Accessibility
This page says what was actually tested on this site, what standard it was tested against, and where it still falls short. It is a statement of fact, not a claim of certification.
What standard, and why
This site targets WCAG 2.1 Level AA, tested through the web requirements of ETSI EN 301 549 v2.1.2, clause 9. Clause 9 §9.5 is the provision that incorporates WCAG 2.1 at Level AA, so the two are the same bar stated in two ways.
EN 301 549 is the harmonised European standard for accessible ICT. It is the reference that Canadian accessibility guidance points organisations to, which is why it is named here rather than only WCAG. Nothing on this page depends on being read in a particular jurisdiction: the measured requirements are the WCAG 2.1 AA ones.
How it was tested
By automated probes against the rendered page in a real browser, measuring the document the way a user agent computes it rather than reading the source:
- Text and non-text contrast measured from computed styles, resolving
oklch()to sRGB through the browser’s own parser so modern colour syntax measures correctly instead of silently returning nothing. - Reflow at 320 CSS px and at 200 % text zoom, checking for horizontal scrolling rather than assuming it.
- Text-spacing overrides (line height, paragraph, letter and word spacing) applied to confirm no content is clipped.
- Keyboard reachability, focus visibility, heading order, language, duplicate ids, and accessible names for interactive elements.
Contrast was measured against the composited background — translucent colours flattened over what is actually behind them — so the numbers below reflect what renders, not what the stylesheet intends.
Results
Measured on this site’s pages. “Not applicable” means the page contains nothing for that requirement to apply to, which is how absence is treated under clause 9.0.
| Requirement | What it covers | Result |
|---|---|---|
9.1.1.1 | Non-text content has a text alternative | Pass |
9.1.4.3 | Contrast (minimum) — 4.5:1 text, 3:1 large | Pass — worst 5.02:1 |
9.1.4.4 | Resize text to 200 % without loss | Pass |
9.1.4.10 | Reflow at 320 CSS px, no two-dimensional scroll | Pass |
9.1.4.11 | Non-text contrast — 3:1 for meaningful graphics | Pass |
9.1.4.12 | Text spacing overrides do not clip content | Pass |
9.2.1.1 | Keyboard — everything operable is reachable | Pass |
9.2.3.1 | Three flashes — no flashing content | Pass |
9.2.4.1 | Bypass blocks — skip link past repeated content | Pass |
9.2.4.2 | Page titles are present and meaningful | Pass |
9.2.4.6 | Headings and labels describe their topic | Pass |
9.3.1.1 | Language of the page is set | Pass |
9.4.1.1 | Parsing — no duplicate ids | Pass |
9.4.1.2 | Name, role, value for interactive elements | Pass |
9.1.2.x | Time-based media alternatives | N/A — no media |
9.2.2.1 | Timing adjustable | N/A — no time limits |
9.2.2.2 | Pause, stop, hide moving content | N/A — nothing moves |
9.1.3.5 | Identify input purpose | N/A — no forms |
9.3.3.x | Input assistance, error identification | N/A — no data entry |
9.1.4.13 | Content on hover or focus | N/A — no overlays |
On the contrast margin. The tightest pair in the palette now measures 5.02:1 against a 4.50:1 floor, up from 4.64:1. The fix was applied to the ground rather than the ink: --surface was darkened from 38 % to 35.5 % lightness, which raised every pair sitting on it at once — muted text (4.64 to 5.10) and all three tag fills (4.62–4.74 to 5.02–5.16). Lifting --muted instead would have cleared the threshold but left it 0.01 lightness from --fg, collapsing the two-tone hierarchy. Anyone editing the palette should re-measure rather than assume the margin survives.
Where it falls short
These are real, and stated plainly rather than rounded off:
- No screen-reader pass. Everything above is machine-measured. No assistive-technology session — NVDA, JAWS, or VoiceOver — has been run against this site. That is the single largest gap in this statement: automated tools cannot judge whether a text alternative is accurate, whether reading order makes sense, or whether an interface is coherent when heard rather than seen.
- Not everything is machine-checkable. Clause 9 runs to fifty requirements at Level AA. The table covers a subset. Requirements that need human judgement — whether alt text is meaningful, whether link text is descriptive in context, whether content is understandable — are outside what these probes can assert.
- No conformance claim. A statement that this site “conforms to EN 301 549” would require a full audit, manual testing with assistive technology, and a documented evaluation. That has not happened, so no such claim is made here.
- One page, one browser. Measurements were taken in Chromium. Behaviour in other engines, and with browser extensions or user stylesheets or forced-colours modes applied, is not characterised.
- The share card image has no text alternative in situ. The social preview image at
/img/og-card.jpgcarries its content as painted pixels — the name, role and summary are drawn into the JPEG rather than being selectable text. Its meaning is conveyed to assistive technology only through theog:image:altandtwitter:image:altattributes, which crawlers and social platforms may or may not surface. The same information is available as real text on this page and the profile page. - Level AAA contrast is not met, and is not targeted. This site targets WCAG 2.1 Level AA. The AAA threshold (1.4.6) requires 7:1, which the muted text tokens do not reach — they measure 5.02–6.95:1. Clearing 7:1 would require
--mutedat roughly 88 % lightness, which collapses it into--fgand removes the distinction the reading hierarchy depends on. Automated audits configured for AAA will report these as failures; that is accurate, and it is a bar this site deliberately does not claim. - Automated tools may misread the palette. The colour tokens are authored in
oklch()and composited withcolor-mix(). Plain sRGB hex fallbacks are declared first for parsers that cannot read modern colour syntax, but a scanner that fails to parsecolor-mix()oroklch()may substitute a default background rather than reporting that it could not measure — producing a contrast failure that does not exist in any real browser. Verify reported failures against the rendered page before acting on them. The ratios quoted in this statement were computed from the tokens directly, withcolor-mix()flattened the way a browser composites it.
Feedback
If something on this site does not work for you — an inaccessible control, a contrast problem, content that reads badly with a screen reader — tell me directly and I will fix it. Email noah@mechanima.tech.
If you need something on this site in a different form, ask. It is a static page; converting it is straightforward.
Details of this statement
- Prepared: 28 September 2026
- Standard applied: ETSI EN 301 549 v2.1.2, clause 9 (§9.5 → WCAG 2.1 Level AA)
- Method: automated browser measurement of the rendered document; no assistive-technology testing performed
- Reviewed by: the site owner, not an independent third party
This statement will be updated when a screen-reader pass is completed, or when the page changes enough to invalidate the measurements above.