Can a city sustain a vertically-integrated alternative to global, attention-extractive social media, and is the form factor replicable?
MyBridge is a hyperlocal community application built for the city of Lethbridge, Alberta. It bundles five surfaces that mainstream platforms typically split across separate apps: a social feed, a peer-to-peer marketplace, a local business directory, a custodial Web3 wallet on NERA, and a gamified participation layer. It also runs as the engagement layer for HLIC 2026, where the same product becomes a conference app with booth scanning, peer scans, session check-ins, reward polls, and voucher redemption. The thesis question: whether a hyperlocal, vertically-integrated social-commerce platform can sustain itself culturally and economically in a market dominated by horizontally-scaled attention platforms.
Mainstream platforms (Meta, X, TikTok, Snapchat, YouTube) optimize for global attention at scale, monetized through algorithmically-targeted advertising. The unit economics require a feed-ranking system that maximizes time-on-app.
Identity is platform-owned: handles are tied to phone or email, effectively non-portable, and account loss is indistinguishable from identity loss.
Moderation scales by machine because human review cannot keep up; appeals processes are infamously broken because the cost-per-decision must stay low.
Local commerce sits across disconnected silos (Instagram, Google Business, Yelp, the merchant's own site), none of which share a graph with the consumer.
Verifiable participation is absent: "I attended X event" is self-report; check-in data the user generates is data the user cannot export or own.
Every user, business, and event is anchored to one city. Discovery is intentionally bounded to local relevance. Directory, deals, marketplace, and event feeds are all geographically pre-filtered. There is no cross-cultural virality objective.
No third-party advertising. Revenue paths are business subscription tiers (free / pro / premium), ticketed events, and conference booth claim codes. Users earn instead of paying: Bridge Points accrue from participation and are exchangeable for on-chain rewards.
Identity is an NNS (Nera Name Service) name backed by a wallet address on NERA. Custodial at signup to keep the on-ramp frictionless; self-custody is an option later. The platform stores convenience metadata but does not "own" the handle the way Instagram owns @username.
Feed tabs are simple and documented in code: For You (30% trending interleaved with 70% recent), Latest (pure reverse-chronological), Following (followed users and businesses only), Live, Reels, Polls. No opaque preference inference, no shadow ranking.
Events have physical locations, ticket URLs, and QR-based check-ins. Conference booths are real tables with printed claim codes. Deal redemption is in-store or deep-linked online. Peer scans link two attendees at an event. Every key digital action is back-stopped by something the user did in the physical city.
Event attendance is minted as an NFT on NERA at check-in time (optionally AI-generated artwork in PRO mode). The user holds cryptographic proof that survives the platform indefinitely: a passport stamp rather than a temporary wristband.
Live broadcasts support direct on-chain tipping in NERA tokens. Each tip is a recorded transaction with a NERA hash. Creators own an audit trail of their earnings that exists independently of the platform.
Following a business surfaces their posts in the user's Following feed exactly the same way following a person does. The same identity is producer and consumer; the same graph powers social and commerce.
Content is moderated at submission (a text moderation pipeline checks posts, reviews, and chat) and again at the community layer (user reports route to an admin queue with contextual snapshots). Because the user population is bounded by city size, human review remains a tractable operation rather than a machine-only scale problem.
Next.js App Router with React Server Components for public detail pages; client-side interactivity layered over top.
Single-container Docker deploy on a DigitalOcean droplet behind Cloudflare. Vertical scale, not horizontal: a deliberate rejection of the scale-first assumption baked into most social platforms.
WHIP-to-RTMPS via MediaMTX on the droplet for live streaming. Direct Cloudflare WHIP cannot produce HLS or recordings reliably on cellular carriers; a bridge box was the pragmatic answer. Glass-to-glass latency around five seconds.
Custodial-by-default Web3 wallet. The on-ramp friction of "manage your seed phrase before signup" excludes most users from crypto-flavored social platforms; MyBridge accepts the custodial tradeoff at signup and unlocks self-custody later, when the user actually wants it.
Denormalized read counters (ratingAvg, followerCount, totalDealViews) recomputed on every write. A deliberate departure from event-sourcing in favor of simpler operations for a small team.
In-memory rate limiting with tiered budgets (read, write, scan, auth). Trades multi-instance consistency for operational simplicity.
CASL-compliant email broadcast layer. The platform operates in Canada and is bound by Canada's Anti-Spam Legislation; consent-tracking and one-click unsubscribe are first-class.
HLIC 2026 field test: observe whether real conference attendees adopt the app as the event's operating layer (booth scans, peer scans, session check-ins, reward redemption).
Measure cross-tab behavior: does presenting both algorithmic and documented feeds change time-on-app vs reported satisfaction?
Test the secondary thesis: is the form factor replicable? If MyBridge works in Lethbridge, instance the same codebase for a second city and compare adoption curves.
Self-custody migration UX: when does a custodial user opt into managing their own keys, and what does that conversion funnel look like?
Open the moderation pipeline: publish the text moderation rules used at submission, so users can see why something was filtered.