Knowledge is undirected
We started out with a simple idea – to make engineering knowledge as flexible as the way a person stores and uses it. If we take Ohm’s Law, it relates voltage, current and resistance as a relation which can be traversed in any direction, not as an equation R = E/I, and to meld logic, and …
The same relation can be used many times – systems can have thousands of resistors, all interconnected. Add multipliers, logical switches, flow switches, and a system can become hard to think about.
This led us to a network formalism, where the direction of traversal is initially undefined – knowledge is undirected.
So we came up with
This may not seem like much, except that the structure can be used to work out the value of any numerical variable knowing the other numerical values and logical states, or a logical state knowing the numerical values – the structure is undirected as to purpose. It emphasises the concept of a logical surface on which knowledge is written. This allowed all of decision-making using numbers and logic to be represented in a dynamically extensible way. It also avoids all the problems of an algorithm – the structure can grow and change itself.
But a natural language is different – relations aren’t just operator boxes like PLUS, but are both objects that can be talked about (and linked to) and relations – the formalism grew more flexible and able to represent more complex, many-layered knowledge.
As well as being able to say “He can swim”, the text itself can have its existence controlled, so we had to add existential logic control to each relation.
We spend our days working out how (not whether) an active, undirected hyperspace network can represent all of natural language and the knowledge it describes.
Here is one idea – the Jacquard loom
It is a machine that uses a program to weave a structure.
An obvious analogy is to use a structure to weave a structure, making it self-extensible
The idea is simple, realising it is not.
We rapidly ran up against the (very severe) limit on conscious processing, which prompted us to think about how we could break that limit.