Our standard

We hold ourselves to the same bar.

This is a HumanityAGSI tool, which means it ships under one test: would we hand it to someone we serve, unprompted, and feel proud our mark is on it? For a site about manipulation, that bar is non-negotiable.

  • No dark patterns. None.

    No countdown timers, no confirmshaming, no pre-checked boxes, no roach-motel flows, no nags. Every link does what it says. A tool that decodes manipulation has no business using it.

  • No tracking, no accounts, no cookies.

    There is no analytics script, no third-party request, and no database. The Enmeshment Audit is scored entirely in your browser and saved only on your device. We couldn’t sell your data if we wanted to — we never receive it.

  • Evidence strength is always on the label.

    Every claim is tagged robust, mixed, contested, or metaphor. We flag failed replications (ego depletion, d = 0.04 across 23 labs) and overclaims (the “dopamine hit”) rather than quietly leaning on them.

  • We represent scientific contestation honestly.

    On charged topics like social media and teen mental health, we report the disagreement instead of picking the scarier side. Leaked internal Instagram research (documented) sits next to Orben & Przybylski’s finding that the screen-use effect is tiny (~0.4% of variance, contested). The Confidence Trap exists to show manipulation by design — not to amplify a moral panic the evidence doesn’t support.

  • Metaphors are marked as metaphors.

    “Enmeshment,” “reptilian brain,” and the “dopamine hit” are framings, not mechanisms. We say so, and point to the more accurate science (e.g. dopamine encodes reward prediction; switching-cost economics carries the Trap).

  • Examples are sourced and neutrally phrased.

    We describe what a design does (“the checkout pre-selects an add-on”), not what a company “is.” Allegations are labelled as allegations. There is no user-submitted “Hall of Shame” — that would carry real defamation risk without moderation and legal review.

  • Built to WCAG 2.2 AA.

    Semantic structure, a skip link, visible focus, color paired with text (never color alone), keyboard-operable controls, and reduced-motion support. Accessible by default, because integrity includes who can use it.

Found something here that falls short of this standard? That’s a bug we want to fix. The credibility of the whole thing rests on it.