|
Hi Tarvo,
I looked at registering at your forum but I cannot guarantee my posts are fully accurate as I
tend to think very laterally? Also I don't know if spam can easily reach my e-mail via web links to forum?
Just: I had to think about the agreement form so seems simpler not to sign while I'm not quite sure about the various legal aspects of the form.
However if I repeat my relevant earlier posts here (as follows) that may be O.K. as "artificial consciousness" could be regarded as a suitable topic on this forum here?
Some people teach the skill of reading, to children, by encouraging them not to get stuck on a word but keep going. My posts may look a bit ragged but that is because rather than stumble or fall at the first obstacle I tend to try and write out what I can figure out while there are still possible pieces not maybe fitting so well in the jigsaw so-to-speak.
***********The Structure Might Look Foamy Or Stringy?
Hi Tarvo,
internet is costly for me but thanks for feedback; hope these comments make some sense:
Quoting:
"It seems that without changing structure we cannot model everything. For example equation x=y*z establishes certain connections between x, y and z, but we cannot write any equation so that it for example will create new variable n and establish new connections between x and n etc, in equations such topology is established once and remains such forever. We can implement Turing machine though with equations (but only with iterative equations) because it is a state machine, and Turing machine can implement all processes, but then such equation doesn't model all processes the way they happen in nature because Turing machine implements processes sequantially while in nature they happen massively simultaneously.
I maintain the derivation in my web page exactly how I once did it, I think I would modify it a bit if I wrote it now. First the dimensions were just a thought experiment. If we come from constructing dimensions (a theory developed in Harvard university) and say that all dimensions can be constructed, then in space with no dimensions there also cannot be anything else than points (with zero dimensions) connected to each other (with connections with no properties).
****COMMENT: I found that "connections with no properties' is what the laws of physics describe. That is why I said they are voluntary and are only laid down by counting in a particular base.
It is possible that "counting in a particular base" is like "Turing machine iterative process". But you are right: such a "machine" perspective does not take into account interactions with "other perspectives", that is "other bases to count from".
When I talk of "counting in a particular base" I am really referring to the idea of "categorising".
Example: "counting in "base" "fruit" would involve labelling items as "so-many" fruit. This is perhaps a kind of "fruit-category Turing machine?
But if you counted them in "base" "colour", you may get a different result as oranges are not the same colour as lemons.
You might construct a "topological space" called "fruit colour" but if you do not distinguish between these two bases ("fruit", and "colour") where you count so-many oranges, so-many lemons, and so forth; I guess you just have a set of points that are connected but how they are connected is not stated?
You just have a set of "fruit-colour" superpositions connected as say points in a "fruit-colour" space?
To get properties (like "lemon has property of yellow colour") would require dimensions in the "fruit-colour" world: say the dimension (or perspective) of "fruit"; and the dimension of "colour".
" Now everything else comes from the condition that the structure must change (what is everything what *can* change there) and it was derived to be the only (simple) possibility for that what enables any change."
****For example: to get complexity in the "fruit-colour" world; you need two "branes" to collide: you need the category "fruit" to collide with the category "colour" giving colours that apply to fruit, and fruits that apply to colours.
This is why I call "branes in physics" basically like "times tables" or "factorisation sheets".
When a topology is constructed for the "surface" called "colour" and the "surface" called "fruit"; it is like a collection of common factors where "fruit" and "colour" share common ground in defining points of contact between items that are fruit and items that are coloured.
"Concerning cellular automata there are patterns what develop forever but only if there is a single such pattern in the whole space, ADS survive a long time in very random conditions."
****For example: the superposition "fruit-colour" itself superposed on "very juicy" gives a "knotty" problem: even if you have a "lemon-yellow" is it "very juicy"? If you have a "lemon-very juicy" is it yellow? (I think in practice ripe (yellow) lemons are the juicier.)
But early on are lemons not green? If you have a "yellow-very juicy" is it "lemon"? It might be some other fruit. If I link the "knot" that describes things that are both "lemon", "yellow", AND "very juicy"; and LINK this to some other such knot (e.g. "yellow, banana, ripe") I may or may not find other connections (in this case "ripe" applies to both knots).
IF "ripe" had not applied then I would have created a "not knot"! SO the new knotty question is: when linking knots: do I get a new common knot or not?
IF I do not get a new common knot; I have projected information about the topolgy or structure of the space these knots inhabit? As I have excluded "ripe" from "yellow lemon very juicy" if do that. So if the "brane": "yellow" collides with the brane: "lemon" collides with the brane "very juicy" collides with the branes "ripe", "banana", I may get a toplogy that "knows" that "very juicy yellow lemons" are "not ripe" if there is no common knot between the knot "lemon/yellow/very juicy" and "yellow/banana/ripe".
By definition "no common knot" would project "knowledge" about the appplicability of "ripe" in this limited definition scenario?
This looks like the double slit experiment: the double slit is "banana/ripe" inside the "surface" : "yellow" in the space-time: "yellow/lemon/very juicy". If you "close one slit" say make it just "banana" inside the surface "yellow" inside the space-time "yellow/lemon/very juicy":
how does the "photon": two versions of "yellow/lemon/very juicy" KNOW you closed the slit?
Because by definition "ripe" is no longer available as a means to differentiate (yellow?very juicy?) lemons from (yellow:?very juicy?) bananas.
By closing the slit "ripe" you actually opened two slits again by leaving unanswered the question: is a banana yellow and a lemon very juicy? Or are they both very juicy AND yellow? Or is a lemon yellow but a banana not yellow but very juicy?
By describing a minimum structure involving a few knots and links you can; if you require that new information be looked at from the perspective of non-common knots you will "photo-copy" the new information on to your starting information. This way your programme can "learn" things in terms of seeing them as a kind of obstacle course its own structure must navigate around?
Add more information and continue to require non-common knots and the complexity grows.
But the programme may quickly grind to a halt as soon as common ground becomes essential?
Once common ground was introduced it would spread through the programme like cancer eating up its topological structure until you had a single "flat space" or quantized field of say A:B:C:D:E:F:G:H:I:J:K:L:M:N:O:P:Q:R:S:T:U:V:W:X:Y:Z superpositions?
"Absolutely dynamic systems (ADS) are not exactly based on Chris Langan's interpretation of consciousness because they are not based on language, language may come only from emerging processes. Also the only thing I saw in common with Hawking's "nuts and bolts" was that the points there also have zero dimensions. But there certainly may be similarities because ADS are derived from the most common principles. Consider this part of my derivation:
> "It can be that both the new knot and the adjacent knot are connected with the same knot, but there can also be no common knot with what both the new knot and the adjacent knot are connected with. This is the simplest possibility to make a distinction between the pairs mentioned above."
> Brilliant!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
> Exactly what I found and what I think Chris Langan found!
This may indeed appear in many theories because this is the most common principle - to find common between two sets. This is the form in what it appears in my derivation. What else comes from it is that if we connect the *different* (not common), we get association and this is the most basic process the ADS are based on.
> The learned words would be understood in the system's own language; it might build up more understanding in further developing its internal language?
If there would develop a language. I taught it only as much as it was necessary for testing the program and I didn't even try to teach it more. The problem is that there is not yet any theory on how to teach a system what knows almost nothing. This seems to be a simple problem as we all think that we can teach, but it isn't. If we try to teach it just by interacting, then it would take too much time. So we must write a "teaching program" what does exactly the same, but for that we need some theory of how to write it, what most likely would come from the principle that such system tries to act so that it can predict the results of its action. So the problem of such "teaching algorithm" is much more complicated than it may seem to be."
****I suggest would it work to teach it to build "cells" of common ground (so allow common knots but separate them by "cell walls" of non-common knots)?
"You and some others seem to know more about physics than I do. This is why I'm somewhat reluctant to develop any principles of how physical systems may be modelled with ADS. What I did was not physics and so whatever may be related to physics in a way or another, I don't want to argue that things *are* in one way or another so that others may just say whether it is right or wrong, or to go in wrong direction. So it's better that people who know more about physics develop theories in physics, even if such models like ADS may be useful for that. It seems that there is at least someone who likes one idea what comes out of it, that propagation of light is a propagation of a process. One thing why it is hard to talk about ADS or anything else what is based on very general principles is that it may be as talking about the whole world. I'm sorry that I didn't comment everything what you said, I think many things were correct, but it's just impossible to talk about everything at once."
****I think many subjects are connected and it can be easier to understand one subject by seeing how it appears in another subject.
MORE COMMENTS
I tried to comment your post as much I could, it's nothing very systematic though but I hope that it may still be useful as some kind of explanation.
> In a mutual word definition-space: the structure COULD be constantly changing; if mapped in a minimum math-way it would appear to be constantly changing.
It can if definitions change when something changes.
> Of course we are talking about mutual awareness of specifying and generalising characteristics in defining something; so such a definition-space implies a mutual awareness among items. The structure overall may seem like it is "conscious" due to the inter-dependent relations.
Inter-dependent relations are necessary but not enough, many things must be possible to haappen.
> By holding each "knot" or "link" as unchanging (equates to Dr. Dick's "conservation of center of mass?) seems to be like saying that any additional detail introduced to the definition-space (such as adding category "D" say: "with tyres" to the aeroplane/wheels/wings interaction) does not change the knots (overlaps of overlaps projecting potentially new space) and links (overlaps) among categories. But the structure itself changes.
Possible.
> This triple effect of three categories meeting might be called "time".
? Time comes from that every generation of new knots would generate new generation of new knots, this may thought to happen instantaneously everywhere."
****I think your system is generating what they call "imaginary time" by its requirement of "not common knots"? Maybe your programme is like an inverse of Dr. Stafford's idea: instead of "set of numbers; add imaginary data; add imaginary data so that minus any one item and set is still of unique items" your system might be: "set of imaginary data; subtract one item, subtract one item so that any imaginary data is still unique"?
(Explanation: "Set of imaginary data" would be your initial few knots and links. "Subtract one item, subtract one item so that any imaginary data is still unique" would be "not common knot defined by two items subtracted after an interaction, so that the two are not common. The few knots and links "surround" the subtracted items effectively making them cells surrounded by membranes? )
"> By introducing a fourth category "things with tyres"; you still can have the three knotted as before and linked as before but you can generate a new space or new dimension on this it appears say:
When whe had "things with wheels" then we would have new knot what means "wheels and tyres", this is association because "things" is common :-)
****Seems O.K.
"> (Double-slit experiment: two views of three categories may be called "photon"? Pass this through double-slit is like asking "which two is which? Two that defines photon or two slits? So you lose track of the photon? It becomes "everywhere" in the experiment? But if you specify "past" and "future" you get "where is which?"? So you get "offer wave" and "confirmation wave" as in John Cramer's interpretation of quantum mechanics?)
> Introduce a fifth category and "space-time" is uncertainly allocated; which of five are the four and which is the extra dimension in space?
Time is not category and space is not category. To simulate space with ADS there must be some nodes where angles between connections are somehow determined (or marked with patterns if process interprets them in a proper way). Now if angles are determined as in 3d space, this is 3d space, if as in 4d space, then this is 4d space. Yep, in fact in any such system would emerge some "processing nodes", relatively independent structures what are connected to each other, though connections may change. These may even be "particles" :-) These nodes may also join, split, new ones may be created, whatever."
****I see that the "non-common ground" requirement may effectively keep "time" and "space" in superposition (not differentaited as to which is which) say? Not sure.
As soon as you introduce restrictions like angles and nodes you would perhaps make your system "stringy" ?
"> It is the existence of that "possibility space" of two other options that makes the "three quark anti-colours" of your ONE choice of three options into a "three-affected" choice. Perhaps this is why "anti-colours" are regarded as "going backwards in time" if "time" is "triple-ness" (self-referent reference like pendulum self-refers by retrace alleged same path)
Don't know enough to comment this. There would always be iterations when a process or sequence of processes cause itself. This may also be as a kind of dynamic memory.
> Is the very definition of "number" entangled here? Or rather, is the definition of "category" caught up with "base" in which numbers are counted ? What is going on?
Ha, what would numbers look like in ADS? Like first position is connected to pattern three and second position is connected to pattern seven...
> It seems like a "space-expansion" system?
Like space-creating system :-)
> This is just like my "discussion model of physics": I call it "transparency" where everything is built from mutual-agreed space so all constructs are built of pure consciousness.
ADS may not always implement artificial consciousness, they may also implement a very "mechanical" system."
****I wouldn't say it implements artificial consciousness rather it builds a hollow structure around things but relies on differences between things being transcribed into its own topology?
"> May I suggest you have inverted the usual arrangement: instead of a rigid geometry in which events are played out; you have made rigid events in which geometry is "played out"?
Maybe, Heraclitus said something that the rivers don't change, but they are a result of change.
> Maybe your computer programme is an inversion of "cellular automata":
Or cellular automata in space with limitless number of connections from every point :-)
> What if "the law of non-contradiction" is the only fixed reference in a dictionary?
Laws are rather something like association and natural selection.
> "knot" becomes "category overlaps with category overlaps with category", so A:B:C
You may consider knot as a set what consists of connections to other knots, this is how sets are represented in ADS.
> (IN QED: the direction of the arrow may be D:E and the length of the arrow may be A:B:C.)
The links have no direction, any direction must come from the process itself."
****I realise they have no direction as such; but in QED direction is only relevant relative to other directions it seems. What counts are the differences in directions and the frequency of those differences I gather?
"> Like dissapearing Chess-move-options which have no logical-compatibility common ground with a new move made by a player?
Like dissapearing the parts of the system what don't fit into the rest of the system. This enables trying the possibilities and association enables generating every possibility in the environment."
****But my impression is you required that "common knots" be not included? This would limit the system. It can only count things?
"> to Chess-move options not yet available as spaces occupied
These are just possibilities considered in derivation, in ADS new knot shall be created only from links what are different to two knots because only this possibility doesn't cause system to constantly increase or decrease.
> If new moves in Chess point to non-common spaces (ones that are not in your game-plan of moves) you have a constantly changing game-plan so "an absolutely dynamic way to play Chess" where you never rule out any move? Like a discussion where every view gets a hearing and no one is left out...
Yes, this way we create something new.
> Like a few cellular automata with no grid to play on?
Almost. In what ADS are similar to cellular automata is that both are self-developing.
> You get a rapidly built grid that keeps changing?
Yes.
> Then it settles down as the grid stabilises? Due to feedback loops between the cells and their new environment?
Yes, because only these parts of system remain what can maintain themselves in certain environment.
> I see why you say it looks like it is conscious.
To satisfy the criteria of artificial consciousness, more about that in my forum.
> Would it not tend towards a Bose-Einstein condensate type state? Eventually it would generate so much feedback that it would break up into a whole lot of individual cells and die?
May happen, it shall collapse when constantly more knots shall be deleted than created, this doesn't suppose to happen when the system is well-developed.
> (The static starting state might cause problems by speading "background noise" through the system? Or ? Not sure...)
If starting structure is minimal then very soon after some interaction the starting state shall have almost no influence and almost all system is only determined by interaction. But yes, a single thing may influence a whole system, this is why such system has some "holographic" properties"
> And the kind of BEC death your model might experience: why ! Cellular division! Runaway cellular division! Cancer!
Yes, at least in theory it can model cellular division. Equations cannot do this well because they cannot directly model changing structure."
****Comment:
I think I understand it to some degree; but I do wonder if the system will eventually require "common knots" to avoid settling into a "flat" grid of static cells or something?
Regards,
Alan
To Tarvo: Sketching Some Ideas Re: Physics Patterns :
looking through my earlier long response I have thought of things:
I suggested that three categories seen again (which I described as conservation of the definition of three categories; so "again" is implied) gives:
"mark" as the "again"? (interaction or observation is implied by "again"?)
The six possible views as "six quarks".
I wrote:
"But where are the three quarks? Of three categories that meet, where a question is asked "which are the "first" two say?" for which there are three possibilities for the say "last" one in the quark "line-up" (Superstring theory!?); and where the categories meet again and the question is asked again:"
Now it occurs to me:
if you generalise "first two" and "three possible as the last one": you get:
(choose first 2 out of 3; 3 options for last one)
times 2 times 3.
This allows the definition of "2" and of "3" to be undecided? But an outside look (count this uncertainty cell by observing it) gives:
five string theories in 3 x 3 that is nine dimensions? (just as they say in physics...)
You get a 2 x 2 aspect that is a "branes" aspect.
You get a (2 x 2 view of 3 x 3) that is a loop with "randomness" (that is with "double-booked or interfering definition overlaps since have 2 + (3 x 3) as "five" so need quantum electro-dynamics (the distributive law mixed through itself by changing roles of addend in brackets with multiplier on outside of brackets).
Idea is choose either, 2 x2 view OR 3 x 3 view; if superpose both must have 3 x 3 IN 2 x 2; AND 2 x 2 IN 3 x 3:
so 9 in 4 and 4 in 9:
so 4 x 9 = 36 dimensions for string?
Quantum: divide the loop gives gravity: condensing the "randomness" into a cell as get breaking of (2 x 2)/ (3 x 3) into (2/(3 x 3) ) + ( 2/ (3 x 3) ) seems like quadratic formula a squared + 4ac + b squared.
Generalise 3,3 as "time/space but if map this "as a new view of 3" as "4 where one of the four is the other view of three" you get five-fold symmetry when interact with a cell of this "3 new 3" seen as "4-ness"?
Dr. Stafford's paper then maps this generalised 4-ness as a 4-geometry so he gets a polygon rotated in n-dimensions (he is counting 4 x4 space that is ways of finding 4 in 4?)
He gets space 4s (polygon) and time 4s mixing cause and effect in his paper's perspective?
A 2,2 view of (3,3) 3 gives fixed grid a 2 with "1" (second 2 seen as a unit i.e. 1) view of 3 x 3 x3 or 27 so get 27 and sub-atomic particle symmetries?
Categories:
I will call a category a "cat." for short.
"cat.cat.cat" means here that these three categories may share common ground so may overlap.
Consider three such cat.cat.cat 's:
cat.cat.cat; cat.cat.cat; cat.cat.cat:
the whole thing might be CAT.
Internally there might be c a t (sort of ghost-cat) that is the non-common ground that separates them inside CAT.
This looks like Dr. Stafford and his "adding unknown data".
Maybe "add unknown data, add unknown data minus any one item to make set items unique" becomes:
add ghost-cat, add CAT so minus any one item to make the cat.cat.cat.; cat.cat.cat.; cat.cat.cat unique SO he quantizes the cats?
Five-fold symmetry:
#1: CAT; #2: cat.cat.cat; #3: cat.cat.cat.; #4 cat.cat.cat; #5 ghost cat.
But Dr. Stafford uses a 4-geometry which gives 4 x4 so polygon in n-dimensions.
A 4 x 5 view gives a way to have negative-mathematics (cancelling mathematics say)?
Not sure.
So have 4 x 5 = 20; but take one (4 x 5) perspective (one definition of 4 x 5?) or of space-time complexity say? get 20 minus 1 = 19 constants in physics standard model?
Generalise "minus 1" to the 4-ness get square root of minus one seen in 5-ness?
Or get 5 theories of "4-1" that is five string theories with dualities (as the 5 is from a 4 and a 1 in "4 minus 1"?
These seen in a 4 + 5 that is in 9 dimensions?
CAT (that is overall category or "supposed universe say") may be seen as "definition"?
ghost-cat (the uncommon ground that keeps cat.cat.cat 's from being muddled inside CAT) may be space between cat.cat.cat; cat.cat.cat.; cat.cat.cat may be seen as "rule"?
Defining CAT (external overview) and ghost-cat (internal space betwen cat.cat.cat 's)
the question is like what Chris Langan says in conspansive duality?
Not knowing how you get CAT or ghost-cat in terms of each other; this might be called "stochaistic process" ("random") as 1/4 turn in cat.cat.cat is what separates it from another cat. in that other cat.'s cat.cat. company?
Two views of cat.cat.cat gives coming together as gravity when take 1/4 turn view (take one cat. view from cat.cat.cat. of other cat.cat.cat.?)
But the other two cat.s in the other cat.cat.cat give a ready available two views of cat. to give a 1/4 angle on cat. and cat.cat.cat.?
Thus gravity (1/4 turn -ness) is quantized here as a potential shell.
To see it (fill the shell) you have to count it: requires another cat.; so you get a loop, a quantizing, a 1/4 turn that is uncertain (possible gravity) but which is which? You are stuck with a "graviton" and that whole double slit business (where is the graviton?). (Double slit business: two cat.'s in a cat. surface in possibly linked to otherwise with another cat.cat.cat.)
The other two cat.cat.'s in the cat.cat.cat. from which you took one cat.'s view: gives a loop?
Or is it a quantum (as it involves two so division)?
But if count this from outside you get opportunity to make up who is to play the role of "loop"; who is to play the role of "quantum", and who is to make the play "gravity"?
Bit messy.
Spinors and twistors:
may regard cat.cat.cat. as "spinor"?
may regard cat.cat.cat. and cat. as "twistor"?
may regard cat.cat.cat. and cat.cat.cat. as spinors and twistors in complex number space IF freeze two of the cat.s?
Super-spinor:
cat.cat.cat.; cat.cat.cat.; cat.cat.cat.
Super-twistor:
ghost-cat (as creates space between the cat.cat.cat.'s so puts a twist on the scene.
-just a rough sketch
-dolphin
|
