Every Open Mic · metro core

How we look

104 ways we look for a mic — and growing. The number of ways is a lower bound, never fixed; we add new ones as we learn them.

47 running now 20 being built 37 planned

Coverage below = how many of our 6899 venues we've actually run a method on. Every gap is a place a mic could be hiding — there is no "done".

How we find them

Venue website 7/12 running

Homepage keyword scanlivediscover 1.0% (70)

limits: homepage HTML only; misses JS, sub-pages, social

Events / calendar / what's-on sub-pageslivediscover not run

limits: only links we guess; JS calendars need render

schema.org/JSON-LD Event markuplivediscover not run

limits: most grassroots venues emit none (Finding n5)

Calendar plugins (Tribe wp-json, Squarespace ?format=json, Wix, Tockify, Google-Cal embed)livediscover 11.4% (789)

limits: per-platform probes

Linked iCal/.ics feedlivediscover not run

limits: rare on small venues

RSS/Atom feed (blog/events)livediscover not run

limits: may be news not events

sitemap.xml event-URL mininglivediscover not run

limits: discovers URLs, still needs extraction

Site blog/news postsbuildingdiscover not run

limits: unstructured; may be stale

Menu / weekly-specials / flyer PDFs ('Tue: Open Mic')buildingdiscover not run

limits: PDF/image OCR needed

Embedded event / ticketing widgetsbuildingdiscover not run

limits: artist-events only

Headless-render JS-only event pagesplanneddiscover not run

limits: RAM-heavy; reserve for known-dynamic sites

QR-code flyer → landing page (decode + follow)buildingdiscover not run

limits: QR-decode into the OCR path

Venue social 7/9 running

Instagram posts/Reels (venue account)livediscover not run

limits: throwaway acct; flyers need vision

Instagram saved Highlights (often an 'Open Mic' highlight)livediscover not run

limits: throwaway acct

Instagram hashtags (#pdxopenmic #portlandcomedy …)livediscover not run

limits: discovers new hosts, not venue-specific

Facebook Page postslivediscover not run

limits: login-walled; throwaway

Facebook Events (incl. recurring)livediscover not run

limits: no public API; throwaway read

Facebook Groups (local music/comedy/poetry groups)buildingdiscover not run

limits: BLOCKER: burner FB account (Phase-2); richest grassroots layer; search-reach partial

TikTok (venue/host clips, on-screen flyer text)livediscover not run

limits: yt-dlp metadata + OCR

X/Twitter postsplanneddiscover not run

limits: low local value

Mastodon / Bluesky cross-postslivediscover not run

limits: most reliable social feeds (RSS-able)

The host 1/4 running

The HOST's own IG/socials (where the mic usually lives)buildingdiscover not run

limits: must first identify the host

Host's Linktree/Beacons (the actual schedule)livediscover not run

limits: single most reliable once found

Comedian/musician profiles ('I host the Tue mic at X')planneddiscover not run

limits: manual research

Comedian/musician personal sites ('I host the {night} mic at X')buildingdiscover not run

limits: search-reach

Link-in-bio 1/1 running

Linktree/Beacons/Carrd parse (__NEXT_DATA__)livediscover not run

limits: must have the bio URL first

Aggregators/directories 1/1 running

Public event directorieslivediscover 0.6% (41)

limits: 21 sources, cross-referenced; never a primary source

Google/structured 2/4 running

Google Events box (SERP 'open mic near Portland')buildingdiscover not run

limits: no API; ToS-grey scrape

Google Places attributes (live_music, popular-times, type)plannedenrich not run

limits: BLOCKER: Places API key (cred-gated); no build

OSM tags (live_music, amenity, capacity)liveenrich not run

limits: undertagged

Geocoded (Photon/Nominatim)liveenrich 1.2% (84)

limits: coordinates only

Reviews/UGC 3/9 running

Google Reviews text ('great open mic Tuesdays')buildingdiscover not run

limits: BLOCKER: Places API key (cred-gated) for live; parser + search-reach subset exercised

Review-site tips mentioning open micbuildingdiscover not run

limits: BLOCKER: Yelp Fusion API key (cred-gated); search-reach subset exercised

Travel-review sitesplanneddiscover not run

limits: low density

Community forums & comment threadslivediscover 0.5% (32)

limits: thin/noisy; PRAW/RSS

Check-in app tipsplanneddiscover not run

limits: fading

Craigslist community/events open-mic postslivediscover not run

limits: reach-only; noisy

Bandcamp/SoundCloud/Mixcloud 'open mic at X' recordingsbuildingdiscover not run

limits: lagging signal

Local music/comedy/poetry Discord servers (disboard/discord.me search)livediscover not run

limits: reach-only; pinned schedules

Nextdoor neighborhood posts/eventsbuildingdiscover not run

limits: reach-only

Direct/human 2/2 running

Someone submits/tells uslivediscover not run

limits: needs a submit path

Agent canvass sweep (legacy)livediscover 3.8% (260)

limits: one-time sweep; see checks detail

Indirect signals 1/5 running

Locally-maintained scene spreadsheets/Google-Sheetsbuildingdiscover not run

limits: find via community

Karaoke/trivia/'weekly events' page co-occurrence (same page lists the mic)buildingdiscover not run

limits: —

City entertainment/sound-permit data (has a stage/live perf)plannedenrich not run

limits: capability not schedule

Stage/PA present (photos, OSM, reviews) — capability priorplannedenrich not run

limits: prior, not proof

Ask an AI answer-engine 'open mics in {city}' as a lead-generator (= the catch-all agent itself)livediscover not run

limits: agent-rule; leads then verify

How we verify them

Verification is its own phase — a discovered mic is marked for verification before it's shown as confirmed. These are the ways we confirm one is real and current.

Venue website 0/1 running

Wayback Machine history (did it run before? seasonal?)planned not run

limits: past ≠ present

Venue social 0/3 running

Instagram Stories ('tonight!' reminders)planned not run

limits: 24h ephemeral; must poll

Instagram location-tag / tagged posts (attendees posting #openmic AT the venue)building not run

limits: noisy

YouTube ('open mic at X' uploads, venue channel)planned not run

limits: lagging signal

The host 0/1 running

Host's email newsletter / Substackplanned not run

limits: subscribe; manual

Google/structured 0/1 running

Google Business Profile (posts / events / Q&A / 'from the business')building not run

limits: scrape the GBP panel

Reviews/UGC 1/3 running

Local music/comedy blogs + news articleslive not run

limits: search-driven

Local scene podcast mentionsplanned not run

limits: manual

Map/review photos showing a stage / 'open mic' chalkboard / signup sheetplanned not run

limits: vision OCR

Direct/human 1/9 running

Phone the venue ('do you host an open mic? what night?')planned not run

limits: human/voice; authoritative

Email the venue/hostplanned not run

limits: courteous; ADR-0007

DM the venue/host on IG/FBplanned not run

limits: throwaway acct

In-person visit / walk-by (stage, sign, chalkboard)planned not run

limits: human; gold-standard

Ask the host directlyplanned not run

limits: human

Ask a regular/performer who attendsplanned not run

limits: human; scene knowledge

Google Street View signage (a posted recurring sign)planned not run

limits: time-stamped; may be old

Meticulous agent research (the catch-all deep facet)live 4.1% (284)

limits: the subscription-driven human-in-loop

Venue confirmation (legacy)planned 0.0% (2)

limits: —

Corroboration 0/8 running

≥2 independent sources agree (day/time/venue)planned not run

limits: dedupe by origin

Day/time consistent across sourcesplanned not run

limits: conflicts → MoE check

How recently each source confirmed itplanned not run

limits: freshness decay

Run consistently for N weeks/monthsplanned not run

limits: needs time-series

Recurring pattern confirmed across multiple weeksplanned not run

limits: —

Same-day 'still on tonight?' volatile-source re-checkplanned not run

limits: cheap, day-of

Posts/photos FROM the mic on the expected night (live proof)planned not run

limits: throwaway acct

Explicit 'no more open mic / cancelled' detectionplanned not run

limits: decreases confidence

Indirect signals 0/1 running

Online performer sign-up lists/formsbuilding not run

limits: strong if found

Retired: ~~Local-model read of captured website text (surface_candidates)~~ · ~~Local-model read of captured Instagram text (surface_candidates)~~ · ~~Local-model read of captured web+IG text (surface_candidates)~~ — superseded by direct agent reasoning (kept for history).