Glyph Melding: A Spatial Approach to Visual Programming

Vision

Instead of the traditional node-and-wire paradigm of visual programming languages (VPLs), QNTX introduces glyph melding - a direct manipulation interface where glyphs physically fuse together through spatial proximity.

Glyphs don't connect via explicit wires or lines. They meld together like magnetic pieces, forming compositions through adjacency. Data flows implicitly through the melded structure, left to right, creating readable pipelines without visual clutter.

Core Concept

When a user drags one glyph near another:

  1. The edges begin to morph toward each other, previewing the meld
  2. Upon release, the glyphs fuse into a single draggable unit
  3. They remain distinct entities within the composition
  4. The composition can be decomposed by pulling glyphs apart

This creates a tactile, intuitive programming experience - like working with physical objects rather than abstract connections.

Design Principles

Direct Manipulation

Type Compatibility

Linear Composition

Reversible Operations

Example Interactions

Creating a Pipeline

User drags [ax] glyph toward [python] glyph
→ Edges morph as they approach
→ Release to meld: [ax|python]
→ Drag [prompt] to the right side
→ Final composition: [ax|python|prompt]

Decomposing

User grabs [python] from [ax|python|prompt]
→ Pulls [python] away from the meld
→ Results in: [ax|prompt] and separate [python]

Implementation Considerations

Visual Feedback

Interaction Model

Type System

Related Work

Tangible Programming

The concept builds on decades of research in tangible programming interfaces, where physical manipulation replaces abstract symbolic programming.

Direct Manipulation & Visual Programming

Key Differentiator

Unlike these systems which use:

Glyph melding uses proximity-based fusion - a more organic, fluid interaction that eliminates connection management while maintaining clear data flow semantics.

Semantic Query Composition (SE → SE)

When SE₁ melds rightward to SE₂, the downstream glyph narrows the upstream search space. SE₁ ("science") defines a broad semantic region; SE₂ ("about teaching") intersects it. Only attestations matching both queries appear in SE₂. SE₁ continues to show its own unfiltered results independently. The downstream similarity score is reported to the user.

Chaining (future): Currently supports pairwise intersection. For chains of 3+ (SE₁→SE₂→SE₃), true transitive intersection (SE₁∩SE₂∩SE₃) requires propagating the full ancestor chain through the meld graph.

Union (future): Vertical SE composition would merge disjoint semantic regions — "machine learning" ↓ "gardening" shows attestations matching either. The dual of intersection: spatial union rather than refinement.

Future Directions

Advanced Melding Patterns

Interaction Enhancements

Visual Language

References

  1. Suzuki, H., & Kato, H. (1995). AlgoBlock: A Tangible Programming Language. MIT Media Lab.

  2. Google Creative Lab, IDEO, & Stanford University. (2016). Project Bloks: Making Code Physical for Kids. Google Research.

  3. Tangible Play Inc. Osmo Coding: Tangible Programming for Children. Demo.

  4. Yu, J., & Garg, R. (2025). Tangible-MakeCode: Bridging Physical Coding Blocks with Web-Based Programming. CHI 2025. Demo.

  5. Victor, B. (2012). Learnable Programming. Essay.

  6. Shneiderman, B. (1983). Direct Manipulation: A Step Beyond Programming Languages. IEEE Computer, 16(8), 57-69.

  7. MIT CSAIL. (2017). Tangible Computing. Engineering Interactive Technologies Course Materials. PDF.