computers in
isolation of their "human subsystem."


but, the actual creation part can no longer be done apart from machines...

to write a piece of software, we still need the OS, text editor, compiler,
....
thus, we can't program computers apart from computers, ...


but, we can craft other tools and objects apart from these objects'.
so, to smith metal, or carve wood, as I see it, is a fundamentally different
process from that of computer use or software development...


in the past, humans were the primary means of information storage and
processing. any information that was to be stored, was stored by humans
in a
human accessible for (human memory, scrolls, codexes, books, ...), all
information processing was done manually (and primarily via human mental
faculties), and likewise, any human creation was done by a human
themselves,
and by tools made through human effort.

so, for example, one person is a blacksmith, and they make hammers and
chisels, and these are sold to a craftsman, who carves wood or chisels
stone. in part, it is the hammer and chisel that does the work, but they
are
weilded by the person, and going back one step (to the blacksmith), they
are
a product of human effort.

Yes, a system computer-human [possibly] has different properties. BUT, we
do not program that system. You make a logical error talking about
programming and programs referring to the larger system. It is as if you
talked about astrophysics being progrmmed because computers (and we)
consist out of burned out star material.


matter is, however, a substrate.
since it is well established that humans can't live (or exist) apart from
matter, it is not necessary to consider it.

more so, raw matter does not exhibit any kind of cybernetic (or, AKA:
homeostatic) properties...


as I see it, modern technology is currently in a gray area between that of
inanimate objects, and that of being animate (or, as it could be said
another way, "living"). however, as I see it, they still have a long way to
go...


so, we rely of machines for these tasks. a machine to make the DVD, and a
machine to understand and utilize its contents...

Even if a C++ caompiler can compile C++ it is still a product of human
activity.


yes, but the human does not actually operate every aspect of the process by
themselves.
they give the compiler some code, and say "compile this", and the compiler
completes this task autonomously.

a person enters the code via a text editor, but here is a problem:
no part of the text editor physically exists, nor is it directly maniputated
by the human.

the direct human interface with the software world actually ceases at the
keyboard and mouse, and the view into this world ceases past the monitor, an
essentially different sort of reality for a different sort of system.


the text editor only exists because the computer keeps running, and it keeps
running because it has certain properties.

one of the more major (and basic) of these properties is that of turing
completeness, where the simple act of adding the requirements for this
property, makes the system suddenly (and drastically) different than it
would have been otherwise.

now the thing is, not everything in a computer is itself turing complete,
but this property can be extended further, with possibly impressive
effects...


turing completeness exists at the basic level, leading to many of the
interesting properties, usages, and costs of these systems, and it is the
usage of this property that characterizes the system as a whole...

apart from this property, computers as we know them could not exist...


yet, even if the computer itself is turing complete, often the compilers and
tools, and a good portion of the software, does not exhibit this property...

and, amusingly, one of the major properties exhibited by a turing-complete
system stands in conflist with the traditional definition of an algorithm: a
turing-complete system is both capable of, and can get useful work done,
while running in an infinite loop, whereas by the traditional definition an
algorithm can't...


and how does the machine keep running? it runs itself... (we can walk
away
from a computer running windows and doing stuff, and what then?... it
keeps
on running windows and doing stuff...).

Which is not different from an endless loop. It does not "produce"
anything, or speaking formally it does not create new states. It
enumerates
some finite, fixed set of states. This set is closed. If you wanted to
open
this set, you would need some *external* factor. For example, a hardware
clock to put time stamps on states. Any "non-triviality" comes from
outsite.


this is a different set of concerns...


Again, it is a question of belief that "non-triviality" of human activity
is "trivial" without random generators used at the neuronal level. Maybe.
That is another story. But computer systems are certainly "trivial," per
design. And any finite combination of them is "trivial" too.


yes, and this is arually one of the major interesting properties of using
TRNGs in some algorithms...

however, though I can observe that TRNGs do interesting things, I am not
sure how exactly to describe or categorize these properties...

but, then one can argue that maybe all the noise that can be mined from the
bus by making use of rdtsc and similar is all deterministic as well, but
evidence does not look this way (too much residual activity happens due to
external factors, making variances in the exact number of clock cycles
between iterations, which can be mined for entropy, among other sources...).


There is a fundamental principle that a logical system cannot reason
about itself. So any proposition like exists (self) as well as not
esits (self)
is automatically either inconsistent or undecidable. It is a wonder
how
humans managed to do this stuff all the time in the history of
pilosophy... (:-))

well, I am not sure the basis of the above claim (that a system is
unable
to reason about itself).

You can see your hand when it is at a distance from your eyes. But you
cannot see your head.

a mirror is sufficient to see my head...

Mirror produces a model of your head. Now you cannot see yourself looking
at the mirror...


I see my head as it looks at the mirror...


Reasoning about self, at full, means that there is a model of self. Does
that model includes the model itself? A model of model? And so on.

so, do you think a homunculus necessarily exists?...

I don't know it. See above. Human brain contains incomputable elements.
Whether their role is trivial (inputs) or not, I cannot tell.


ok.

well, personally I don't believe something equivalent to a homunculus
exists, because that would make a problem:
how could this exist without itself containing a homunculus...

this would keep going on either infinitely, or maybe there is a homunculus
with the one inside it being itself (AKA: it itself is shaped sort of like
out good old friend the klein bottle...).

I think Leibniz had the idea that this was a kind of "monad", but then there
is a problem: as I see it, no kind of monadic entity would have sufficient
state to retain properties like conscious identity, ...


a simpler idea is to assert that no such entity exists, and that "self" is
more of an emergent property (I would seem to exist, but no single part of
myself is "myself" as such, ...).

if you take a person and hack them apart, you don't have a "trimmed down"
person and a bunch of spare parts, all you have is the parts... (although,
presuming they are dead, one can say that the "self" part of them flies away
at death, but oh well...).


part of what I was going on was derived from my understanding of
Descartes, ...

If Descartes knew about Gödel and Cantor's diagonal proofs, he probably
would not become a philosopher.

these proofs prove what they describe, but not what they don't.
not every system is of a logical or mathematical nature.

Descartes would reject this statement, because it means that some things
cannot be comprehended or logically reasoned about.


potentially, but as I see it, there are systems that can be reasoned about,
just not all of them with logic as such (logic, as it is understood, is then
of a number of possible systems, each with their own capabilities and
limitations...).


but, alas, on this point I think my thinking is different from Descartes...

for example, both him and Aquinas had some rather interesting points (along
with Pascal, ...), but alas, they operate within a world subtly different
from my own.

actually, some of these points have just been thrown in, and in the case of
my world serve as structures keeping the whole thing from imploding
(actually, parts of my world can sometimes be compared to a submarine, with
lots of terribly useful axioms thrown in the mix to keep the whole thing
from collapsing under its own weight...).


many systems are, by definition, "incomplete", but this does not prevent
their usefulness or continued operation.

All formal systems are. Your statement is different. It was that there are
things we cannot, shall not reason about. This statement would shake
Descartes. But the statement is true, as Gödel showed. And anyay it is
impossible to comprehend nonsense.


speaking of nonsense... I recently beat together a textual analyzer and
generator based on a 'Hidden Markov Model'... it generates text that in some
cases almost makes sense, even if it is all just a big set of associative
graphs and random walking to produce the text (AKA: there is no
"understanding" in the system...).

but, alas, an HMM is something which is far from being complete, but can
still get "useful" work done...


if it works, it works...

No, it does not. The answer depends on whether "it works" is decidable in
the logical system used. And it is not. So you don't know if this
statement
is true. It is merely a belief.


that things like foundational arithmetic have some rather gaping holes in
them does not prevent the convinient use of counting, or the implementation
of ALUs...

nothing stops a monkey from learning to pick locks either, even if it is
possibly the case that the monkey will be able to fully understand or reason
about the locking mechanism...


that something works is sufficient grounds to make use of it, even if the
logic behind it seems to be incomplete or broken...


this is in contrast apparently to regarding the computer in its own
right,
and considering the relation between the user and computer (or the
computer
and reality) as much more bidirectional (like, the computer and the
software
world already exist, I more interact with it and try to work on it and
make
things better...).

OK, but that is not programming. If the computer reality already exists
(like in a computer game, or less exciting, like in an office package),
you
do not program it. You simply interact the reality. It is the difference
between programmer and user. User is a subject of reality. Programmer
creates realities.

the programmer does both. they write code within the existing computer
reality...

You mean "human" here. Programmer is a role. It is not both. A human can
play different roles at once.


maybe, but it does not mean they are not a programmer if they operate within
the confined of reality...


for example, a programmer will utilize an existing compiler, OS, and
libraries?...
well, they likely did not write these components...
and they didn't make the hardware either...

thus, the computer "reality" already exists...

Sure. There are many of them.

I had partly reworked the notion of an NN a little for my uses, more
into a
kind of associative node-graph with possible embedded procedural logic
and
customizable neuron behavior (each neuron may have a type containing
customized behaviors). still doesn't make them all that useful for much
(and
there is a major weak point that this particular design can't be
"templated"
or "instanced", which is a little bit of a detractor for many possible
uses).

Looks more like a good old decision tree.

maybe, but not necissarily...
a decision tree is typically laid out in code...

this is a bunch of objects linked together with a kind of
accumulate/fire/backprop thing going on, which is a little different...

Yes, I meant this.

But it is interesting. I always played with an idea of a sort of AI for
equivalent tansformations of program (mainly for optimization, but also
for
correctness checks). But I never came to any conclusion.

I had some "interesting" experiences today messing around with using
Hidden
Markov Models to analyze and synthesize text (basically, it spews out
masses
of largely incoherent rambling... but some of it was interesting... but
more
in the cases where the text was almost workable but otherwise absurd).
doesn't necessarily mean it will be all that useful though...

actually, I have been left wondering a little if there is much possible
use
for this kind of thing in any of the coding I will be doing (at least
HMM's
give tolerable performance, unlike GA and GP, but building and storing a
higher-order associative model is not exactly cheap as far as memory
goes,
and along with the costs of building the model, somewhat limits the
practical scalability of the modelling and synthesis capabilities...). it
is
left as a mystery if there is a good way to hybridize GP and HMMs...

Probably not in coding, but indirectly, to achieve higher abstraction
levels. Presently much of work has to be done manually. Some AI support
like: "a program like this, but ..." For refactoring etc. Much of generic
programming is like that.


maybe...

reducing the amount of coding is one use of autogenerated code, since a
piece of code can be written for a general task, and specific situations can
be given as data, and the algos can figure out how to write the code needed
to get it done.

however, I am now thinking that HMMs are probably better for perception than
synthesism, as they could be used for recognizing and responding to
situations. combined with some feedback, this could be a fairly effective
way to implement some kinds of learning behavior (absent needing GP or NNs
as such...).


The answer is much less efforts. This is the major reason why we
chose
Ada.

now, can you convince the rest of the world of this?...

It would be quite useless to try. Software developing industry works
in a
way that selects technologically inferior solutions. This includes
languages and libraries. There is no way anybody could change this.
Maybe
US Congress could, by making all vendors of commercial software fully
liable. Since we have a global crisis, and many software vendors will
vanish anyway, so it would less painful to do this right now, but
surely
they would not. So far I am glad to be able to convince some of our
customers not to push for C. This alone took years for work.

it selects typically for the cheapest options, whatever these are...

No, that is not true. It tries to get as much money from the customer as
possible. He buys the crap f0r $1 and then spends $99 for support,
maintenance, personal training, off times. The system is built in a way
that $99 is not considered as a loss.

that is the vendors, but the course things in general do not necessarily
take the vendors' sides...

anyways, I meant 'cost' here not in an economic sense...

But it is all economic in the end. A technology that is not economically
feasible cannot sustain for a considerable time, even if you managed to
pay
others for it, as in the case of software products. That only agravates
the
problem, because the real expenses are invisible. We already had the IT
buble. But the software buble is much bigger...


hmm... commercial software blows up and everyone uses opensource...


but, yeah, typically the "cheapest" answer is that which can be achieved
with the minimal net impact, and has the most generally needed tradeoff of
properties.

so, for example, the languages which tend to rise to influence tend to have
a lot in common with the last language which had influence, many other
technologies make incremental adaptations, many times things developed for
one purpose are just sort of kludged over to do something different, ...


as a simple example, consider all the data being stored in JPEGs that is
clearly not photographic images:
texture maps;
normal maps / deform maps / ...;
height maps / bump maps / parallax maps;
light maps;
vector fields;
....

so, a person writing a 3D engine is like "well, I have JPEG, and it stores
graphical data", where graphical data soon comes to mean "pretty much
everything that can be represented as a grid of low-precision normalized
vectors".

likewise, the GPUs graphical facilities become generalized, and so textures
can hold all sorts of data, which is typically not drawn directly but ran
through pixel shaders.

....

oh yes, and now we can do physics on the GPU, ...

and other people are like "wow, the GPU is a good way to do neural nets", or
other people have the idea of doing whole bunches of financial calculations,
cryptology, ...


--
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de


.

Relevant Pages

  • Re: The Promise of Forth
    ... but that can simply reflect the difficulty with language. ... And when your language includes internal contradictions, ... But when you find contradictions in your ... But there are nonstandard logics that can accommodate this just as there are non-Euclidean geometries. ...
    (comp.lang.forth)
  • ESSLLI 2009 - Final Call for Participation
    ... The European Summer School in Logic, Language and Information ... Non-deterministic Multi-valued Logics - Arnon Avron and Beata Konikowska ... Foundations and main properties - Philippe de Groote and Sylvain Salvati (Introductory course, ... An introduction to minimalist grammars - Greg Kobele and Jens Michaelis (Advanced course, ...
    (comp.specification.z)
  • Re: The Promise of Forth
    ... logical contraditions, it's sloppy. ... but that can simply reflect the difficulty with language. ... But our common culture still holds logic in a superstitious regard it manifestly doesn't deserve. ... (there are many "logics" these days, just as there are many "geometries") ...
    (comp.lang.forth)
  • WoLLIC 2010 - Call for Papers
    ... WoLLIC 2010 17^th Workshop on Logic, Language, Information and Computation /* ... Each meeting includes invited talks and tutorials as well as contributed papers. ... It is sponsored by theAssociation for Symbolic Logic, the Interest Group in Pure and Applied Logics, the The Association for Logic, Language and Information, the European Association for Theoretical Computer Science ... Proposed contributions should be in English, and consist of a scholarly exposition accessible to the non-specialist, including motivation, background, and comparison with related works. ...
    (comp.specification.z)
  • Re: some ideas about db rheory
    ... language level and intrinsic properties are not. ... or what do you think your app is?), there are obvious first steps that various people might re-arrange in ways they prefer: 1) find out what answers are needed from the db, 2) find out what facts are available to infer them from, 3) reconcile the results from 1) and 2) and 4) organize the facts. ... I don't know why you might suspect I had anything to say about truth. ... To the man or woman in the street they will proclaim various self-serving 'logics' to do with propriety of some kind, usually the proper management of the most basic human motives and inescapable behaviours. ...
    (comp.databases.theory)