[B7] Social adjacent: taste affinity, side-by-side ratings, following a list

Backlog. Social features that came out of the gamification discussion but are not gamification themselves. All gated behind SOCIAL_ENABLED. ## Ideas parked here - Taste affinity score between two users, computed from ratings on works they have both rated. Sticky and genuinely useful, and non-competitive — but expensive to compute well. Possibly a premium-profile stat (see [B9]). - Side-by-side ratings on a work page: your friends' ratings next to yours ("you 8 · Marc 4 · Julie 9"). Simple, and it makes the social layer feel alive. - Following a list. Does not exist today — a list can be shared but not followed. It is a prerequisite for any "list followed by N people" achievement, and useful on its own. ## Note Review is already the single source of truth for ratings (ReviewService.getRatings), which is what makes the first two feasible at all.

[B6] Collection and saga completion (TMDB collections)

Backlog, viability to be confirmed. TMDB exposes a film's collection membership (belongs_to_collection): Alien, Star Wars, Fast & Furious. That gives a bounded, non-farmable object to complete. ## Why it is interesting - A "5/9" gauge on a saga is a natural discovery objective, not a consumption counter. - It answers the completion problem directly: a saga is a bounded set, so completing it means something, unlike "I finished one series". - It is as much a catalogue feature as a gamification one — the gauge is useful even without an achievement attached. ## To check - Coverage and reliability of the TMDB collection data. - Equivalents in the other domains (game series via IGDB, book series via Open Library) — probably much weaker. - One achievement per saga would flood the catalogue; aggregate them ("3 complete sagas"). Depends on: [G2]/[G3] achievement system.

[B5] Advanced progression: per-domain levels, leaderboard seasons, multi-cadence streaks

Backlog. Extensions to the progression system once the MVP is calibrated in the wild. Depends on: [G1] XP ledger, [G7] Leaderboard. ## Ideas parked here - Per-domain levels alongside the global one (cinephile 4 / reader 2 / player 1). Gives an RPG-like "skill" dimension, lets someone dominate one domain without leading overall, and keeps the global level from flattening identity. Recomputable, but harder — the XP has to be attributable per domain from the start, which is worth checking in [G1]. - Leaderboard seasons with an archived podium and an achievement for placing. Prevents early adopters from locking the top forever, and creates a recurring event. Recomputable from the ledger, so the MVP (month/year filters) does not close this door. - Multi-cadence streaks. A daily streak only really makes sense for video; for books and games the honest unit is the week. One mechanic, two cadences depending on the domain. Calibration is the hard part and is not obvious — parked deliberately. ## Related constraint Real playtime is not measured outside the Steam import, which limits any game-based cadence.

[B4] Challenges, mini-games and easter eggs

Backlog. Everything opt-in and playful that rewards XP. Depends on: [G1] XP ledger. ## Ideas parked here - Challenges. A challenge is an objective plus a completion rule ("12 countries in 12 months", "one film per decade", "clear these 20 games"). Reward is XP, so it plugs straight into the level system. Implementation option to evaluate: a challenge is essentially a List + a rule, rather than a new primitive. Shape: one shared monthly challenge for everyone, plus one extra actionable challenge reserved for premium (see [B9]). - Daily mini-game — a Wordle-style media guessing game or a logic puzzle. This replaces the idea of daily quests, which is rejected: "rate something today" is pure habit manufacturing, disconnected from actually watching or reading anything, and it pushes users to log junk that pollutes the data. - Easter eggs, some of them unlocking secret achievements (the secret mechanic itself is built in [G2]/[G3]). - "Séance à l'aveugle": the app proposes a work deliberately outside your habits; watching it earns XP. Discovery and play in one gesture, but it depends on a recommendation engine that does not exist yet — far out. Challenges are not a near-term priority.

[B3] Recaps and comparatives: yearly Rewind, monthly retro, anonymous percentiles

Backlog, explicitly late. ## Ideas parked here - Yearly Rewind. Your year in numbers, top rated, the genre that exploded, longest streak, plus a shareable image. High potential, but it demands real design work and, above all, accounts with several months of history — which does not exist yet. Premium candidate (see [B9]). Should be one of the last features, not one of the first. - Monthly retrospective. Lighter version, pushed through the existing NotificationDigestService infrastructure. Lower enthusiasm, but it is the incremental path toward the Rewind. - Anonymous percentiles — "you are in the top 10% of documentary watchers". All the satisfaction of comparison, none of the leaderboard's effects: no name, no race. Complements [G7] rather than duplicating it. ## Existing groundwork The stats/ module already computes most of the aggregates: video-temporal, rating-distribution, decade, cross-domain-totals.

[B2] Sharing and social proof: member card, public hall of fame, referrals

Backlog. Everything that makes progression leave the app. ## Ideas parked here - Shareable member card. A generated image of the profile — level, 3 pinned badges, top works, seniority — in a cinema membership-card register. To be folded into the existing profile share button: instead of showing only a QR code, show a shareable profile card (with or without the QR embedded). This is the Rewind for a product with no history yet: it works after one month of use. - Public hall of fame for the hosted instance (reviewers of the month, active profiles). Worth nothing until there is a user base — hosted only, well after launch. - Rewarded referrals. Real growth lever, but needs an invitation flow, and it mixes growth with play — the front door for fake accounts polluting the leaderboard. Probably only after premium ships. ## Dependencies Profile share button (exists), [G4]/[G5] for what the card displays.

[B1] Cosmetic rewards: equippable titles, themes, avatar frames

Backlog. The non-inflationary reward currency: unlike XP, you can add a lot of these without unbalancing anything, and it fits an app that refuses to be a points game. Depends on: [G2]/[G3] achievement system. ## Ideas parked here - Equippable titles shown next to the pseudo. Explicitly not one title per achievement — a title is a rare item, rarer than a badge. The 3-badge showcase already covers "show what I did"; titles cover "show what is hard to get". Membre fondateur (see [B9]) is the ultra-rare example. - Alternative themes / accents. The "Séance" identity is built on a single ownable accent (projector amber); alternates are a natural unlock, and a clean premium candidate since they are purely cosmetic. - Avatar frames, profile banners. ## Rule No achievement is ever behind a paywall. Cosmetics may be; rewards themselves may not.

A button to add an item directly from a list

Today, its not possible, on a list view, to add an element. Add a plus button to show a search bar displaying content on a popover ? see design frontend skill to determine best choice for UX UI.

Rework Games / Books details page

Native push notifications (Capacitor)

Native mobile push notifications via Capacitor, to complement the PWA. Currently blocked on the dev side: no mobile toolchain available.

Jellyfin/Plex/Trakt integration (webhook / scrobble)

Automatically mark an episode or movie as watched from a Jellyfin/Plex webhook or a Trakt scrobble, instead of manual tracking. Strong potential as a self-hosted "TV Time replacement".

Podcast tracking

Add podcast tracking via the iTunes Search or Podcast Index API. The domain already exists in the data model, pending implementation on the catalogue and screens side.

Board game tracking

Add board game tracking via the BoardGameGeek API (free, no key required). The domain already exists in the data model, pending implementation on the catalogue and screens side. Backloggd (video games) has no official export yet — shouldn't be built against scraping.

Undo a destructive action

After a frequent irreversible action (removing a title from your library, deleting a review or a list), show a toast with an "Undo" button for a few seconds, like Gmail/Linear/Notion.