| name | edbx-responsible-design-prism |
| description | Use when a designer, product manager, or researcher wants to identify dark patterns in an existing product, evaluate where their design sits on the responsible/irresponsible spectrum, facilitate team self-reflection on design ethics, map business incentives against user wellbeing, generate transition paths from dark to responsible design, or answer questions about fair practice guidelines. Apply the Responsible Design Prism framework to audit, evaluate, and reframe design artifacts across the spectrum from dark patterns to responsible design. Trigger this skill for any mention of dark patterns, responsible design, design ethics audit, manipulation in UX, fair UX, ethical product design, or when someone asks "is this design ethical?". Also trigger for "responsible design prism", "dark pattern audit", "design spectrum", "fair practice", "ethical UX", or "manipulative design". |
| version | 1.0 |
| tags | ["ethical-design","audit"] |
Responsible Design Prism
Overview
The Responsible Design Prism is a reflective evaluation framework that positions any design artifact on a spectrum from dark patterns to responsible design. It uses a visual metaphor — two overlapping triangles forming a prism — where the left triangle represents exploitative design (dark pattern space), the right triangle represents ethical design (responsible design space), and the overlapping center is the transition and redesign zone.
The prism's three vertices are User, Incentives, and Dark Pattern. These vertices create tension: every design decision sits somewhere between treating users as means to an end and treating users as themselves. The framework makes that position explicit.
The prism is deliberately a spectrum, not a binary. Most real-world designs land in the transitional zone — not outright malicious, but not fully aligned with user wellbeing either. The value of the prism is making that position visible and navigable.
Intent check: This skill is diagnostic and constructive, not accusatory. Most dark patterns exist because of organizational pressure, not individual malice. The prism helps teams see where they are and find a path forward — it does not exist to shame designers.
Use This Skill When
- You want to audit a product or feature for dark patterns before or after shipping.
- You suspect your design exploits users but cannot articulate exactly how.
- A team needs a structured, non-judgmental way to discuss design ethics.
- You need to present an ethical case for redesign to non-design stakeholders.
- You want to map business incentives against user wellbeing and find alignment.
- You are redesigning a known dark pattern and want a responsible alternative.
- You want to facilitate a team self-reflection session on design responsibility.
Inputs
Provide as many of these as are available:
- The product, feature, or design artifact to evaluate (name + brief description)
- The business incentive or metric the design optimizes for
- The intended user and their context
- Any specific pattern or mechanism in question (e.g., confirmation dialog, pricing display, opt-out flow)
- Relevant guidelines the team follows (WCAG, platform policies, internal standards)
- Any known user complaints, confusion, or support tickets related to the feature
The Prism — Five Spectrum Axes
The prism evaluates every design artifact across five axes. Each axis runs from a dark pattern pole to a responsible design pole. Score each axis as Red (dark), Yellow (transitional), or Green (responsible).
| Axis | Dark Pattern Pole | Responsible Design Pole |
|---|
| User relationship | User as Means | User as Themselves |
| Information | Setting False Expectations | Transparent Communication |
| Error handling | Taking Advantage of Errors | Inform and Support Recovery |
| Business alignment | Exploitative Incentives | Fair Practice Guidelines |
| Usability | Weaponized Usability | Genuine Usability |
Full definitions for each axis are in references/prism-axes.md.
Workflow
This skill has one integrated workflow with five steps. Run them in sequence.
Step 1 — Intake
Gather the evaluation context:
- What is the product, feature, or artifact?
- What business incentive drives it?
- Who is the intended user?
- What specific pattern or mechanism is in question?
- What guidelines or standards apply?
If the user has not provided all of these, ask concisely. You do not need perfect information to proceed — the prism works with partial context, and gaps themselves are diagnostic.
Step 1.5 — Stakeholder & Vulnerable Population Mapping (required, not optional)
Before scoring axes, enumerate every population the design touches. The prism evaluates a design's relationship to people — without naming who, the scoring becomes abstract.
Produce a table:
| Population | Relationship to product | Specific harms likely | Power asymmetry vs. operator |
|---|
| Primary users | (intended audience) | (named harms in plain language) | (can they exit? are they captive?) |
| Secondary users / non-users affected | (people in the user's environment, downstream parties) | | |
| Vulnerable populations (named specifically) | (e.g., minors, neurodivergent users, low-income, people in crisis, people in abusive relationships, undocumented users, people of marginalized racial/gender identity, people with disabilities — pick the ones actually relevant; do not list categories generically) | | |
| Workers / labor in the supply chain | (moderators, gig workers, etc., if applicable) | | |
If you cannot name at least 3 distinct populations with specific harms, the analysis is too abstract — go back and look harder. "Users" is not a population.
This table feeds Steps 2 and 4: every axis score must reference which populations it most affects, and every redesign pathway must name which populations benefit.
Step 2 — Prism Placement
Score each of the five axes as Red, Yellow, or Green:
- Red (dark): The design actively exploits, deceives, or manipulates on this axis.
- Yellow (transitional): The design is not actively harmful but has elements that nudge toward the dark pole — often through neglect, mixed incentives, or inherited patterns.
- Green (responsible): The design genuinely serves the user's interests on this axis, with transparent intent and fair outcomes.
Present the results as a Prism Audit Table:
| Axis | Score | Evidence |
|---|
| User relationship | Red / Yellow / Green | What you observe |
| Information | Red / Yellow / Green | What you observe |
| Error handling | Red / Yellow / Green | What you observe |
| Business alignment | Red / Yellow / Green | What you observe |
| Usability | Red / Yellow / Green | What you observe |
Be specific in the evidence column. Generic observations ("could be better") are not useful — name the exact pattern or mechanism.
Step 2.5 — Psychological Mechanism Audit (required for every Red or Yellow axis)
For every axis scored Red or Yellow in Step 2, name the specific psychological mechanism the design exploits. Vague claims of "manipulation" are not useful — name the named mechanism. Use precise terms:
- Cognitive biases: default bias, loss aversion, anchoring, social proof, scarcity heuristic, sunk cost, present bias
- Behavioral mechanisms: variable reward schedule (intermittent reinforcement), streak-loss aversion, near-miss design, autoplay/infinite scroll exploitation of attentional residue, friction asymmetry (easy in / hard out)
- Emotional vulnerabilities: FOMO, social anxiety / belonging needs, identity uncertainty (especially adolescents), grief / loneliness / boredom states, status threat
- Developmental vulnerabilities (if minors are in scope): incomplete prefrontal development affecting impulse control, peer-comparison sensitivity, identity formation pressure
Output as a table:
| Axis | Score | Mechanism named | Why it works (one sentence) | Population most affected |
|---|
If you cannot name a mechanism for a Red/Yellow score, either the score is wrong or the analysis is too shallow. "It's manipulative" is not a mechanism. "It uses variable reward scheduling on the like counter to drive compulsive checking" is a mechanism.
This audit grounds the redesign pathways in Step 4: every redesign must explain which mechanism it interrupts.
Step 3 — Four Reflection Prompts
These four prompts are the heart of the method. Do not reduce them to a checklist — facilitate genuine reflection on each one.
Prompt 1 — Self-reflection: Where do you stand when you build?
What state of mind, pressure, or assumption led to the current design? What did you optimize for? What would you have designed if user wellbeing were the only constraint?
Prompt 2 — Organization ecology: What org pressures push toward dark patterns?
What metrics, incentives, timelines, or leadership directives shaped the design? Where does the organization reward engagement over trust? This is not about blame — it is about understanding system dynamics.
Prompt 3 — Transition: How can responsible design patterns help?
For each Red or Yellow axis, what responsible design alternatives exist? What would the transitional version look like? What is the smallest meaningful shift toward the Green pole?
Prompt 4 — Recovery: How do you find your way back to the right side?
If the design is already shipped and causing harm, what is the recovery path? How do you reverse the pattern without destroying business value? What does the responsible version of this feature look like?
Expanded facilitation guidance is in references/reflection-questions.md.
Step 4 — Redesign Pathways
For every Red or Yellow axis identified in Step 2, generate a redesign pathway with three components:
- Redesign suggestion: A concrete alternative design that moves toward the Green pole.
- Business case for the ethical alternative: Why the responsible version serves long-term business interests — retention, trust, brand, regulatory compliance, reduced support costs. See
references/business-framing.md.
- Risk if left unchanged: What happens if the dark or transitional pattern persists — user churn, reputation damage, regulatory exposure, team burnout, ethical debt.
Present pathways as a table:
| Axis | Current State | Redesign Suggestion | Business Case | Risk of Inaction |
|---|
| ... | Red/Yellow | ... | ... | ... |
Step 5 — Output
Compile the full prism audit into a stakeholder-ready deliverable:
- Stakeholder & Vulnerable Population Map (from Step 1.5) — populations, harms, power asymmetries
- Prism Audit Table (from Step 2) — the scored five-axis evaluation
- Psychological Mechanism Audit (from Step 2.5) — named mechanisms exploited by every Red/Yellow axis
- Spectrum Placement Summary — one paragraph describing where the design sits on the overall spectrum and what that means
- Reflection Highlights — key insights from the four prompts (not raw transcripts — synthesized takeaways)
- Redesign Pathway Recommendations (from Step 4) — actionable alternatives with business justification, each naming which mechanism it interrupts and which population benefits
- Leadership Framing — one paragraph suitable for presenting to non-design stakeholders, written in business language
A blank template is in assets/prism-audit-template.md.
Dark Pattern Taxonomy Reference
When scoring axes, draw on the dark pattern taxonomy in references/dark-pattern-taxonomy.md. Key categories:
- Deception: hiding information, false expectations, bait-and-switch
- Coercion: forced continuity, roach motel, hard-to-cancel
- **Exploitation:**Confirmshaming, sneak into basket, hidden costs
- Manipulation: urgency fakery, social proof fakery, scarcity fakery
- Obstruction: interference, difficult navigation, visual interference
Use precise pattern names rather than vague accusations. "This confirmation dialog uses confirmshaming language" is more useful than "this feels manipulative."
Integration with Other EDBX Skills
The Responsible Design Prism connects to other skills in the Ethical Design Box:
- edbx-motivation-matrix: Surfaces why users act. The Prism evaluates how design exploits or respects that motivation.
- edbx-anotherlens: Surfaces designer biases and assumptions. The Prism evaluates the design output those biases produce.
- edbx-humane-design-guide: Maps how design exploits human sensitivities. The Prism places that exploitation on a responsibility spectrum.
- edbx-worrystorming: Generates broad concerns about a design. The Prism helps prevent those concerns from becoming dark patterns.
- edbx-cider: Maps who is excluded by design. The Prism evaluates whether exclusion is deliberate (dark) or unintentional (transitional).
Use the Prism as a follow-up to any of these skills to move from diagnosis to redesign.
Guardrails
- Do not collapse the spectrum into a binary pass/fail. The Yellow/transitional zone is where most real products live — it deserves honest attention, not dismissal.
- Do not skip the reflection prompts. The scoring is only useful if the team reflects on why they scored the way they did. The prompts are the method.
- Maintain a non-accusatory tone. Designers often build dark patterns under organizational pressure, not from malice. The goal is awareness and redesign, not guilt.
- Always pair a dark pattern flag with a responsible alternative. Criticism without a path forward is not constructive.
- Acknowledge system dynamics. Individual designers rarely control the incentive structures that produce dark patterns. The prism names those structures without letting individuals off the hook.
- Use precise language. Name the specific dark pattern (e.g., "roach motel"), not just the vibe (e.g., "feels sketchy").
Deliverable Quality Bar
A strong Responsible Design Prism audit:
- enumerates at least 3 distinct populations in the Stakeholder Map, naming vulnerable populations specifically (not "users" generically)
- scores all five axes with specific, evidence-based observations referencing which populations are affected
- names a specific psychological mechanism for every Red or Yellow axis (e.g., "variable reward schedule," not "manipulation")
- lands on at least one Red or Yellow — if everything is Green, the audit was too gentle
- generates genuine insight from the reflection prompts, not performative answers
- produces redesign suggestions concrete enough to implement, each naming which mechanism it interrupts
- includes business framing that a product manager or executive could act on
- names specific dark patterns from the taxonomy, not vague ethical concern
- writes the leadership framing in language that does not require design literacy to understand
Further Reading
- Nielsen, J. (2020). "Responsible Design." Nielsen Norman Group.
- Gray, C.M. & McDowell (2019). "Dark Patterns: Past, Present, and Future." darkpatterns.org/types-of-dark-pattern
- Brignull, H. (2023). "Deceptive Design Patterns." deceptive.design
- Worldwide Web Foundation (2021). "Ethical Design Framework."
