edbx-anti-heroes

name	edbx-anti-heroes
description	Use when a designer needs to critique a design, flow, feature, or concept for dark patterns and anti-user values, run reverse brainstorming to expose worst-possible solutions, generate ethical concept alternatives, or facilitate ethical dialogue in a team using shared Anti-Hero / Hero language. Use the Anti-Heroes card method to surface manipulative design intentions and pair them with ethical Hero counterparts. Lean toward this skill whenever the user mentions dark patterns, manipulative design, deceptive UX, ethical review, anti-patterns, designer responsibility, or wants to "break my design" before shipping.
version	1.0
tags	["ethical-design","forecast"]

Anti-Heroes

Overview

Designers continuously balance user, shareholder, and business values, and the design choices they make inscribe those values into the product. The Anti-Heroes method, from Mehta, Chivukula, Gray, et al. (2024), exposes the manipulative roles a designer can play (Anti-Heroes) and pairs each with the ethical role that counters it (Hero). The point is not to label a designer as bad; it is to make the value trade-off visible so the team can choose differently.

Use this skill to put designs through a Hero / Anti-Hero check, generate worst-possible solutions on purpose, develop ethical concept alternatives, or facilitate a team conversation that names manipulative moves out loud instead of letting them slide.

For the full card deck (Anti-Hero / Hero pairs and what each one does), read references/cards.md. Read it whenever the user asks for a critique, a reverse brainstorm, or wants the deck explained — most use cases need at least a few specific cards named, not just "this looks dark-pattern-y."

Use This Skill When

• A team wants to evaluate a design, flow, feature, or concept for manipulative or anti-user patterns before shipping.
• A reviewer wants principled language for what feels off about a design instead of vague "this seems gross."
• A team is doing reverse brainstorming and wants to deliberately design the worst version to surface anti-goals.
• A designer wants to generate ethical concept alternatives that counter an Anti-Hero pattern they have already noticed.
• A team meeting needs a shared vocabulary to call out manipulative moves respectfully (the "football referee" use described in the source).
• A discovery team is reviewing competitor or industry trends for ethically problematic patterns to avoid.

Inputs

Provide as many of these as are available:

• the design under review: screenshot, wireframe, flow description, copy, prototype link, or a verbal walkthrough
• the user's primary goal in this design and the user's likely emotional state
• the business goal or stakeholder pressure behind the design
• any specific Anti-Hero or Hero card the user wants to focus on
• the use mode the user wants: evaluation, reverse brainstorm, conceptualization, or ethical dialogue
• constraints that genuinely cannot move (regulation, hard business deadlines, accessibility minimums)

If the user has not said which mode they want, infer it from how they describe the task and confirm in one short sentence before going deep.

The Four Modes

Anti-Heroes is one method with four working modes. They share the same card vocabulary but produce different outputs.

Mode 1: Evaluation

Critique an existing design and identify which Anti-Hero patterns are present, then propose the matching Hero moves.

1. Restate the design's required user task and the user's likely emotional state.
2. Walk the design and tag each suspicious element with a candidate Anti-Hero card (Trap-Setter, Empathy Manipulator, Nickeling-and-Diming, etc.). Cite the specific UI moment, copy, or interaction — not vibes.
3. For each tag, name the Hero counterpart and describe one concrete change that flips the move.
4. Note where Anti-Hero patterns are unavoidable (e.g., a regulator-required friction step) so the team does not over-correct.
5. Hand back: list of Anti-Hero hits, Hero counter-moves, and a short risk note.

Mode 2: Reverse brainstorming

Deliberately design the worst possible version to surface anti-goals before they sneak in by accident.

1. Pick 2–4 Anti-Hero cards that are most plausible for this product or feature.
2. For each card, generate one or two worst-version moves the design could make. Be specific: what does the screen, copy, or flow actually do?
3. Cluster the worst moves into anti-goals ("we are pushing the user toward X by exploiting Y").
4. Flip every anti-goal into a Hero move and a guardrail.
5. Hand back: anti-goals named, Hero flips, and watch-outs for a future review.

Mode 3: Conceptualization

Use Hero / Anti-Hero pairs as generative prompts when shaping a new concept.

1. Frame the concept and the user value it claims to deliver.
2. Pick 2–3 Hero cards relevant to the value proposition (e.g., Empowerer, Liberator, Life Coach) and sketch a concept move that embodies each.
3. Stress-test each Hero move by naming the Anti-Hero version it could slide into under business pressure.
4. Adjust the concept to make the Anti-Hero slide harder or impossible (defaults, copy, incentives, dark-launch metrics).
5. Hand back: 2–3 concept directions, each with the Hero move, the Anti-Hero risk, and the design that prevents the slide.

Mode 4: Ethical dialogue

Help a team talk about an ethically charged design without finger-pointing.

1. Restate what is being discussed and what is at stake for the user.
2. Offer 2–3 Anti-Hero cards as candidate names for the move under discussion, with the wording the team can use ("this feels like a Trap-Setter — once a user is in, the off-ramp is hidden").
3. Offer the matching Hero card as a constructive counter ("the Empowerer version of this would be …").
4. Suggest a short script the team can use in the meeting itself ("I want to call a Trap-Setter on this — can we look at the cancel flow together?").
5. Hand back: candidate cards, dialogue scripts, and one or two follow-up questions to keep the conversation moving forward.

Workflow

Regardless of mode, follow this overall arc:

1. Detect the mode. If the user has not stated it, infer and confirm in one sentence.
2. Anchor in user evidence. Restate the user's task, their emotional state, and the business pressure pushing on the design. Without this, card naming becomes labelling theatre.
3. Pick the cards. Read references/cards.md and select cards that match the actual moves in the design. Two to five cards is usually enough — naming every card dilutes the critique.
4. Cite the move. For each card, point to the specific element, copy line, default, or step that triggers the tag. The cite is what makes the critique survive disagreement.
5. Pair with Hero counterparts. Every Anti-Hero tag should have at least one concrete Hero move that addresses it.
6. Mark unavoidable friction. Some Anti-Hero-shaped patterns are required (legal warnings, fraud checks, age gates). Call these out so the team does not strip them by accident.
7. Output the appropriate format for the mode (see below).

Output Format

Default to this layout. Trim sections that do not apply to the chosen mode.

1. Frame

• Design under review (or concept under development)
• User task and emotional state
• Business goal / stakeholder pressure
• Mode in use

2. Anti-Hero hits

A table with columns: card · what is happening · where in the design · why it is anti-user.

3. Hero counter-moves

A table with columns: paired card · concrete change · why it restores user value.

4. Mode-specific section

• Evaluation: risk note and "first three changes I would ship."
• Reverse brainstorming: anti-goals named, Hero flips, watch-outs.
• Conceptualization: 2–3 concept directions with the Anti-Hero slide each one risks.
• Ethical dialogue: candidate cards, dialogue scripts, follow-up questions.

5. Necessary friction kept

List anything Anti-Hero-shaped that should stay (regulatory, safety, fraud, accessibility).

Prompt Patterns

Use or adapt prompts like these:

• Evaluation: "Here are screenshots of our trial-cancellation flow. Run an Anti-Heroes evaluation and tell me which cards apply and what the Hero versions would look like."
• Reverse brainstorming: "We are designing a notification system for a finance app. Use Anti-Heroes to imagine the worst possible version, then flip each move into a Hero."
• Conceptualization: "We want to build an AI coach for new managers. Use Hero cards to shape three concept directions, and flag the Anti-Hero each one could slide into."
• Ethical dialogue: "My team is split on whether to add a default-on data-sharing toggle. Use Anti-Heroes to give us shared language and a short script for the meeting."
• Single-card lens: "Apply only the Empathy Manipulator card to this onboarding screen and tell me whether the move is present."

Guardrails

• Card names are diagnostic tools, not moral verdicts. Keep the language descriptive ("this acts as a Trap-Setter") rather than accusatory ("the designer is a Trap-Setter"). The source method explicitly frames this as designer reflection, and the deck works only if people can hear it without going defensive.
• Cite the move. A card claim without a specific UI element, copy line, default, or step is not credible and the team will reject it.
• Do not flatten unavoidable friction into Anti-Hero patterns. Legally required confirmations, fraud checks, age gates, and safety pauses can look manipulative but exist for the user. Mark them explicitly so they are not stripped in the rewrite.
• Avoid card spam. If every element gets tagged, none of the tags carry weight. Two to five cards per critique is usually right; pick the ones that matter most.
• Pair every Anti-Hero with a concrete Hero move. A critique with no flip is rant fuel, not design feedback.
• Hold space for genuine value tension. Some designs deprioritize user values for defensible reasons (cost, scale, regulation). Surface the trade-off honestly instead of pretending Hero moves are always free.
• Stay grounded in the source. The full deck and intended use are documented in Mehta, Chivukula, Gray, et al. (2024), arXiv:2405.03674. When in doubt, defer to the cards in references/cards.md rather than inventing new ones.

Quality Bar

A good output:

• names the user task and emotional state before tagging anything
• uses 2–5 cards, each with a specific cite from the design
• pairs every Anti-Hero with a concrete Hero counter-move
• distinguishes manipulative friction from necessary friction
• ends with something the team can actually do in the next sprint, not just labels

Scale guidance: for a single screen or flow, expect 2–5 Anti-Hero hits, 2–5 Hero counter-moves, and 1–3 friction-to-keep items. If the design is sprawling, scope the review to one journey at a time.

Example Output

Frame

• Design under review: free-trial cancellation flow for a fitness app
• User task: cancel before being charged; user is anxious and short on time
• Business pressure: trial-to-paid conversion is the team's top OKR this quarter
• Mode: Evaluation

Anti-Hero hits

Card	What is happening	Where	Why it is anti-user
Trap-Setter	Cancel button is buried under three nested menus	Settings → Subscription → Manage → "More options"	User has to discover the path; entry was one tap
Empathy Manipulator	"Are you sure? You'll lose your 14-day streak" with a sad-mascot illustration	Step 2 of cancel	Weaponizes loss aversion against a user who already decided
Nickeling-and-Diming	Offers a 50% discount only if user clicks "Stay" twice	Step 3	Makes the cheaper price contingent on staying, not on asking

Hero counter-moves

Paired card	Concrete change	Why it restores user value
Empowerer	Cancel link appears next to "Renews on …" wherever the renewal date appears	Symmetry with how user got in
Life Coach	Replace mascot with one neutral line: "Your data stays for 30 days if you change your mind"	Information, not pressure
Unveiler	Show the discount on step 1 as an honest option, not a reward for resisting	Transparency over coercion

First three changes I would ship

1. Move "Cancel subscription" to the top of the Subscription screen.
2. Cut the streak-loss illustration and copy.
3. Show the retention discount upfront, framed as "stay at 50%" or "cancel — both fine."

Necessary friction kept

• Two-step confirm for the actual cancel click stays — prevents accidental cancels and is standard for paid services. Just keep the steps neutral.