fix: erase all em-dashes (—) and en-dashes (–) / AI double dashes everywhere; use single hyphens per badge hygiene rule

Author: qectorlab

ai.txt

@@ -1,4 +1,4 @@
-QECTOR Decoder 3 — quantum error correction decoder package for Python and Rust.
+QECTOR Decoder 3 - quantum error correction decoder package for Python and Rust.
 
 Maintained by Guillaume Lessard / iD01t Productions.
 

source/README.md

@@ -1,4 +1,4 @@
-# QECTOR — qectorlab.github.io
+# QECTOR - qectorlab.github.io
 
 Lightweight static bridge + marketing site for **QECTOR Decoder v3** (high-performance decoder library) and **Qector Workbench** (free professional GUI) under the **qectorlab** brand.
 
@@ -8,8 +8,8 @@ Lightweight static bridge + marketing site for **QECTOR Decoder v3** (high-perfo
 
 - Quick-start: pip install qector-decoder-v3
 - Free Workbench: https://github.com/qectorlab/qector-decoder-workbench/releases/tag/v3.4.0
-- Key claims + evidence: Exact PyMatching parity d=15, +35.7% gains, GPU batch — all on GitHub
-- Commercial: Source-Available; licenses for production use — see qector.store/commercial
+- Key claims + evidence: Exact PyMatching parity d=15, +35.7% gains, GPU batch - all on GitHub
+- Commercial: Source-Available; licenses for production use - see qector.store/commercial
 - Full docs & artifacts: https://github.com/GuillaumeLessard/qector-decoder
 - PyPI + dynamic version: https://pypi.org/project/qector-decoder-v3/
 
@@ -38,7 +38,7 @@ npm run preview     # preview the production build locally
 npm run lint         # eslint
 ```
 
-Node 18+ is recommended. The production build type-checks the whole project before bundling — a failing `tsc -b` will fail the build.
+Node 18+ is recommended. The production build type-checks the whole project before bundling - a failing `tsc -b` will fail the build.
 
 ## Project structure
 
@@ -66,14 +66,14 @@ src/
 
 ### Shared component pattern
 
-Cards, section headers, and evidence blocks are intentionally centralized (`components/*.tsx`, not `ui/*`) so that every page — Home, Decoder, Benchmarks, Evidence, Pricing, Evidence reports — renders the same visual language instead of re-implementing card markup per page. When adding a new page, prefer reusing one of these over writing new markup; if none fit, extract a new shared component rather than duplicating.
+Cards, section headers, and evidence blocks are intentionally centralized (`components/*.tsx`, not `ui/*`) so that every page - Home, Decoder, Benchmarks, Evidence, Pricing, Evidence reports - renders the same visual language instead of re-implementing card markup per page. When adding a new page, prefer reusing one of these over writing new markup; if none fit, extract a new shared component rather than duplicating.
 
 ## Accessibility
 
 The site targets WCAG 2.1 AA. Notable patterns in place:
 
 - Skip-to-content link and route-change focus management (`Layout.tsx`).
-- Keyboard-operable tab groups (`role="tablist"`/`"tab"`/`"tabpanel"`, arrow-key navigation) — see the product switcher on `Pricing.tsx`.
+- Keyboard-operable tab groups (`role="tablist"`/`"tab"`/`"tabpanel"`, arrow-key navigation) - see the product switcher on `Pricing.tsx`.
 - `NeuralReveal` (the scramble-in heading effect) exposes the final text via `aria-label` and hides the animated glyphs from assistive tech with `aria-hidden`; it also honors `prefers-reduced-motion` and skips the scramble entirely.
 - Decorative SVGs and background video are `aria-hidden`; data-bearing SVG charts use `role="img"` with a text `aria-label` that states the actual values, and the underlying data is also present in an adjacent HTML table where applicable.
 - Form fields use explicit `<label htmlFor>` / `id` pairing, not implicit/visual-only association.
@@ -83,12 +83,12 @@ When adding new interactive UI, match these patterns: real semantics over ARIA w
 
 ## Content & claims policy
 
-Numbers used across the site (exact MWPM parity to d=15, +35.7% Belief-Matching LER reduction, 98.3% optimal shots at d=9, 832/832 tests) are simulation-validated with reproducible artifacts in the decoder repository. Do not edit these figures without updating the underlying validation report — the site's credibility depends on every public number being traceable to a citable artifact linked from `Evidence.tsx` / `Benchmarks.tsx`.
+Numbers used across the site (exact MWPM parity to d=15, +35.7% Belief-Matching LER reduction, 98.3% optimal shots at d=9, 832/832 tests) are simulation-validated with reproducible artifacts in the decoder repository. Do not edit these figures without updating the underlying validation report - the site's credibility depends on every public number being traceable to a citable artifact linked from `Evidence.tsx` / `Benchmarks.tsx`.
 
 ## Deployment
 
 The site is a static Vite build deployed to GitHub Pages from `qectorlab/qectorlab.github.io`. `npm run build` outputs to `dist/`; the deployed site is served from that output. SEO/crawler files (`robots.txt`, `sitemap.xml`, `llms.txt`, `llms-full.txt`) and `.well-known` metadata live in `public/` and are copied as-is.
 
 ## License
 
-Site content and code are © Guillaume Lessard / iD01t Productions. The QECTOR Decoder itself is dual-licensed (PolyForm Noncommercial for community use, commercial licenses available) — see [`/license`](https://qector.store/license) on the live site for current terms.
+Site content and code are © Guillaume Lessard / iD01t Productions. The QECTOR Decoder itself is dual-licensed (PolyForm Noncommercial for community use, commercial licenses available) - see [`/license`](https://qector.store/license) on the live site for current terms.

source/public/ai.txt

@@ -1,4 +1,4 @@
-QECTOR Decoder 3 — quantum error correction decoder package for Python and Rust.
+QECTOR Decoder 3 - quantum error correction decoder package for Python and Rust.
 
 Maintained by Guillaume Lessard / iD01t Productions.
 

source/src/components/AlgorithmCard.tsx

@@ -1,15 +1,15 @@
 type BadgeColor = 'cyan' | 'green' | 'purple' | 'gold';
 
 interface AlgorithmCardProps {
-  /** e.g. "Exact MWPM — Validated" */
+  /** e.g. "Exact MWPM - Validated" */
   title: string;
   /** Short technical tag shown as small mono text above the title, e.g. "Blossom + Belief-Matching" */
   tag?: string;
   /** Colored pill shown next to the title instead of `tag`, e.g. { label: 'Best LER', color: 'cyan' } */
   badge?: { label: string; color: BadgeColor };
   /** What it is and why it matters, plain language first */
   desc: string;
-  /** Optional single concrete fact/number to anchor the claim, e.g. "d = 3–15, exact parity" */
+  /** Optional single concrete fact/number to anchor the claim, e.g. "d = 3-15, exact parity" */
   proof?: string;
   /** Dims the card for research-stage / experimental entries */
   muted?: boolean;

source/src/components/ErrorBoundary.tsx

@@ -35,7 +35,7 @@ export default class ErrorBoundary extends Component<ErrorBoundaryProps, ErrorBo
             </div>
             <h1 className="text-xl font-bold text-primary mb-2">Something went wrong</h1>
             <p className="text-secondary text-sm mb-6">
-              This page hit an unexpected error. This has been logged locally — reloading usually resolves it.
+              This page hit an unexpected error. This has been logged locally - reloading usually resolves it.
             </p>
             <button onClick={this.handleReload} className="btn-cyan">
               Reload homepage

source/src/components/MetricCard.tsx

@@ -5,7 +5,7 @@ type MetricCardVariant = 'default' | 'gold' | 'compact';
 interface MetricCardProps {
   /** Short label/headline for the metric, e.g. "+35.7% Advantage" */
   label: string;
-  /** Optional large prominent value shown between label and desc, e.g. "d=3–15" */
+  /** Optional large prominent value shown between label and desc, e.g. "d=3-15" */
   value?: string;
   /** Supporting explanation, one to two sentences */
   desc: string;

source/src/pages/About.tsx

@@ -65,7 +65,7 @@ export default function About() {
           </div>
           <h1 className="text-4xl md:text-6xl font-extrabold tracking-tight leading-[1.1] mb-6">About QECTOR</h1>
           <p className="text-secondary text-lg md:text-xl max-w-2xl mx-auto leading-relaxed mb-8">
-            Independent quantum error correction research, engineering, and commercialization —
+            Independent quantum error correction research, engineering, and commercialization -
             built by <span className="text-cyan-300 font-semibold">Guillaume Lessard</span> with full transparency,
             reproducible artifacts, and no speculative claims.
           </p>
@@ -132,34 +132,34 @@ export default function About() {
             <div ref={(el) => addRef(el, 2)} className="card-surface">
               <h3 className="text-cyan-300 font-semibold text-sm uppercase tracking-wider mb-4">Software Stack</h3>
               <ul className="space-y-2 text-secondary text-sm leading-relaxed list-disc pl-5">
-                <li>Rust / PyO3 — core decoder engine</li>
-                <li>Python 3.10+ — API, CLI, benchmarking, simulation</li>
-                <li>Stim — circuit-level noise simulation</li>
-                <li>OpenCL / CUDA — GPU-accelerated batch decoding</li>
-                <li>PyPI distribution — <code className="text-cyan-300 text-xs">qector-decoder-v3</code></li>
+                <li>Rust / PyO3 - core decoder engine</li>
+                <li>Python 3.10+ - API, CLI, benchmarking, simulation</li>
+                <li>Stim - circuit-level noise simulation</li>
+                <li>OpenCL / CUDA - GPU-accelerated batch decoding</li>
+                <li>PyPI distribution - <code className="text-cyan-300 text-xs">qector-decoder-v3</code></li>
                 <li>HMAC-JWT commercial license enforcement</li>
-                <li>NiceGUI / CustomTkinter — SATI OS desktop interface</li>
+                <li>NiceGUI / CustomTkinter - SATI OS desktop interface</li>
               </ul>
             </div>
             <div ref={(el) => addRef(el, 3)} className="card-surface">
               <h3 className="text-cyan-300 font-semibold text-sm uppercase tracking-wider mb-4">Engineering Philosophy</h3>
               <ul className="space-y-2 text-secondary text-sm leading-relaxed list-disc pl-5">
-                <li>Reproducibility first — every claim ships with verifiable artifacts</li>
-                <li>Honest documentation — Decoder is Source-Available (not free); Workbench GUI app is free</li>
+                <li>Reproducibility first - every claim ships with verifiable artifacts</li>
+                <li>Honest documentation - Decoder is Source-Available (not free); Workbench GUI app is free</li>
                 <li>Simulation-validated before any public benchmark claim</li>
-                <li>Full provenance trail — DOI, SHA-256 bundles, IBM job IDs</li>
-                <li>No vendor lock-in — pluggable CodeProvider architecture</li>
+                <li>Full provenance trail - DOI, SHA-256 bundles, IBM job IDs</li>
+                <li>No vendor lock-in - pluggable CodeProvider architecture</li>
                 <li>Transparent validation reports with non-pass classification</li>
               </ul>
             </div>
             <div ref={(el) => addRef(el, 4)} className="card-surface">
               <h3 className="text-cyan-300 font-semibold text-sm uppercase tracking-wider mb-4">QECTOR Product Scope (qectorlab)</h3>
-              <p className="text-xs text-secondary mb-2">This site and QECTOR Decoder v3 + Workbench are the high-performance decoder library + professional GUI. Broader QECTOR vision (LCL-833, full SATI CODEX theoretical work) is research context — see GitHub for details.</p>
+              <p className="text-xs text-secondary mb-2">This site and QECTOR Decoder v3 + Workbench are the high-performance decoder library + professional GUI. Broader QECTOR vision (LCL-833, full SATI CODEX theoretical work) is research context - see GitHub for details.</p>
               <ul className="space-y-2 text-secondary text-sm leading-relaxed list-disc pl-5">
                 <li><strong className="text-primary">QECTOR Decoder v3</strong>: Rust/Python multi-algorithm QEC decoder library (Source-Available)</li>
                 <li><strong className="text-primary">Free Qector Workbench v3.4</strong>: Professional GUI, MCP, benchmarks (fully free)</li>
                 <li><strong className="text-primary">SATI OS</strong>: Full commercial stack on top (GUI + HALs + support)</li>
-                <li><strong className="text-primary">Research</strong>: SATI CODEX, LCL codes, "Mastering QEC" book — cross-linked on GitHub</li>
+                <li><strong className="text-primary">Research</strong>: SATI CODEX, LCL codes, "Mastering QEC" book - cross-linked on GitHub</li>
               </ul>
             </div>
           </div>
@@ -173,7 +173,7 @@ export default function About() {
                 { year: '2025 Q1', event: 'SATI CODEX LCL-832 framework developed: [[832,10,4]] CSS code on genus-5 surface. IBM Quantum hardware runs on ibm_fez. Artifacts on GitHub.' },
                 { year: '2025 Q2', event: 'QECTOR Decoder v3 core engine written in Rust / PyO3. Initial PyPI releases. Belief-Matching, BP-OSD, Union-Find decoders integrated. SATI OS desktop UI scaffolded.' },
                 { year: '2025 Q3', event: 'SATI v18 Titan-Class [[72,12,6]] BB QLDPC code environment. OpenCL and CUDA GPU backends achieving byte-for-byte identical corrections to CPU.' },
-                { year: '2025 Q4–2026', event: `v${pypiVersion} Decoder (Source Available) + free QectorWorkbench GUI v3.4 (25 tools, 10/10 polish). SATI OS full suite on top.` },
+                { year: '2025 Q4-2026', event: `v${pypiVersion} Decoder (Source-Available) + free QectorWorkbench GUI v3.4 (25 tools, 10/10 polish). SATI OS full suite on top.` },
                 { year: '2026 Q2', event: 'SATI OS v1.0.0 (build 1.0.0.0) first GA release. Features 39-panel desktop GUI, FastAPI server, dual CLIs, MCP server, and LCL-free open core with optional premium LCL-832 plugin. 1204 tests passing.' },
               ].map((item) => (
                 <div key={item.year} className="flex gap-4 items-start pb-4 border-b border-gridline/50 last:border-0">

source/src/pages/Benchmarks.tsx

@@ -51,7 +51,7 @@ export default function Benchmarks() {
           </h1>
           <p className="text-secondary text-lg md:text-xl max-w-3xl mx-auto leading-relaxed mb-8">
             Head-to-head against PyMatching on rotated surface codes using Stim circuit-level noise.
-            Results are SHA-256 sealed and available on GitHub — run them yourself with the artifacts.
+            Results are SHA-256 sealed and available on GitHub - run them yourself with the artifacts.
           </p>
           <div className="flex flex-wrap justify-center gap-3">
             <a href="https://github.com/GuillaumeLessard/qector-decoder" target="_blank" rel="noopener noreferrer" className="btn-cyan">GitHub Artifacts →</a>
@@ -75,11 +75,11 @@ export default function Benchmarks() {
               </thead>
               <tbody>
                 {[
-                  { algo: 'QECTOR-Blossom (MWPM)', dist: 'd = 3 – 15', ler: 'Exact parity', speed: '6.9–7.7× faster', status: 'Validated' },
+                  { algo: 'QECTOR-Blossom (MWPM)', dist: 'd = 3 - 15', ler: 'Exact parity', speed: '6.9-7.7× faster', status: 'Validated' },
                   { algo: 'Belief-Matching', dist: 'd = 5', ler: '\u221235.7% LER', speed: 'Comparable', status: 'Validated' },
                   { algo: 'QECTOR-Blossom', dist: 'd = 9', ler: '98.3% optimal shots', speed: 'Faster', status: 'Validated' },
                   { algo: 'GPU Batch Decoder', dist: 'Any', ler: 'Bit-identical to CPU', speed: 'Native CUDA / OpenCL', status: 'Available' },
-                  { algo: 'Union-Find', dist: 'd = 3 – 21', ler: '~1.5× higher LER', speed: '10–50× faster', status: 'Validated' },
+                  { algo: 'Union-Find', dist: 'd = 3 - 21', ler: '~1.5× higher LER', speed: '10-50× faster', status: 'Validated' },
                   { algo: 'BP-OSD', dist: 'qLDPC', ler: 'Code-dependent', speed: 'Slower (BP iterations)', status: 'Validated' },
                 ].map((row, i) => (
                   <tr key={i} className="border-b border-gridline/50 hover:bg-surface/50 transition-colors">
@@ -101,17 +101,17 @@ export default function Benchmarks() {
           {/* Detailed Results */}
           <div ref={(el) => addRef(el, 1)} className="grid grid-cols-1 md:grid-cols-2 gap-6">
             <div className="card-surface">
-              <h3 className="text-cyan-300 font-semibold mb-3">MWPM LER Parity (d=3–11)</h3>
+              <h3 className="text-cyan-300 font-semibold mb-3">MWPM LER Parity (d=3-11)</h3>
               <p className="text-secondary text-sm leading-relaxed mb-4">
                 QECTOR-Blossom achieves exact logical error rate parity with PyMatching across all tested code distances.
                 The adaptive-k MWPM implementation uses a union-find preprocessor followed by exact Blossom V matching on reduced graphs.
               </p>
               <div className="font-mono text-xs text-muted-foreground space-y-1">
-                <div>d=3:  LER = 0.0117 (PyMatching: 0.0117) — exact match</div>
-                <div>d=5:  LER = 0.0079 (PyMatching: 0.0079) — exact match</div>
-                <div>d=7:  LER = 0.0051 (PyMatching: 0.0050) — exact match</div>
-                <div>d=9:  LER = 0.0030 (PyMatching: 0.0031) — exact match</div>
-                <div>d=11: LER = 0.0018 (PyMatching: 0.0018) — exact match</div>
+                <div>d=3:  LER = 0.0117 (PyMatching: 0.0117) - exact match</div>
+                <div>d=5:  LER = 0.0079 (PyMatching: 0.0079) - exact match</div>
+                <div>d=7:  LER = 0.0051 (PyMatching: 0.0050) - exact match</div>
+                <div>d=9:  LER = 0.0030 (PyMatching: 0.0031) - exact match</div>
+                <div>d=11: LER = 0.0018 (PyMatching: 0.0018) - exact match</div>
               </div>
             </div>
             <div className="card-surface">
@@ -132,13 +132,13 @@ export default function Benchmarks() {
           {/* Speed Comparison */}
           <div ref={(el) => addRef(el, 2)} className="card-surface">
             <h3 className="text-cyan-300 font-semibold mb-4">Speed Comparison (Relative to PyMatching)</h3>
-            <p className="text-muted-foreground text-xs mb-4">Blossom figures are simulation-validated. Union-Find and GPU Batch figures are hardware-dependent — regenerate locally before citing.</p>
+            <p className="text-muted-foreground text-xs mb-4">Blossom figures are simulation-validated. Union-Find and GPU Batch figures are hardware-dependent - regenerate locally before citing.</p>
             <div className="space-y-4">
               {[
                 { label: 'QECTOR-Blossom (d=5)', factor: '6.9×', desc: 'Adaptive-k preprocessing reduces graph size before exact matching. Simulation-validated.' },
                 { label: 'QECTOR-Blossom (d=9)', factor: '7.7×', desc: 'Larger gains at higher distance due to sparser reduced graphs. Simulation-validated.' },
                 { label: 'Union-Find', factor: 'Faster', desc: 'Near-linear scaling; approximate but extremely fast. Exact multiplier is hardware-dependent.' },
-                { label: 'GPU Batch', factor: '6.9–7.7×+', desc: 'CUDA/OpenCL batch decoding; gains exceed CPU at suitable batch sizes. Bit-identical to CPU MWPM. Hardware-dependent.' },
+                { label: 'GPU Batch', factor: '6.9-7.7×+', desc: 'CUDA/OpenCL batch decoding; gains exceed CPU at suitable batch sizes. Bit-identical to CPU MWPM. Hardware-dependent.' },
               ].map((item) => (
                 <div key={item.label} className="flex items-center gap-4">
                   <div className="flex-1">
@@ -155,7 +155,7 @@ export default function Benchmarks() {
           <div ref={(el) => addRef(el, 3)}>
             <InteractiveChart
               type="distance"
-              title="LER vs Code Distance (d=3–11, p=0.001)"
+              title="LER vs Code Distance (d=3-11, p=0.001)"
               subtitle="QECTOR-Blossom vs PyMatching · Circuit-level depolarizing noise · Rotated surface code"
             />
           </div>