KNOWLEDGE COMPUTER

 The idea of the knowledge computer is that the unit of calculation is a knowledge unit.  In a digital computer the unit of calculation is a bit, or binary digit, a “0" or a “1".  These two numbers can be added, subtracted and all the other mathematical operations.  They can also be expanded to sequences of 0’s and 1's to form larger numbers using the binary code to the base 2.  One advantage of this binary system is that any mechanical device that can have two states can represent the 0 in one state and the 1 in the other state, such as “on” and “off” of magnetic “North” and magnetic “South”.  The next step is to do logic operations such as “and”, “or” and “not”.  This can be done with a binary code, and allows computers to do logic operations in mathematical form in binary code.  Storage and retrieval are a matter of writing and reading storage media, and doing operations on retrieved data in a central processing unit, and displaying the results in a form humans can understand, such as a computer screen or a punch tape.

 The knowledge computer is fundamentally different.  Its main unit is a unit of knowledge.  Knowledge units, defined below, are stored in a four dimensional descriptive form.  Knowledge units are searched and retrieved, not by an address, but by their initial condition and prime attribute.  Retrieval is like asking for something that begins in this condition, changes in certain ways and results in something that has that attribute.

 A knowledge unit is a description of how to cause change.  It begins with an initial condition, describes what to do to change the initial condition by organizing its parts into a new whole that has the prime attribute, and ends with a description of the prime attribute.  When searching for a knowledge unit, the search can be either for knowledge units that have that initial condition, or that prime attribute or both.

 Knowledge units can be linked together to form steps in a solution process.  Each knowledge unit takes the process one step.  The prime attribute of one knowledge unit becomes part of the Initial condition of the next knowledge unit.  The process continues step-by-step until the ultimate goal (prime attribute) is reached.

 How can such a knowledge unit be represented?  The solution presented here involves three sets of rules, which presumably (big word) describe everything in the universe.  These are the rules of the container (space-time), the rules of the substance contained (energy) and the rules of organization (based on category structure).

 The space-time signature of an object is the size and shape of the space it occupies at any instant.  A three dimensional representation of an object at an instant or unit of time would be the three-dimensional space it occupies, its size and shape during that time.  A four dimensional description would be the size three dimensional size and shape, including any changes in size and shape over a sequence of units of time.  A knowledge unit is a description of how to cause change.  A knowledge unit’s space-time signature is a description of its size and shape and how they change over time as the change occurs from initial condition to emergence of the prime attribute.

 A knowledge unit has an energy signature.  The parts of the initial condition are made of various forms of energy.  Each form of energy is represented by a color.  There are many forms of energy, and many colors in the spectrum, as well as many combinations of colors.  There are rules about how a form of energy can interact with other forms of energy, for example a glass bottle can contain a gas at certain temperatures and pressures.  The glass bottle is a form of energy and the gas inside the bottle is a form of energy as well as the air outside the bottle.

 A knowledge unit has an organizational signature.  The parts in the initial condition are organized in a certain way to form the new prime whole which has the prime attribute.  The prime attribute does not exist, except as an attribute of the whole organized this way.  If the parts are put together to form the whole organized this way, the prime attribute emerges.  If the whole is broken down into its parts again, the prime attribute ceases to exist.  The wetness of water at room temperature is an example.  Two dry gases at room temperature, hydrogen and oxygen combine with a certain organization to form a water molecule and the attribute of wetness emerges.  If the water is broken down again into its constituent parts, the attribute of wetness ceases to exist, and two dry gases remain.

 How shall a knowledge unit be represented in a computer, which allows logic operations, storage and retrieval?  The parts of the knowledge unit are represented in terms of their signatures in the three sets of rules.  The space-time signature is represented by the space-time occupied by the parts and whole and attribute.  The forms of energy of the parts are represented by colors.  The organizational structure is represented by how the parts are organized to form the new whole.  Only the surface of an object is represented.  Its interior is irrelevant.  If its interior is relevant to the process of changing the parts into the prime whole and attribute, then smaller parts of it are used.  Only molar parts are used, that is only parts that have a relevant attribute as a whole.  Then their parts or interiors are irrelevant and a representation of the surface only will capture all the relevant parts for the change process in the knowledge unit.

 A part of a knowledge unit looks like a transparent, colored shape in relations to other transparent colored shapes.  This is like transparent soap bubbles of different shapes that change over time in shape and relation to each other (space-time data) and are different colors (energy type data).  The way the soap bubbles are related to each other over time gives their organization (category structure).  The result is the prime entity which has the prime attribute.

 Such a knowledge unit described in transparent “soap bubbles” of different colors and shapes can be stored and processed as light of different colors.  Recent work in data storage in a holographic medium has suggested that a knowledge computer is possible.

 Knowledge units might be stored in a holographic storage medium.  Retrieval might be accomplished by pattern comparison with light images of objects described in terms of the three sets of rules.  These would appear as different colored shapes organized in certain ways and producing over time a prime attribute.

 The May 2000, edition of Scientific American included an article by Jon William Toigo entitled Avoiding A Data Crunch.  The article was focusing on digital data storage, but knowledge unit storage could be possible using these methods.  The article stated:

 “Each of these pages of data is stored in the form of an optical-interference pattern within a photosensitive crystal or polymer material.  The pages are written into the material, one after another, using two laser beams.  One of them, known as the object or signal beam, is imprinted with a page of data to be stored when it shines through a liquid-crystal like screen known as a spatial-light modulator.  The screen displays the page of data as a pattern of clear and opaque squares that resembles a cross word puzzle.
 “A hologram of that page is created when the object beam meets the second beam, known as the reference beam, and the two beams interfere with each other inside the photosensitive recording material.   Depending on what the recording material is made of, the optical-interference pattern is imprinted as the result of physical or chemical changes in the material.  The pattern is imprinted throughout the material as variations in the refractive index, the light absorption properties or the thickness of the photosensitive material.
 “When this stored interference pattern is illuminated with either of the two original beams, it diffracts the light so as to reconstruct the other beam used to produce the pattern originally.  Thus, illuminating the material with the reference beam re-creates the object beam, with its imprinted page of data.  It is then a relatively simple matter to detect the data pattern with a solid state camera chip, similar to those used in modern digital video cameras.  The data from the chip are interpreted and forwarded to the computer as a stream of digital information.
 “Researchers put many different interference patterns, each corresponding to a different page of data, in the same material.  They separate pages either by varying the angle between the object and the reference beams or by changing the laser wavelength”, p. 70.

 Storage media may include photographic films and even solid crystals made of lithium niobate and other inorganic substances and photoreflective, photochromatic and photochemical polymers.  No media yet have yielded the mix of performance, capacity and price to be commercially successful.

  “After data are stored to a holographic medium, a single desired data page can be projected that will reconstruct all reference beams, for similarly patterned data stored in the media.  The intensity of each reference beam indicates the degree to which the corresponding stored data pattern matches the desired data page.
 “‘Today we search for data on a disk by its sector address, not by the content of the data.  ‘Coufal explains’.  ‘We go to an address and bring information in and compare it with other patterns.  With holographic storage, you could compare data optically without ever having to retrieve it.  When searching large databases you would be immediately directed to the best matches’”, p. 71.

 This suggests that if a four dimensional image of the colored surfaces of parts and wholes in a knowledge unit could be stored in a hologram, and pattern matching could be used to retrieve similar initial conditions, solutions, prime entities and prime attributes, then a knowledge computer would in principle be feasible.  If one had a set of initial conditions and one wanted to change them to produce a certain prime attribute, all knowledge units with similar initial conditions and prime attributes could be retrieved.  The best matches would have the most intense reference beam.

 Then the reference beam could be used to recover the signal or object beam with the information on it, such as a page of data, or the description of a change process that would accomplish the desired change.  In this way a knowledge computer is a “how to” computer.  It stores solutions, and is searchable, so one can find the solution to go from the initial conditions to the prime whole and prime attribute.  Even processes of several steps could be stored and retrieved.

 The theory of this type of computer is that when knowledge of how to cause change is stored in knowledge units, which may be linked together to produce a series of changes that results in an ultimate prime attribute, then many things can be accomplished.  First a single knowledge unit can be retrieved and used to accomplish the change.   Secondly an initial condition with a remote prime entity (requiring two or more stored knowledge units to accomplish when linked together) could be entered, and the knowledge computer could make a sequences search.  First it would produce a change from the initial condition.  Then it would take the prime entity and prime attribute as the initial condition and the ultimate or target prime attribute as search for a match.  By trying many sequences, it might find a path or set of linked knowledge units that would accomplish the desired change step by reliable step.

 This is similar to the way a mathematical or logical proof is done.  One starts at the beginning and works step by step toward the end.  Sometimes one starts with the end and works back toward the beginning, making a bridge moving out from each end.  When the final piece is in place, the bridge is complete and one has a description of how to go step-by-step from initial conditions to ultimate prime attribute.

 In this way the knowledge computer can search all its knowledge units to find a step-by-step solution path to accomplish the goal programmed into it from the initial conditions given it.  In addition the knowledge computer can determine when it has all the steps a certain way across the bridge from one end, and from the other end, but is missing some steps in the middle.  When these missing steps, or one of them is identified, it can be put into place.  If the knowledge computer included all world knowledge, and a person somewhere in the world contributed a new knowledge unit (a new way to cause a certain change), that person could get credit for it, the new knowledge unit would fit into at least one requested change sequence, and the creator would be acknowledged and rewarded for contributing something new to human knowledge.

 Other types of processing could be developed for the knowledge computer.  Above are suggested pattern matching for retrieval of existing knowledge of how to cause a certain change, and perhaps a method for finding paths of change sequences that would lead step by step through several knowledge units from initial conditions to a final goal.  What else could a knowledge computer do?  The knowledge computer could create new knowledge.  Variations on organization of parts could be done to develop new emergent prime attributes.  These could be tested in the knowledge computer to see how they would interact with various objects to develop a set of emergent attributes for new prime wholes.  Much creative computing could be done this way.  By combining path searching with creative computing, novel solutions to existing problems could be found.  Some solutions could be found in principle, even if they never existed physically before.  They could be created and tested without ever existing physically.  Then the best new solution could be built physically to see if it really worked that way.

 This is the essence of the efficient rise of complexity using symbolic knowledge in the age of knowledge evolution beyond genetic evolution, where a succession of generations of the new object must be physically built and tested by natural selection.  The genetic way is old, slow and wasteful of energy resources compared to the knowledge evolution way using symbolic knowledge to try concepts out before committing a large amount of time and energy to them.

 As human knowledge grows, more humans could solve more and more problems by retrieving and using all relevant human knowledge on any subject.  The knowledge computer makes this possible.

 The knowledge computer is not dependent on any human language.  Its basic units are stored in representations of physical quantities, which presumably (that big word again) are the same everywhere in the universe.  In this way a knowledge computer, with a suitable front end to communicate with humans, could be used by any human of any language back ground.  It would be similar to hearing music.  However, other non-human, intelligent species could understand it too, because it would be a faithful representation of physical quantities, if they could understand the color code for forms of energy.

 A question arises whether the study of the forms of energy would result in our ability to define any attribute at any level in terms of forms of energy.  An attribute, such as a prime attribute, is meaningful and relevant, because it causes a change in some target entity.  Light from the red of an apple causes certain chemical changes in cells in the retina, which is part of the process by which we “see” the apple.  We say the apple has the attribute of red color, but we experienced that through a series of physical interactions.  Therefore an attribute should be describable in terms of the physical changes it causes in target entities.

 If this is true, then colors and shapes should be enough to store all the data of a knowledge unit.  If not, some method of defining attributes and their effect on objects would be necessary.  However, if all is one substance (energy) in the macro world, just in different forms which interact with each other in characteristic ways, perhaps attributes can be defined in terms of the interaction of forms of energy.  Does that interaction produce still other forms of energy and changes which are the prime attributes?  Probably yes.  If so, then this representation system (colors for energy forms in shapes in space-time, organized in certain ways) should work to describe objects, relationships and attributes.

 In terms of category structure a form of energy in space-time can be either an object or an attribute, but not both in the same category.  The category of all red things contains members which are things (objects) which have the attribute of redness (attribute).  Red color can also be an object or a member of a category.  The category “colors” includes red, yellow and blue, for example, which have the attribute of being electromagnetic radiation within a certain range of energy levels or frequencies.  The colors are certain forms of energy in certain amounts.  In this category the colors are members or entities or “objects”.  They possess attributes which allow them to be aggregated together in the category “colors”, because they all meet the category membership definition and everything else in the universe does not.  In principle anything, certainly any form of energy, can be either a member or an attribute, but not both in the same category.  Any form of energy can be an object (member of a category) or and attribute of a member of a category, which is included in the membership definition, but not both in the same category.

 In a particular knowledge unit a form of energy will function either as an object or as an attribute, but not both.  I believe this to be true, but it would have to be proved.  If knowledge units are limited to being represented by arcons and molecons (objects, relationships and attributes of objects and relationships), then it is true that the same energy form cannot be both an arcon and an attribute of the arcon.  Similarly a rule cannot be both a relationship between two arcons and an attribute of a relationship.  The rules of organization are limited by the rules of category structure.

 The light from the red apple shines on certain cells in his retina and changes them.  The apple is the object.  Red is the attribute.  The apple possess the red color.  The red color is a property of the apple.  Take the apple away and the red color will no longer be there.  Take the red color away, for example by peeling the apple, and the apple is still there, but now has a different color associated with it.

 It is important to keep objects and their attributes distinct.  The initial conditions of a knowledge unit are generally object descriptions, which include attributes.  The goal of the knowledge unit is to change the initial conditions in such a way that a new organized whole (the prime whole) is created which possesses the prime attribute.  Only when the prime attribute has been produced is the solution process complete.

 One interesting question occurs, if the hologram storage is to be used for a human knowledge computer on Earth.  How could a copy of the crystal or other storage medium be made, so that a daughter knowledge civilization could receive a complete copy of the mother knowledge civilization’s knowledge base?  If two or more knowledge civilizations came together to form a new knowledge civilization which had their combined knowledge, how would such a combined knowledge base be made, and how would a copy of it be made to give to the new knowledge civilization.  Indeed the several parent civilizations would want a copy too.  I believe that something more fundamental than hologram storage will be developed.  However building a knowledge computer using knowledge units stored in hologram form would be a good place to begin to create this type of knowledge computer.

 How does the knowledge computer do operations?  The key question always is “Will a step cause the desired change?”  The three sets of rules are programmed into the knowledge computer so that a proposed new step must follow all the rules.  If it does and the desired change occurs, then a copy of the change can be made as a new knowledge unit.  If the change violates any of the rules, it cannot occur in the physical world and is rejected.  Changes in forms of energy within certain space-time limitations and organizational limitations can be checked against what is possible within the rules.  If a change involves one form of energy passing through another form of energy which is a container for the first form, the change will not work.  The knowledge computer would have to find some other solution.  The CPU of the knowledge computer would act like a logician looking for steps that would work to bridge the gap between the initial conditions and the prime attribute with steps that would work.  At each step it would search its memory to find a solution that would take it closer.

 Evaluation algorithms could be used to speed its decisions about whether to pursue this path or try something else.  The story of the way the man solved Fermat’s Last Theorem is illustrative of how a knowledge computer would work, but it would have all its memory of solution steps available for finding a solution to each step.

 It may be that the knowledge computer will have to have a goal setting algorithm to set interim goals toward which it can aim for the solution to the next step in a long chain of solutions to the ultimate prime attribute goal.  Newell and Simon’s work with General Problem Solver and the Understand program may be helpful here.  The value system of a mathematician or logician would be helpful.  Seeing a possible solution path and pushing it as far as possible is a good strategy.  Looking at solution paths starting from both ends and meeting on common ground in the middle is another good strategy.  These strategies could be programmed into the knowledge computer and then evaluated by the computer depending on what was available in its memory.  Different solution paths could be tried until a complete one was found, or the shortest one was found.  Then a goal could be set to extend one side of the bridge a certain way, and that solution sought.  A list of solutions to be sought could be kept.

 When someone added a knowledge unit to the knowledge base that completed that step, credit could be given to the discoverer.  Then the knowledge computer could then set the goal for the next step toward the solution path for that problem.

 It may be that there could be many CPU’s operating out of one central memory.  Many could read the memory, but only one or a few could write to the memory, or the knowledge computer would automatically write to its memory whenever it found a new knowledge unit that worked, i.e. followed all the rules and successfully produced the prime attribute from the initial conditions.  What a wonderful democratizing object the knowledge computer would be, if everyone had access to all its knowledge, and to its creativity simultaneously.

Cataloging Forms Of Energy And Their Interactions And Attributes

 One of the many areas that needs to be developed for the knowledge computer to reach its full potential is a much deeper understanding of forms of energy.  What they are, what attributes they have under various circumstances, how they interact with other forms of energy, what changes space-time limitations have, and what happens when they are combined with other shapes of forms of energy?  All these issue need to be studied.  I expect that when we do, we will find some general rules that apply to forms of energy of which we are unaware now.  These rules will also help us understand why attributes emerge as they do.  Predictive power from this knowledge will help guide interim goal setting algorithms when an ultimate goal is known, and several steps are needed to produce it.

 This is a separate study all by itself, probably a new academic discipline, or sub-discipline within physics.  What are the forms of energy, how are they produced, how are they changed, how do they interact with other forms of energy, and what universal rules can we discover about forms of energy?  Some might say this is physics, but what is sought here is a catalog of these answers for use in a knowledge computer.  For example what forms of energy and interactions are possible within a certain energy range?  Does variation in energy available for interaction determine energy forms to be formed and attributes to emerge in some rule governed way?  Does a linear increase in energy produce reliable forms of energy, objects and emergent attributes at certain points along the way?  Why do new attributes emerge, and why in the way that they do?  What role do space-time and organization have on emergence of attributes and interaction of attributes at certain levels of organization?

 If a knowledge computer could represent a form of energy as a colored shape related to other colored shapes in certain ways that change over time to produce the prime attribute (and how would it be represented?), then the search techniques of the hologram storage system could work to allow massive parallel searches of knowledge units in a knowledge base (a crystal or set of crystals).  The color of the shape would have certain interaction characteristics with a certain other colored shape.  The energy interaction would be rule governed, so that the knowledge computer would reject any interaction that violated the rules.  Solid shapes could not pass through walls, or defy gravity.  All the physical rules would have to be respected.  The rule checking program would be extensive.  Unless some clever way could be developed to do it by color comparison, the rule checking process could slow the knowledge computer down dramatically.  Energy intensity or energy amount in a shape might be color coded; e.g. more energy, higher frequency; lower energy, lower frequency.  General rules comparing colors might speed the process.  If the other object in the proposed interaction has a certain range of color, the interaction will not work.  Perhaps colors could represent forms of energy, so that a form of energy with a higher frequency color could not penetrate a form of energy with a lower frequency color, or vice versa.

 The study of forms of energy with this understanding in mind should be very fruitful in producing new human understanding.  Hopefully that new understanding will be useful in making a knowledge computer.  The key is being able to represent objects and relationships within the three sets of rules and do operations on those representations.

Further Reflections On The Knowledge Computer

5-19-2000 jb What will the colors represent, and what variables can the light carry?  Light variables include hue, intensity, phase, and perhaps others.  What parameters of energy should be included, if time and space are represented by sequential variations in shape.  The space-time rules, the rules of the container must apply to the sequence of shapes.

 The color of the shapes, or light characteristics carry the energy information.  This energy information shows the form of energy and its state(temperature; gas, liquid, solid or plasma; light absorption or reflectivity, etc.)

 Determining what characteristics of energy are, will be a make advance in science.  It may lead to an understanding of why and how new attributes emerge from new organized wholes made up of smaller parts of energy objects, organized in a certain way.  Why are hydrogen and oxygen gasses at room temperature?  Why does the combination of these two dry gasses into a water molecule form into a liquid at the same temperature?  Why is gold gold color, silver silver color and most metals metallic color and sulfur yellow and some elements white powder?  What energy changes produce these attribute changes?  Can these attributes be defined in terms of these energy changes?  Can the knowledge computer only show energy attributes and have them describe attributes including the prime attribute in energy terms?  Can the color gold be expressed in energy terms describable as light, which can be searched by a reference beam?

 The thrust of the knowledge computer in hologram form is that the rules of the container (space-time) are contained in the sequence of shapes described as surfaces of whole objects with attributes as a whole (arcons).  The space-time relationships of the shapes shows the space-time relationships of the shapes to each other.

 The rules of the substance contained (energy) are expressed in light which forms the surface of the shape.  Think of a see-through glass Christmas tree ornament that has a slight color in the glass.  This is how the energy attributes of a shape are represented in the hologram.

 The rules of organization are represented by the way shapes are organized together.  The water molecule is represented by two hydrogen atoms forming a triangle with an oxygen atom, with the angle H O H being 105 degrees.  There may be a way to represent a water molecule as a whole as a  colored shape.  The sequence would begin with the separate parts, flow into the parts being still separate, but organized together in a certain way, and then the new prime whole would be represented as a single shape with its molar attributes.

 What would a prime whole’s light signature be to describe its attributes as a whole?  How would you distinguish the prime attribute from its other attributes as a whole (rels: relevant molar (as a whole) attributes).  This is the problem of being able to describe emergent attributes at their unique level.  Everything in the knowledge unit (part level and whole level) is relevant to the prime attribute.  Everything that does not change in the process of going from the parts in the initial condition to the formation of the prime whole with the prime attribute is irrelevant to the knowledge unit. In order for the relevance issues to be managed, the prime attribute must be describable and its presence or absence determined.  Some methods of representation of new, unique attributes, such as the wetness of water, or gold color must be developed.

 How does one describe emergent attributes in terms of energy?  That is the probable path to success in developing this part of the knowledge computer.  An answer seems to be in the interaction of the prime attribute with another entity to change the other entity.  Once we discover how to describe attributes, can we use the knowledge computer or other computers to predict what new attributes would be like.

 This would be a powerful new tool for the rise of complexity through the application of symbolic knowledge.  The place to begin is the understanding of the attributes of energy itself, and its various forms and how they can vary.  Then combine smaller wholes (chunks of energy) into to larger wholes and explain the emergent attributes of the larger wholes in terms of the organization of the energy attributes of the parts organized in that way, or how new forms of energy are produced, which have the emergent attributes.  This understanding of fundamental physics may make it easier to describe attributes in terms of color and shape in a knowledge computer using hologram storage and processing.

 This is a vast, new, very important field of study in physics.  Some scientists at the turn of the last two centuries said there was nothing more fundamental to discover in science.  This answers those claims.  Those who think there is nothing more to discover in fundamental science stand beside an ocean of ignorance with their eyes shut.