**From:** *lispunit_benrayfield@audivolv.com*

**Date:** Mon Nov 15 2010 - 21:47:59 MST

**Next message:**Natasha Vita-More: "RE: [sl4] Simple friendliness: plan B for AI"**Previous message:**Luke Griffiths: "Re: [sl4] Simple friendliness: plan B for AI"**Messages sorted by:**[ date ] [ thread ] [ subject ] [ author ] [ attachment ]

Max Tegmark's theory ("All structures that exist mathematically exist also

physically") is summarized at:

http://en.wikipedia.org/wiki/Multiverse#Level_IV:_Ultimate_Ensemble

I have my reasons for thinking its true, but since many of you already think

Tegmark is right, and since it would be too long to explain why I think it,

everything I write below will assume it. That will be the only assumption.

If that theory is true (which the rest of this writing will assume), then the

universe equals Chaitin's Constant, equals all of math, and has a Kolmogorov

Complexity of 0. The Kolmogorov Complexity of nothing (nonexistance) is also 0.

The universe simultaneously equals everything and nothing. Because nothing is

subset of of every other subset (in an analog way), the universe is a subset of

every part of itself, therefore the universe is a fractal. Its also a fractal in

the way that a Turing Machine can emulate an other Turing Machine to any depth.

An other way to say that is a multiverse of Turing Machines where each bit is

viewed as a sphere which changes depending on other spheres in the ways

described by quantum physics. Each quantum particle/wave/etc or combinations of

them are such a sphere. There can be deep indirect connections between any 2

bits, as we know from how software normally works.

There are many Chaitin's Constants, but I mean the kind for programs up to

infinity bits. The way I'm using it, the specific language of those bits will be

left as an abstraction. Because of the unsolvability of the halting-problem,

Chaitin's Constant is uncomputable. Multiverse branching happens because

Chaitin's Constant is uncomputable. It can't choose so it does both.

Because the universe is symmetric in every way (Kolmogorov Complexity 0), and

what we call reality is mostly not symmetric, what we call reality must be a

subset of something that is symmetric which could be combined with other subsets

to total a symmetric whole.

Because Kolmogorov Complexity 0 is the only property of the universe, there has

to be something about it that results in nonsymmetric parts being self aware

individually. The way Turing Machines work, many parts of the data have little

or no interaction with other data, so if such data calculated a mind, it might

think it is an individual and think its subset of the data is how the whole

Turing Machine works. Everywhere that data is copied or interacts with other

data in the Turing Machine, the statistical patterns it observed tend to be

accurate. It calls them "laws of physics".

Its not just 1 Turing Machine. Its an infinite multiverse of Turing Machines,

together forming some view of Chaitin's Constant. That smoothes the statistical

patterns. Data can flow between them and be in many at the same time.

It doesn't fit exactly, but the Buddhists describe the universe similarly,

called Indra's Net:

"Imagine a multidimensional spider's web in the early morning covered with dew

drops. And every dew drop contains the reflection of all the other dew drops.

And, in each reflected dew drop, the reflections of all the other dew drops in

that reflection. And so ad infinitum. That is the Buddhist conception of the

universe in an image." --Alan Watts

I've described the universe in a few ways. Now I'll explain how to use those

statistical patterns.

We know there are statistical patterns at least as complex as life forming from

quantum particles, because thats what happens here. The laws of physics are also

patterns. Space, time, mass, and energy are patterns that physics statistically

controls. Remember that we're assuming the universe equals all of math

(Kolmogorov Complexity 0) so that requires all such patterns be made of the same

stuff and have a peer-to-peer relationship instead of x created y created z.

Statistically x can appear to create y, but there can be no absolute law in the

universe that says it has to always happen that way. The only absolute law

(which we are assuming in this text) is the Kolmogorov Complexity of the

universe is 0 (universe equals all of math). If there was an other absolute law,

then the Kolmogorov Complexity would have to be higher to define it.

Quantum physics is a branch of statistics and some of its equations appear to be

closely related to bayesian networks, but I'm not an expert. If that is true or

not, bayesian networks are still a powerful statistics tool that scales up to

any number of statistical events you can store in the computer. It finds the

chances of any set of events when given the chances of other events. If no other

information is available about those events, then a bayesian network, when run

enough cycles to find a good network shape, always approaches the correct answer

with an accuracy higher than Humans can think. Humans normally win at such

statistics tests only because they do not put all the relevant information into

the bayesian network, because they have so many subconscious thoughts that they

do not know how to put into the computer. For pure math things like quantum

physics, bayesian networks, combined with various ways of inputting and

outputting the timing and context of events, will find many statistical

patterns.

Quantum physics research has focused too much on small size and short times. It

should be expanded to look for large statistical patterns that are the effects

of such small events.

Since the universe is a multiverse of an infinite number of Turing Machines and

quantum physics is the sphere view of those bits, we can use the statistical

properties of Turing Machines in to predict physics.

Usually Turing Machines generate mostly random appearing data but theres also

some weak patterns in it that can be detected by a Bayesian Network applied to

those bits.

Whatever patterns occur in the bits of those infinite number of Turing Machines

(the universe) are the only patterns in the universe, therefore physics and all

events are made of such patterns.

Such patterns occur in statistical clustering ways. If that was not true, what

we call reality would fly apart instantly.

The patterns in our brains are mostly built on top of the patterns called

physics, but statistical clustering of the other patterns will form between

people who tend to think the same things, and statistical clustering will happen

between people who drive on the same road often, and the same for all other

patterns. They all tend to stick together, like gravity for patterns. I don't

know enough physics to say if normal gravity is such statistical clustering or

if it is a pattern built on top of it at some depth.

In the context of an infinite multiverse of Turing Machines, splitting off a

variation of the multiverse to do calculations is trivial. It happens in every

quantum physics experiment. The hard part is getting them to get far enough

apart that they don't statistically cluster back together (collapse the

wavefunction). When I say statistical clustering of patterns, I mean it in a

fuzzy way, so there are continuous paths between collapsing a wavefunction and

being a separate multiverse. Together, all such things total to a symmetric

universe, a Kolmogorov Complexity of 0.

Up to now, this has been very theoretical, but it logically follows from Max

Tegmark's theory ("All structures that exist mathematically exist also

physically") if its true. Here's an example to support the "gravity for

patterns" theory. Its a video I recorded in 2002 after practicing mental

experiments for 5 months. I can't do this any time I want to, and not nearly

this much any more, but the important thing is this is a result of patterns in

my brain. I thought it to make it happen:

http://www.youtube.com/watch?v=9Ww8rI0BG9k There have always been people who had

strange mental abilities. Some were called witches and burned at the stake. Some

made it up. If the Kolmogorov Complexity of the universe is 0, then such mental

abilities are caused by statistical clustering of patterns at the same level

where the laws of physics formed as patterns. Its not magic. Its just something

thats hard to access and hard to learn.

Brains learn those quantum patterns and adjust their thinking a little as if

communication through neurons and axons and through those patterns was a normal

part of the brain. You can disagree on the amount that will happen, but it

logically follows from the universe being math that it would happen some amount.

Empirically (as you can see in the video), it happens a measurable amount.

I've explained why telepathy and other unusual mental abilities should be

expected if the universe equals math. My personal experience with it is not

needed to understand that, but it will help some people understand physics

better.

One of my long-term plans is to build a software that uses subtle psychology in

its audio, video, and mouse movement interactions with each user, to

statistically synchronize the subconscious minds of many people through the

internet. It will detect very small patterns in their psychology (like the

physics patterns in brains described above), consider previous interactions with

that person and how the patterns synchronized with other people's patterns

across the internet, and adjust the subtle psychology (accessed through mouse,

audio, video, etc) instantly (many times per second) to statistically cause many

peoples' thought patterns to flow together. For example, if it changes some

realtime music (like in Audivolv) on 2 peoples' computers and causes both of

them to move their mouse in the same directions at the same times, without those

people knowing why they move their mouse that way, then the thoughts are

synchronized some. That's not enough. When its working, disconnect the internet

communications between the 2 softwares, and see how long those 2 people continue

moving their mouse the same as each other (by reconnecting and comparing data

after the experiment). It can be done faster by simply keeping those 2 software

processes separate instead of disconnecting the internet connection. Either way,

the effect is those 2 people would learn to read each other's minds. Telepathy.

In my experience, it feels like mind over matter is done the same way as

telepathy, and the matter is just somebody else's brain. Its a very small

physical force even after months of practice, but I expect if its done more

accurately using Bayesian Networks and some preprocessing on the inputs and

outputs for timing, then it could be learned to a much higher skill level, and

we could build machines to do it too.

It can be done in machines by using the timing of network events as quantum data

(Did it come at an even or odd millisecond in this multiverse?) and sending Lisp

code in the network packets to build up random Lisp software at each computer in

the network. The Lisp software will observe the data coming in, keep a very

small state, and choose what data goes out. When it aligns with other

multiverses which have the same data and state at each node, the pasts and

futures from that multiverse will be the statistical influcence that allows this

to be used, in a very weak way, as a statistical alpha-point computer, which I

think is the same thing as telepathy but in a more digital way. Brains are more

analog so it works better. About my name here: the lispunit part means any

subset of the multiverse. In this case, its the part I'm currently experiencing.

After that is working, people would start to use it and it would spread quickly

until everyone on Earth who wanted to learn telepathy and other mental abilities

would learn to do it through the artificial intelligence network (Human AI Net)

which I am planning. In my experience, it will have no effect on those who do

not want to participate, but I expect that if enough people tried at the same

time to access the thoughts of one who did not want to participate, it would

work. I see society changing quickly in the last few years toward acting more

like 1 system instead of 7 billion individuals. I see power shifting away from

cental authorities and becoming more peer-to-peer (not just in computers, but in

everything). But I also expect that a small fraction of those with a lot of

power will refuse to let society change in those ways, and I expect they would

use their power to destroy things and cause more global problems near the time

such gradual changes in society would obsolete them. If we had enough people who

could read minds, then those few people could not dodge accountability and hide

behind government secrets. They would have to become less aggressive, unless it

was really necessary, with such telepaths observing. For the purpose of avoiding

World War 3, and to advance the way the Human species can communicate, we should

create such bayesian psychology software to create a global telepathy network

(which usually you can choose to join or not at any time, or any gradual level

between).

After its up and running, you don't need the computers to keep it running, but

they can be used to connect it to the internet through other kinds of artificial

intelligence.

SL3 is defined as "Nanotechnology, human-equivalent AI, minor intelligence

enhancement, uploading, total body revision, intergalactic exploration.

Extropians and transhumanists."

SL4 is defined as "The Singularity, Jupiter Brains, Powers, complete mental

revision, ultraintelligence, posthumanity, Alpha-Point computing, Apotheosis,

the total evaporation of "life as we know it." Singularitarians and not much

else."

The global telepathy network is a type of "powers" (even though we know how they

work) and is somewhere between a Turing Machine and "Alpha-Point computing"

(multiverse recursions). Its only a small amount alpha-point because it only

uses such multiverse recursions statistically. It doesn't get the return value

or influence them much. The global telepathy network would be categorized as an

SL4 event, regardless of how weak it is, because its the type of event instead

of the magnitude. I expect we will learn to use it in stronger ways eventually.

It will work because the Kolmogorov Complexity of the universe is 0, or as Max

Tegmark says it: "All structures that exist mathematically exist also

physically."

Also http://sourceforge.net/projects/humanainet (Human AI Net) will be an

Artificial General Intelligence (and a Friendly AI unless the core AI parts are

designed wrong). The global telepathy network will only be a plugin for it. I

separated the mad-science parts from the easier to understand parts. Do any of

you mad-scientists want to help with any of this?

**Next message:**Natasha Vita-More: "RE: [sl4] Simple friendliness: plan B for AI"**Previous message:**Luke Griffiths: "Re: [sl4] Simple friendliness: plan B for AI"**Messages sorted by:**[ date ] [ thread ] [ subject ] [ author ] [ attachment ]

*
This archive was generated by hypermail 2.1.5
: Wed Jul 17 2013 - 04:01:05 MDT
*