JIDEL — a programming language that describes
and executes the flow of thought

Research Proposal

A language that began with a single silkworm

Spring 2011, a nephew’s holiday homework — observing silkworm larvae

The idea for JIDEL began in the spring of 2011, with the observation of silkworm larvae that was a nephew’s holiday homework.

Watching the silkworm spin its cocoon over several days, the designer saw that a living creature’s behavior divides into a few layers, with branches of judgment between them. That observation became a long-kept note on the classification of behavior, and grew into a single question — why is there no language that can record the structure of thought without losing it? Natural language is rich but does not execute; programming languages execute but were far too narrow to hold thought.

Its development is guided by simplification, abstraction, and consistency, and its long-term role is to become a shared tool for describing and experimenting with the concepts of Neutom Network, BOSIFWEM, and ECONAS.

Research Proposal

Five design axes

The five pillars on which its grammar and execution model stand

JIDEL’s grammar and execution model rest on the following five axes. Among them, postfix notation is the language’s first impression — the order of a Korean sentence is the order of execution.

Research Proposal

Relation-oriented — meaning is born from connection

A noun’s meaning = the bundle of edges hung on its node

JIDEL sees concepts, data, and code as a web of interconnected relations. The meaning of a noun is not a definition written somewhere, but is revealed in what it is connected to, and how.

Implementation Plan

Code is information too — from JIDEL to Neutom Network

Code = Data = Information · code written in JIDEL flows into Neutom Network as information

JIDEL treats bundles of commands, routines, and functions like values — binding data and code into one through addresses (pointers), so that code can be compared, evaluated, and composed (the code = data principle). From here it goes a step further. The source code and intermediate-language code (IBC) written in JIDEL flow directly into a kind of the primary information that Neutom Network handles. On Neutom Network’s list of source information, alongside physical, mathematical, conceptual, and BOSIFWEM information, 「JIDEL information」 takes its own place. Code is a citizen equal to data and information — and therefore ‘first-class’.

Beyond von Neumann — toward a looser, freer paradigm

Ever since von Neumann, computing has long carried on the idea of handling instructions (code) and data separately. To see code as data is to tear down that rigid boundary — a shift of view toward a looser, freer kind of programming.

This third path has a long precedent — LISP, half a century ago, already treated code and data as the same structure (homoiconicity), cracking the von Neumann dichotomy for the first time. JIDEL widens that tradition into the age of neural networks: loosening the rigid boundaries among code, data, and model so that all may cross freely as a single kind of information (a three-way union). When a program can treat even itself and its learning model as information, then — repairing itself (self-modification), reflecting on its own thoughts (metacognition), and changing itself toward a goal — will at last becomes possible.

Implementation Plan

Jactor — an active object that extends the Actor

A unit of computation with relations, urgent messages, and dynamic modeling · a JIDEL keyword

Thinking is a process of simulation, moving images about in the mind. For that, units of computation must exchange messages on their own and change their own structure as the situation demands. The Jactor is a JIDEL keyword that extends the Actor of other languages — a stateful unit of computation that independently exchanges asynchronous messages. To the Actor it adds relations, urgent (priority) messages, and dynamic modeling, becoming an active object that forms relations on its own, handles urgent signals first, and reshapes itself while running.

Implementation Plan

Logic of six states — making the unknown a full logic value

Not two — true and false — but six

JIDEL’s logic values are not two — true and false — but six. The key is that the ‘unknown state’ is not an error but a full-fledged logic value.

• We are exploring encoding the six states in 8 bits together with a confidence channel. The layout is arranged so that a value initialized to 0 becomes Unevaluated (I).
• Operations are handled quickly via lookup tables, and the operational laws (associativity, distributivity, and so on) are checked by exhaustive machine verification over all combinations. We built a verifier that exhaustively checks all 2.16 million combinations, and it passed 216 consistency conditions.
• Which rules to adopt is managed as a decision log — thirty-four items are on it and seventeen are settled; the rest are left open. If verification turns up a counterexample, we revise the design.

Implementation Plan

The spiral of knowing — the order in which the six states turn

It begins at Unevaluated and comes back to Unknown

The six states are not a list laid out flat; they have an order in which they turn. It begins at Unevaluated (I), not yet asked; passes through Superposition (S), where candidates overlap; is judged True (T) or False (F); reaches Paradox (P) when the grounds conflict; and from there comes back to Unknown (M). That unknown is not the unevaluated of the beginning — for it is an unknown obtained after asking.

So this logic takes no single state as a terminus. Confidence accumulates, and it is also demoted — a counterexample brings it down, and the moment it reaches for ultimate knowing it steps back of its own accord. Where Western logic has computed with settled states alone, the long Eastern notion of self-cultivation enters as a rule of computation. Humility is built in not as a moral exhortation but as a rule of calculation — this is the keynote of the design.

Implementation Plan

Relative-value operations — reading the space between extremes

Interpolar theory · not "187 cm" but "kind of tall"

People don’t think "that person is 187 cm tall" — they just feel "kind of tall." A value isn’t an absolute scale; it’s placed by where it falls between the two extremes (min and max) of everything experienced so far: (x − min) / (max − min). And that position shifts depending on which group you’re in — the same 187 reads as "average" on a basketball team and "tall" elsewhere.

Unverified Hypothesis

Predicate frames — a sentence is a frame that opens slots

Nouns as processes, not values

This is a recent research direction. Nouns are treated not as fixed values but as processes evaluated according to conditions, and a sentence is seen as a frame of interrogative slots opened by the predicate.

• Lazy evaluation of nouns — a noun is a lazy-evaluation function of [condition → phenomenon], and the act of referring to it is the observation. Particles (i/ga, eul/reul, eseo…) are operators that attach conditions, and the sentence-final predicate is the trigger of evaluation.
• Interrogative slots — who, what, when, where, why, how, plus to whom, with what, how much, so-what, and others: fourteen slots form the sentence’s blank structure. An unfilled slot is not an error but remains an Unknown (M) value.
• Curiosity engine — we are exploring ways to support, at the language level, the act of seeking out and filling empty slots. To ask a question is to bring an unknown slot into the open.

Research Proposal

The possibilities this language opens

Beyond the technical spec — what it means for users and supporters

Implementation Plan

Short-term goals and its place

Documentation, and a working interpreter that runs small examples

The short-term goal is to document the grammar system and execution semantics, and to build a working interpreter that actually runs small examples.

• Core grammar: postfix notation · relation-oriented expression · code as data
• Execution model: stack-based execution · context-dependent operations · relative-value handling · many-valued judgment
• Implementation: a minimal interpreter · a set of examples · a specification document · a logic verifier

In the order of the whole research, JIDEL is the first brick — only once JIDEL is implemented can Neutom Network be built with it, and only when BOSIFWEM theory and Neutom Network come together can we head toward ECONAS.

Research Proposal

See also

This page is a visitor-facing summary of the JIDEL design documents. Research records are made public on the research board.

JIDEL is being designed atop the influence of concatenative languages of the Forth·Factor family and languages such as C, Python, Prolog, and Lisp, with the addition of Korean grammatical theory and many-valued logic research.