Editorial policy

Purpose and scope of our documentation

Every guide and tool explanation published on type-type is written to help readers solve a concrete typing or keyboard hardware question — whether that is raising net typing speed on the typing test, diagnosing a dead matrix column or chattering switch on the keyboard tester and key chatter test, or verifying how scoring formulas behave on the WPM calculator. We do not publish news roundups, scraped product summaries, or rewritten manufacturer marketing copy.

How guides and tool explanations are created

Before a guide is added to our guide library, it must meet five internal drafting standards:

1. Direct answer first: The opening paragraph states the conclusion, formula, or troubleshooting order plainly so a reader never has to scroll past filler to find the answer.
2. First-principles verification: Every equation (such as standard 5-character word conversion, character-adjusted Net WPM vs. full-word penalty Net WPM, USB 2.0 500 mA power budgets, 8-byte USB HID boot reports, or switch work-energy integrals) is derived and checked with a worked numerical example.
3. Hands-on browser and hardware testing: Claims about DOM KeyboardEvent.code values, event.repeat filtering, performance.now() interval resolution, and modifier latching are tested directly against our live tools across desktop browsers.
4. Honest caveats where numbers are approximate: Human typing benchmarks, switch actuation tolerances (typically ±10–15 gf from nominal factory ratings), and ergonomic angle ranges vary from person to person. Whenever a range is empirical rather than a mathematical identity, we state it explicitly as an approximate working range rather than an absolute rule.
5. Zero fabricated citations or personas: We never invent fictional authors, stock-photo biographies, unverified academic studies, or made-up user testimonials. Authorship on our About page uses only verified maintainer details or the plain project byline.

Review dates, updates, and corrections

Every guide displays a visible Last reviewed date below its heading. We review existing articles whenever we add or update a diagnostic tool, and whenever browser keyboard event handling or operating-system accessibility defaults change.

When a reader reports an inaccuracy, typo, or edge case to contact@type-type.app (or via the instructions on our contact page), we reproduce the issue, update the text or formula, run our automated SEO and content quality checks, and update the review timestamp.

To keep our reference articles distinct and useful rather than repetitive, each guide in our library covers a separate technical question — for example separating matrix circuit ghosting from mechanical switch contact bounce, and separating beginner column memorization from advanced bigram error analysis — and is checked by an automated similarity and completeness gate before publication.

Independence and separation of advertising from content

type-type is free to use. When display advertising (such as Google AdSense) is enabled on eligible pages, advertisements are strictly separated from both our interactive tools and our editorial text:

• No paid reviews or affiliate bias: We do not accept money, free hardware, or affiliate commissions from keyboard brands in exchange for favorable coverage. When we mention switch types (such as linear, tactile, clicky, optical, or Hall-effect) or firmware ecosystems (such as QMK, VIA, or ZMK), we do so strictly to explain technical mechanics and troubleshooting steps.
• Ads never alter tool geometry or active tests: Ad slots are never placed inside the active typing viewport or keyboard diagram, and any configured placements are suppressed while a typing test or keyboard check is running.
• Clear labeling and consent: Any display ad is served automatically by the ad network under the consent rules described in our privacy policy and terms of use, and never disguised as editorial guidance. Diagnostic pages where screen space or timing precision is sensitive keep ad containers unmounted when no placement is configured so layout shifts and main-thread work remain at zero.