Psychology & Persuasion Framework
The other frameworks catalog what a platform does and how it's built. This one answers a
narrower, sharper question: what specific psychological mechanism is each feature exploiting,
and is that mechanism disclosed/benign or a genuine dark pattern? It formalizes what the
ux-ui-framework.md glossary already gestures at (Fitts's Law, loss aversion, etc.) into three
named, citable frameworks instead of a loose vocabulary — so claims are traceable to an actual
body of research, not vibes with academic-sounding labels attached.
Purpose and framing, stated up front: this is a diagnostic/audit capability — naming and explaining mechanisms that are already live in shipped products, the same way a security researcher documents a known vulnerability class. It is not a manipulation playbook. Every platform section should read like an audit, not a how-to. If a request is actually asking "how do I make my app more manipulative/addictive," that's out of scope for this skill — redirect toward the ethical, disclosed side of behavior design (the Hook Model is a legitimate product framework used for both honest and dishonest products; the dark-patterns section exists specifically to name where a feature crosses from the former into the latter).
Produce psychology-and-persuasion.md when the request has any psychology/persuasion/addiction/
dark-pattern/manipulation/"why do I keep scrolling" angle — this is likely the single
most-requested angle for a general audience, given how often "what makes these apps addictive"
comes up as a trigger phrase for this skill in the first place.
1. Cialdini's Six Principles of Persuasion
Robert Cialdini's Influence names six levers that reliably move behavior. For each principle,
identify the specific, already-observed feature that implements it — don't just assert the
principle exists, point at the exact mechanism from feature-catalog.md/ux-ui-catalog.md.
| Principle | What it is | What to look for |
|---|---|---|
| Reciprocity | People feel obligated to return a favor | Free tiers, "free award" mechanics, algorithmic gifts (a platform surfacing a rare positive signal, like a "your post is performing well" alert) that create a sense of being given something |
| Commitment & Consistency | Once committed (even publicly, even in small ways), people act consistently with that commitment | Streaks, karma, profile-completion meters, public follow/join actions, "Open to Work" — any mechanic where a small public act creates pressure to keep it up |
| Social Proof | People look to others' behavior to decide their own | Like/comment/view counts, "X and 11 others," follower counts, "trending," aggregate reaction breakdowns |
| Authority | People defer to credible, credentialed sources | Verification badges (paid vs. earned — note which kind, since that changes what "authority" is actually being signaled), admin/mod distinguishing marks, official/verified organization treatment |
| Liking | People are persuaded by people/things they like | Personalized "who to follow," algorithmic content matching to existing taste, avatar/identity customization (Snoovatar, profile photo systems) |
| Scarcity | People value what's rare or time-limited | Limited-time notes/awards, "X people viewed your profile in the last 90 days" curiosity gaps, ephemeral content (Stories), countdown-style edit windows |
Also note Unity (Cialdini's later-added seventh principle — shared identity, "people like us") where relevant: community/subreddit membership, tagged group identity, platform-exclusive in-jokes/formats.
2. The Hook Model (Nir Eyal, Hooked)
Trace the actual habit loop per platform as a cycle, not a list — the output should read as "trigger leads to action leads to reward leads to investment which primes the next trigger," using real, already-documented features at each stage:
- Trigger — external (a push notification, a badge count) and internal (boredom, FOMO, social anxiety the product has learned to prompt). Name the specific notification categories or re-engagement mechanics already found in the feature catalog.
- Action — the simplest behavior done in anticipation of reward (opening the app, one tap to refresh/scroll). Tie to the platform's actual entry-point friction (or lack of it).
- Variable Reward — Eyal's three flavors are useful to distinguish explicitly:
- Rewards of the Tribe — social validation (likes, comments, reactions)
- Rewards of the Hunt — the unpredictable content itself (an algorithmic feed's next-post unpredictability, a "you never know what you'll find" scroll)
- Rewards of the Self — mastery/completion feelings (karma milestones, streak numbers, profile-completion, daily-puzzle scores)
- Investment — an action that increases the odds of the next trigger and improves the service for next time: content posted, a follow made, a profile field filled in, karma earned, a streak maintained. This is the stage that makes the loop self-reinforcing rather than a one-off — always close the loop by naming what investment feeds back into a future trigger.
3. Dark Patterns Audit
Based on Harry Brignull's deceptive-design taxonomy (deceptivedesign.org) — a real, named, widely-cited classification, not an invented one. Work through each category and state plainly whether the platform does or does not exhibit it, with the specific feature as evidence. Absence is worth stating too — don't force a finding into every category.
| Category | Sub-patterns to check for |
|---|---|
| Sneaking | Hidden costs revealed late, forced continuity (hard-to-cancel subscriptions), sneaking an action into a flow the user didn't ask for |
| Urgency | Countdown timers, "act now" framing, artificially time-boxed offers |
| Misdirection | Confirmshaming (guilt-tripping copy on a decline option), visually asymmetric CTA weighting (one option styled as obviously primary, the other de-emphasized to the point of being hard to find), trick questions, pressured upselling |
| Social Proof (misused) | Fake or unverifiable activity notifications, testimonials of uncertain origin — distinct from honest social proof (a real like count), which is not a dark pattern on its own |
| Scarcity (misused) | Fabricated low-stock/high-demand messaging — distinct from genuinely time-limited features |
| Obstruction | Roach motel (easy to get in, hard to get out — e.g. a multi-step cancellation/deletion flow vs. a one-tap signup), hard-to-find privacy controls, comparison prevention |
| Forced Action | Forced enrollment (having to create an account/give data to access basic function), nagging (repeated interruption after a decline — e.g. an app-install modal that reappears every navigation despite being dismissed), gamification designed to obscure rather than genuinely reward |
Calibration note: not every persuasive design choice is a dark pattern. A large, high-contrast
primary CTA next to a smaller secondary option is completely normal visual hierarchy (Fitts's
Law, ux-ui-framework.md) — it becomes a dark pattern specifically when the asymmetry is
designed to produce a mistake or extract a non-consensual action, not merely when one option is
more prominent than another. Keep this distinction explicit rather than calling every design
choice manipulative.
Output structure
For each platform's psychology-and-persuasion.md:
- Cialdini's Six — table or short sections, each principle mapped to real evidence
- The Hook Model — the full trigger → action → reward → investment cycle, narrated as a loop
- Dark Patterns Audit — worked through every Brignull category, findings and honest absences
- What's disclosed vs. what isn't — a short closing take: which of this platform's persuasion mechanics are things the platform has been open about (e.g. Reddit's own anti-gamification messaging sitting next to its karma system) vs. mechanics that are only visible through direct observation, never acknowledged by the platform itself