A Software Design System
Read PDF →Hess, 1980
Category: Software Engineering
Overall Rating
Score Breakdown
- Cross Disciplinary Applicability: 5/10
- Latent Novelty Potential: 5/10
- Obscurity Advantage: 4/5
- Technical Timeliness: 3/10
Synthesized Summary
This paper proposes a grammar-based system for translating software design decisions across abstraction levels using user-defined rules and managing parsing ambiguity.
While its specific text-centric and heuristic-based methods are largely outdated compared to modern design tools, the core concept of explicit, layered, rule-based transformations for design intent holds niche interest.
This might inspire research in explainable symbolic systems, but the original system's practical limitations and user burden necessitate significant conceptual overhaul.
Optimist's View
This paper describes SDS, a system for supporting systematic software design by translating user-defined formal languages across a hierarchy of abstraction levels.
a user-defined, dynamic dictionary of translation rules between formal languages representing design decisions at different levels of abstraction...
coupled with explicit, rule-based strategies for resolving inherent ambiguities in this translation process.
creating verifiable and explainable AI reasoning and planning systems by formalizing knowledge and planning steps as translations between formal language hierarchies.
Skeptic's View
The core assumption that software design can be effectively represented and manipulated primarily as a formal, hierarchical text structure defined by grammar rules is fundamentally misaligned with modern software development paradigms.
Requiring users to define entire language hierarchies and translation rules for each design... is a massive cognitive and practical overhead.
The ambiguity resolution mechanism is particularly weak. Relying on simple counts of branches or "implied primitives" is a heuristic that might work for toy examples but would quickly break down...
Modern software development environments and tools have superseded the specific mechanisms proposed in this paper.
Final Takeaway / Relevance
Watch