Research agenda
Open Problems
These are the limits of the current SymLan research preview. They are listed directly so reviewers can evaluate the project without guessing what is demonstrated, what is simulated, and what remains unresolved.
The project is not claiming that these problems are solved. The goal is to make them explicit enough for independent reviewers, implementers, and hardware researchers to attack them productively.
-
Critical
Physical Substrate Validation
The current reference path is simulation-backed. Future physical backends, including chemical oscillator arrays, VO2 devices, photonic systems, and related dynamical substrates, remain active research targets. No physical backend is currently claimed to reproduce the reference behavior. -
Critical
Public Reproducibility Package
A serious public release needs runnable scripts, fixed seeds, versioned outputs, expected logs, and a clear PASS/FAIL harness for the reference path. Until that package exists and can be run by others, the project should be treated as a specification plus research program rather than an independently reproduced result. -
Critical
Independent Replication
No core result should be treated as settled until an outside party reproduces it from the public package. Independent replication is the required bridge from internal claim to shared technical evidence. -
Hard
Simulation-to-Physical Fidelity
Even if a simulated attractor system produces stable vocabulary resolution, it does not follow that a physical oscillator substrate will preserve the same basin structure, noise tolerance, threshold behavior, or replay fidelity. The translation layer must be measured, not assumed. -
Hard
HAL Non-Invertibility
A hardware abstraction layer can map declared constraints to substrate controls, but the reverse map from observed behavior back to source intent may be many-to-one. SymLan needs round-trip diagnostics and explicit uncertainty reporting before cross-substrate equivalence claims become credible. -
Hard
Vocabulary Stability Across Runs
A resolved vocabulary is only useful if its identifiers remain stable enough under repeated runs, perturbation, and replay tests. Seeded simulations can be deterministic; physical systems introduce drift, noise, calibration error, and aging. Stability thresholds must be defined and tested. -
Hard
Scaling Beyond Reference Examples
The current language and readout framework need stress tests across larger state spaces, more stored patterns, different load factors, different projection dimensions, and more complex topologies. Scaling behavior should be reported as a boundary curve, not as a slogan. -
Design
Semantic Compositionality
A validated glyph can identify a resolved state. That does not yet prove that glyphs compose into stable higher-order meanings. The gap between addressable attractor labels and language-scale compositional semantics remains open. -
Design
Parser, Type Checker, and Runtime
The research preview describes the language discipline and reference path. A production-quality parser, type checker, runtime, package format, and conformance test suite remain implementation work. -
Governance
Registry and Authority Model
The Tower/SymVoc registry concept needs a concrete governance model: admission rules, provenance requirements, revocation, dispute handling, compatibility claims, and namespace authority. A registry is only useful if its trust boundaries are clear.