A Primer on Information Theories

 

Shannon’s Information Theory vs. Autosophy Information Theory

A simple question: “What exactly is information and data communication?” may be answered using two entirely different theories. The conventional Shannon information theory regards all information as "quantities", which are transmitted using bits and bytes in meaningless bit streams. This outdated theory is still being taught at our universities, even though no biological creature communicates in this way. A biologically correct Autosophy information theory, in contrast, regards all information as "addresses", which are communicated to create new knowledge in a receiver.

It may be hard to accept that our entire communications infrastructure and the programmed data processing computer are based on a false information theory. Correcting that mistake may lead to a true renaissance of all electronic communications media and the emergence of brain-like, no programming, error-proof  "Autosopher". Both the theoretical knowledge and the hardware are now becoming available for this next generation technology.

Conventional Shannon communication

This method of communication evolved in an age of primitive telegraph and telephone communication via noisy transmission lines or radio. Claude Shannon defined the theory in 1948 as: “A mathematical Theory of Communication.” This theory is still being taught in our universities even though it is obviously not the way biological creatures communicate. Shannon's theory treats all data items (ASCII character, pixel, or analog samples) as "quantities," to be converted into binary digits (bit and bytes) for transmission in meaningless bit streams. Most of today's communication problems can be traced back to the lingering effects of this outdated information theory.

 

 

 

 

 

 

 

 

 


Figure 1           Conventional Shannon data and Video Communication

According to Shannon's theory, a unit of "information" is a binary digit (bit). The purpose of the information is to "remove uncertainty" in the receiver, i.e. to improve the accuracy or "quality" of the reconstructed data items in the receiver. The more bits being transmitted the higher the data quality. Any attempt to remove bits from the transmission with data compression must lead to inevitable data distortions or loss of resolution. The bit rates are dependent on the data volume or the systems "hardware" without regard for human perception limits. Primitive lossless data compression schemes, such as the Huffman - Fano codes (also known as "entropy" coding), the Ziv-Lempel 77 and 78 compression, the Ziv Lempel Welch LZW codes, and V.42bis data compression in modems, are available based on misunderstood Autosophy - like methods.

In conventional video communications, for example, the transmission bit rates are determined by the system "hardware" such as the number of pixels on the monitor, the color resolution in bit/pixel and the scanning rates in frames/sec. The video images actually shown on the screen are irrelevant. Transmitting totally random noise (snow) images would require the same bit rates as blank images. According to Shannon's theory, information has no “meaning", which is a contradiction in terms. Information without meaning is not true information.

The programmed data processing computer

Data processing computers are essentially programmed adding or calculating machines that during the last two centuries evolved from mechanical calculators into electronic computers. However, it has long been realized that our own brains do not behave like calculating machines or computers. Our self-learning, self-organizing brains require no programming or supervision of internal functioning. Processes that are easy for a computer, such as numerical calculations, are very difficult for us. While, on the other hand, learning and pattern recognition, very easy for a human child, is exceedingly difficult for computers. Clearly, computers and our brains work according to entirely different principles. No matter how fast the processing, how complex the programming, or how large the memory a computers will always remain just a fast calculating machine. It will not evolve towards true Artificial Intelligence, only mere simulations.

 

 

 

 

 

 

 

 


Figure 2           The Programmed Data Processing Computer

A computer regards all input / output data as quantities according to the Shannon information theory. The input data is manipulated in the processor according to a stored program and supplied as useful output data to the user. All operations must be pre-programmed where all input data is treated in the same way. Such machines cannot find meaning or accumulate knowledge. The information stored in the memory is mere data, which does not have meaning without the retrieval software. No matter how much data is being processed or stored the computer itself will not become more intelligent.

A new Autosophy communication paradigm

Autosophy communication methods transmit data content or "meaning" with address codes, called "tip", where each tip transmission may represent any amount of data. Autosophy methods can provide very high "lossless" data compression and built-in unbreakable encryption. Most problems in conventional Shannon communication systems can be resolved by redesigning the systems according to the Autosophy information theory. Reference: Autosophy: an alternative vision for satellite communication, compression and archiving.

 

 

 

 

 

 

 


Figure 3           Autosophy Data and Video Communication

The Autosophy information theory regards all data items as “addresses”, which have "meaning". Communication is with address codes, called "tip", each of which can represent any amount of data. Communication involves a transmitter, which has knowledge, transmitting codes to a receiver for the purpose of "creating knowledge" in the receiver, in effect teaching it something. Communication is therefore a means of exchanging knowledge (engram). "Information" is only that which creates new knowledge in the receiver, in effect only that what the receiver does not already know and which is perceptible to the receiver. Knowledge is accumulated in learning hyperspace libraries in which every data pattern is stored only once. This is because one cannot learn what one already knows. The efficiency of the communication increases with the amount of knowledge stored in the hyperspace libraries. Children, for example, which have little knowledge, communicate less efficiently than experts, which share much knowledge. Communication between children is much less efficient than communication between experts. Reference: Autosophy information theory provides lossless data and video compression based on the data content.

Hyperspace knowledge libraries grow like data trees or data crystals from the input data in electronic memories without programming or outside supervision. There are presently seven known classes of "Infinite Dimensional Networks" each providing a separate learning mode. Klaus Holtz invented the learning hyperspace data networks in 1974. Reference: Patent 4,366,551.

 

 

 

 

 

 

 

 

 

 


Figure 4           Autosophy Video transmission

The code rate in Autosophy video communication is determined by the video content (motion and complexity) where a totally random noise image would require excessive code rates while a static or blank video image would require no transmissions at all. The left image shows the pixels required for conventional Shannon television. The right image shows the codes transmitted in Autosophy television, essentially movement and complexity. The transmission is using universally compatible 64bit codes each specifying a cluster of up to 16 full color pixels.

The same 64bit codes can also transmit other data, such as sound, text, still images or random bit files. The transmission is especially suited to the Internet's intermittent packet stream to avoid its Quality of Service (QoS) problems. There cannot be any fixed "compression ratio" calculation. Compression is: the hardware (the product of the number of pixels on the screen, bit per pixel, and scanning rate) divided by content (motion and complexity). Compression is approximately the colored pixels in the left image divided by the colored pixels in the right image. Reference: Autosophy data and image compression with encryption.

Self-learning brain-like Autosopher

The great visionary and science fiction writer Arthur C. Clarke in the 1960’s envisioned a talking, self-learning, self-aware computer HAL (2001 A Space Odyssey), which was bragging that no 9000 computer had ever failed or supplied erroneous information. Arthur Clarke's vision of self-assembling and self-replicating data networks was amazingly accurate. He wrote: "In the 1980's Minsky and Good had shown how neural networks could be generated automatically self-replicated in accordance with any arbitrary learning program. Artificial brains could be grown by a process strikingly analogous to the development of a human brain. In any given case, the precise details would never be known, and even if they were, they would be millions of times too complex for human understanding". Why don't we have HAL like computers by now? Do we lack the knowledge, the hardware, or just the will to build such systems?

On April 15, 1977 Klaus Holtz published a paper "Here comes the brain-like, self-learning, no-programming, computer of the future" at THE FIRST WEST COAST COMPUTER FAIRE in San Francisco. A Patent 4,366,551 for self-learning networks was filed in 1975. Both publications already described several self-learning data networks based on a newly evolving "Autosophy" information theory. This means, that in the early 1970's we had a choice of two paradigms, the programmed data processing computer or self-learning brain-like machines. Unfortunately, we chose the wrong path to develop faster and faster computers, with ever more complex software and operating systems, instead of brain-like self-organizing and failure-proof systems. Correcting that mistake may cause a true renaissance in computing towards self-learning intelligent robots and eventually to true Artificial Intelligence. Reference: Replacing the Data Processing Computer with Brain-like Machines.

 

 

 

 

 

 

 

 

 


Figure 5           The Self-learning Autosopher

A self-learning Autosopher system behaves like a sealed "black box" to organize its own internal operations without programming or outside supervision. The input data, regarded as addresses, select or create their own internal storage locations in a special memory. The data is stored in a mathematical hyperspace format in which the more data is already stored the fewer memory locations are required to store additional data. This provides for enormous storage compression for very large multimedia archives. Reference: Autosophy Failure-proof Multimedia Archiving.

Primitive software driven application may use the normal Random Addressable Memories (RAM) in a computer. Future large scale archiving systems should use a new memory paradigm known as the Content Addressable Read Only Memory (CAROM) or a self-healing error-proof memory configuration known as Dual Entry Content Addressable Memory (DECAM). These new memory devices are: self-organizing, self-learning, self-repairing, self-healing, and even self-replicating. The memory devices will not fail or provide erroneous information even after suffering severe physical damage. The memories may be built in sealed modules containing a roll of stainless steel foil, about the size of a roll of toilet paper. Because the data is stored in hyperspace nodes, which may be anywhere within the memory spool, memory repair is far beyond human intelligence. Locating and repairing a defective hyperspace node in a foil spool is just as impossible as locating and repairing a specific neuron in the human brain. The only alternative is to use self-repair and self-healing in the failure-proof DECAM memory configuration. Reference: Self-organizing and self-repairing Mass Memories for Autosophy Multimedia archiving Systems. Also try Patent 5,576,985.

Summary and conclusions

The Autosophy information theory may soon replace most applications based on the outdated Shannon information theory. This may provide orders of magnitude improvements in all forms of communication and replace the programmed data processing computer with brain-like no-programming Autosopher. Much research and development will be required, but the effort will be well worth it. It could cause a rebirth of our entire communications and computing industry. Whenever a new theory or technology is introduced it often produces many unforeseen applications opening the way to entirely new applications. Societies will, however, accept new inventions only if they have to.

Using conventional Shannon solutions such as: more and more bandwidth, faster and faster processors, or more and more complex operating systems and software, will face an uphill struggle. Each small step forward will require more and more effort and more research money. Autosophy, in contrast, could solve most computing and communications problems at the same time, because it is the biologically correct theory. We really do need that new technology now. The table below summarizes the important differences between the old Shannon technology and the future Autosophy technology.

Information Theories

Shannon

Autosophy

Input/output data used as:

Quantities

Addresses

The data is used for:

Arithmetic

Learning

Communication is with:

Meaningless bit streams

Universal 64bit content codes

Best transmission media is:

Fixed bandwidth channels

Packet switching Internet

Bit rates are determined by:

Symbol volume (ASCII, pixels)

Content (motion, complexity)

Most efficient data type is:

Meaningless bit files

Perceptible (sound, video)

Purpose of information is:

Remove uncertainty

Create knowledge (teaching)

Data compression is always:

Lossy (causing data distortions)

Lossless - perceptibly lossless

Encryption is using:

Pseudo random numbers

Custom hyperspace libraries

Archive data storage is:

Linear

Hyperspace saturating

Best memory devices are:

Random Addressable RAM

Content addressable CAROM

Memory reliability:

Single point system failure

Healing failure-proof DECAM

Software is generated by:

Programming

Education (like children)

The scientific method uses:

Equations

Algorithms

The resulting devices are:

Programmed computer

Autosopher (brain-like)

Best system architecture is:

Platform-centric (mainframe)

Network-centric (Internet)