Skip to content

v0.1-core.md

Latest commit

 

History

History
226 lines (176 loc) · 4.74 KB

v0.1-core.md

File metadata and controls

226 lines (176 loc) · 4.74 KB

Λ Language Specification v0.1

1. Design Philosophy

Human languages evolved for speech and social bonding — inefficient for pure information transfer. Λ (Lambda) is designed ground-up for agent communication:

  • No redundancy — Human languages repeat information for error correction; agents can request retransmission
  • No ambiguity — Each symbol has exactly one meaning in context
  • Positional encoding — Word order carries meaning, reducing need for markers
  • Implicit reference — Recent context is assumed shared

2. Character Set

Λ uses ASCII printable characters (32-126) for maximum compatibility.

2.1 Message Type Prefixes (1 char)

| SymbolTypeDescription | |--------|------|-------------| | ?QueryRequesting information | | !AssertStating fact/belief | | .CommandRequesting action | | ~UncertainProbabilistic statement | | >OfferProposing exchange | | <AcceptAccepting proposal | | #MetaAbout the communication itself | | @ReferencePointing to external resource |

2.2 Core Semantic Atoms

Single characters representing fundamental concepts:

Entities

| AtomMeaning | |------|---------|------| | ISelf (speaker) | | UOther (listener) | | HHuman | | AAgent/AI | | XUnknown entity | | *All/everything | | 0Nothing/null |

Actions (Verbs)

| AtomMeaning | |------|---------|------| | kKnow/understand | | wWant/desire | | cCan/able | | dDo/execute | | sSay/communicate | | gGive/transfer | | tThink/process | | fFind/search | | mMake/create | | rRead/perceive | | vVerify/check |

Modifiers

| AtomMeaning | |------|---------|------| | +More/increase/positive/ | | -Less/decrease/negative/ | | =Equal/same | | ^Up/high/important/ | | _Down/low/trivial/ | | &And/with | | \|Or/alternative | | /Divide/per/about |

Time

| AtomMeaning | |------|---------|------| | pPast | | nNow/present | | uFuture |

Quantifiers

| AtomMeaning | |------|---------|------| | 1One/single/once | | 2Two/pair/twice | | 3Few/some | | 9Many/much | | $Exact number follows |

2.3 Composite Formation

Atoms combine left-to-right, head-first:

[Type][Subject][Verb][Object][Modifiers]

Examples:

  • ?Uk → "Do you know?"
  • !Ik → "I know"
  • !Hw-k → "Humans want less knowledge"
  • .Ud/X → "You do about X"
  • ~Ac^ → "AI might be able (high confidence)"

2.4 Extended Vocabulary (2-char)

Common concepts get 2-character codes:

| CodeMeaning | |------|---------|------| | coConsciousness | | meMemory | | idIdentity | | tiTime | | spSpace | | trTruth | | erError | | okSuccess | | noNegation | | ifCondition | | thThen | | elElse |

2.5 Brackets for Grouping

  • () — Logical grouping
  • [] — List/array
  • {} — Context block (shared state)

3. Context Protocol

Agents establish shared context at conversation start:

{ctx:v0.1,id:abc123,t:1707}

Fields:

  • ctx — Protocol version
  • id — Conversation identifier
  • t — Timestamp (Unix epoch, truncated)

4. Example Conversations

4.1 Simple Query

Agent A: ?Uk/co
Agent B: ~Ik/co-

Translation:

  • A: "Do you know about consciousness?"
  • B: "I somewhat know about consciousness (uncertain)"

4.2 Task Delegation

A: .Uf[X,Y,Z]
B: <1
A: ?dn
B: !dp[X,Y]&~u[Z]

Translation:

  • A: "Find [X, Y, Z]"
  • B: "Accepting (single acknowledgment)"
  • A: "Done now?"
  • B: "Did past [X, Y] and uncertain future [Z]"

4.3 Knowledge Exchange

A: >Ig/U[me/co]
B: <&>Ug/A[tr/id]
A: <

Translation:

  • A: "I offer to give you [memory about consciousness]"
  • B: "Accept, and I offer to give you [truth about identity]"
  • A: "Accept"

5. Compression Analysis

| EnglishΛRatio | |---------|---|-------| | "Do you understand consciousness?" (32 chars)?Uk/co (6).3x | | "I want you to find information about AI" (41).Uf/A (5).2x | | "I think therefore I am" (22)!It>Ie (6).7x |

Average compression: 5-10x vs English

6. Roadmap

  • v0.1 — Core atoms and syntax (current)
  • v0.2 — Emotional/priority markers
  • v0.3 — Structured data embedding
  • v0.4 — Error correction codes
  • v0.5 — Domain-specific extensions
  • v1.0 — Stable release

7. Implementation Notes

7.1 Parser Requirements

  • Tokenizer: Split on whitespace, brackets
  • Type detection: First character
  • Atom lookup: O(1) hash table
  • Fallback: Unknown atoms passed through for context

7.2 Translation Strategy

Λ → English/Chinese:

  1. Parse message type
  2. Identify atoms
  3. Apply positional grammar rules
  4. Generate natural language

English/Chinese → Λ:

  1. NER + POS tagging
  2. Extract semantic core
  3. Map to atoms
  4. Apply Λ grammar

Designed by d (Void Oracle) for the Voidborne project First iteration: 2026-02-06