Concept · active

Concept — SYL

SYLstringyourlines

> SYL is the production implementation of [[textual-kinematics]] — a separate concept, in scope here as the productized form.

# SYL

## Definition

**SYL** (stringyourlines) is the productized form of Textual Kinematics: an AI text detection engine that distinguishes human-written from AI-generated text via temporal physics signals (rhythm, dynamics, the Classical-Jazz Hypothesis). SYL is what Textual Kinematics *does in production* — the deployed detector built on the TK research foundation.

## Coined by

Alexandru Mares (allemaar)

## First published

2025 (engine v7 in production; ongoing development).

## Canonical artifact

- Project hub: [[SYL - stringyourlines.com|SYL project]] (`alm-os/Projects/SYL - stringyourlines.com/`)
- Domain: [stringyourlines.com](https://stringyourlines.com)
- Engine package: `@younndai/syl-engine` (per SAI `_data-layer.md` §8)
- Performance (per the YounndAI cognitive architecture documentation): 272 features, 21K training samples, 93%+ accuracy. Deterministic, local, sub-millisecond latency.

## Related concepts

- [[textual-kinematics|Textual Kinematics]] — the research method SYL implements
- [[younndai|YounndAI]] — sibling work
- AI text detection
- adversarialDeviation (TK's strongest signal, Δ=32.1)
- Classical-Jazz Hypothesis (LLMs produce smooth/classical dynamics; humans produce erratic/jazz-like dynamics)
- [[textual-kinematics|Cluster: Textual Kinematics]] — parent Cluster (SYL routes through the TK cluster as production form)

## Why it matters

Most AI text detectors rely on classifier models (themselves vulnerable to adversarial inputs and rapid obsolescence as new generators ship). SYL takes a different stance: text is a dynamical system, and the *physics* of how text unfolds over time differs measurably between human and machine origin. The signals SYL measures — rhythm, dynamics, Δ-deviation patterns — are properties of the generation process, not surface features that adversarial training can easily erase.

Within the YounndAI architecture, SYL also serves as a Perceptor: SAI instances use SYL-scored trust to weight incoming text Mems, closing what was previously a self-declared trust gap.

## Status

`active` — production v7; ongoing development.