Network of Functions


Document the world's entities like we document the functions of code, and build their network.


How the world works can be understood by viewing at all things as functions, and building their network. For example, a shop nearby takes in electricity, people's work, monetary inputs from customers, etc., and returns products to customers, and a number of other "side effects". A shop outside, thus, is a function with I/O.

The same applies to companies. A company takes something in (e.g., natural resources), and returns something else (e.g., products, or their parts).

This is also true to schools, hospitals, etc. A hospital takes in sick people, and returns healthy people.

There are many other examples of functions, e.g., a mobile phone takes in electric power, and input, voice gestures, microwave radiation, etc., and returns things like differences in pixel luminosities, frequencies of speaker vibrations, other electromagnetic impulses and so on., even based on the input of code.

The very surface of a stone nearby you takes in photons of one type of characteristics, and returns photons of different type of characteristics.

Anything that interacts can be thought of as a function.

The idea would be to document the world, as if the world were to be build out of functions.

Hopefully, this would help us really understand how the world works, and what is possible, in terms of abstract state spaces. This is actually related closely to the previous idea of Technology Maps ™ on Halfbakery, and inspired by thinking that in fact, the idea of "Function Networks" (FNs) is much broader than the idea of "Neural Networks" (NNs), in a sense that the NNs are composed of narrower set of possible functions. For example, mostly linear functions, and with a couple of non-linear ones as activation functions, and thinking of -- what if we would simply allow all the functions, in the most general sense, and built a network of them.

In fact, this extends quite broadly, and branches very fast, because -- every website, every protein, every lab and institution is a function too, and this type of documentation of the world would best be done by a collaborative effort, perhaps starting with the cooperative open project by world's nations, open knowledge organizations, and largest search engines, taking into account the fact, that openness needs to be with wise constraints.

We had some progress with websites in fact, in a sense, that recently websites had increasingly taken on to implement their APIs to document themselves.

However, most of the world remains remains undocumented.


(suppress notifications) (Optional) Please, log in.

I like this idea. Could be combined with business cloud. One of my ideas:

So, thinking along these lines, I tried to abstract the things that we collect here at 0oo. Basically:

  1. Category: Function class, e.g., Goal, Category, Question
  2. Method: Function prototype, e.g., Idea, Invention, Transformation
  3. System: Function instance, e.g., Plan, Project, Agent, Organization, Team, Person, Equipment, Tool, Resource, Instrument
  4. Task: Function call, e.g., Task, Request, Order.
  5. Place: Function parameter, e.g., Location, Address, Account.
  6. Result: Function response, e.g., System Log, Event, Report, Executed task, Operation, Work result, Demo, Transfer, Transaction, Log, Blog post, Press release, Product, Service deployed.

It seems, that this corresponds well to the established concepts in CS:

2. OPERATORS (i.e., functions)
5. OPERANDS (i.e., parameters)

    : Mindey
    :  -- 
    :  -- 


Apparently, OPERAND and VALUE, i.e., the 5th and 6th merge into one, as they are very similar, and those who think that a relation is not an object, are mistaken.

// OPERAND and VALUE, i.e., the 5th and 6th merge into one //

Doing that would mean not distinguishing between a "variable" (a place for value) and the "value" (thing itself). So, perhaps it makes sense to stay separate.

Thats an excellent idea. Thats the premis of what im doing. Ill create another thread explaining my approach. But yes, everything can be modelled with funcs. Funcs and differential equations are equivalent. Quantum mechanics can be formulated thru funcs. Our programming languages and qm are very similar, imo.

    :  -- 
    : Mindey
    :  -- 


It can be abstracted in different ways. Relationships can be modelled as particle entanglements, funcs, analagous to qm. Reactive funcs constraining the state. But then, a constraining func can be represented as an object, with props as two particle refs, and a symbolic operand. Thats a descriptor of a link in graph. Now we got graph, and we can build and manipulate it graphically, in a low code development system

Of course, there can be relationships between systems of particles, like many to many and one to many and all that. They are all entanglements. Any math experts? I need to classify a complete set of operands, to represent all ways to write constraining funcs, atomic style.

    :  -- 
    : Mindey
    :  -- 


// Any math experts? I need to classify a complete set of operands [..] atomic style

I think you'd benefit from experts in logic and symbolic reasoning, experts in ontologies and knowledge engineering, not so much from mathematicians.

[skihappy], I get a feeling that what comes to your attention, gets through the "prism" of your system that you're building, but you're not explicitly mentioning it, so readers won't understand what you're talking, exactly... so, could you at least mention subjects in your sentences explicitly, for example, saying those few sentences, like:

"In order to construct a knowledge graph that has (some representations and capabilities), we'll need A, and it get it, we'd need B, and C.",

Don't be lazy, contextualization makes thought.

Ok. I do tend to get fixated. I reread your idea. You suggest that everything is broken down into structure, which is arrangements of funcs, and each func is documented, and it can be seen how the things are build by examining the structure. Funcs are easy to document. How these funcs fit together into structure is another subject, dont you think so? Just initial thoughts. But i do like the idea of trying to organize knowledge into a sensible structure, with layers, where one can traverse functional scopes to zero in on specifics. It seems like a good idea, but how would it be different from existing practice. Be nice to have a sample. I think you talking about breaking up knowledge in atomic peices, each with its description. Is that right?

    : Mindey
    :  -- 
    :  --