Railgun Streaming Plan

Companion to the Railgun 2026 Master Plan. That plan sets revenue-first strategy across all five product families; this document is the deep operating manual for Railgun Live — the audience-acquisition engine at the top of the funnel. Rendered publicly at plan.railgun.chat/streaming-plan.

The decision: Build one anonymous media brand—Railgun Live—as a live show, clipping machine, and owned-community funnel. Broadcast one program feed everywhere, but concentrate editorial effort on YouTube, short-form discovery, and the Railgun-owned audience. The platforms are rented distribution; Railgun's audience, email list, community, and products are the asset.

0. Expectations at a glance

Read this first, then read the whole plan before trusting any single number. These are operating scenarios, not promises. The unconditional outcome for a brand-new anonymous channel is worse than the pessimistic column; every figure below assumes the team actually ships 16 quality streams and ≥64 approved clips every month without a single identity leak.

Two-year optimistic ceiling: ~2.48M followers, ~234M monthly views, ~62K CCV, ~$1.4M/month. Two-year pessimistic floor (full cadence, slow learning, no breakout): ~19,800 followers, ~1.4M monthly views, ~212 CCV, ~$14K/month.

Where the money comes from (and the goal to de-risk it): Month 6 is ~35% direct supporters, 25% platform ads/features, the rest sponsors/affiliates/product. By Month 24 no single platform should exceed 25% of attributable revenue, and ≥30% should be Railgun product gross profit (Pro at $8/mo·$80/yr, Operator at $20–30/mo, Meets, paid groups at a 10% platform fee). Product-adoption target by Jul 2028: 3,000–7,500 paying Railgun users on a 30,000-person owned email list ($24K–$80K product MRR-equivalent).

The point of the whole exercise: streaming earns attention cheaply and builds trust that a brand-new privacy product cannot buy. Attention converts to an owned audience; the owned audience becomes Railgun's first paying users. Platform revenue is a bonus, not the foundation — the foundation is the list, the community, and the products. See §9–§10 for the full monetization ladder and the month-by-month 24-month table.

1. What Railgun is building

Railgun Live should feel like friends with sharp opinions sitting around a digital war-room table: funny, candid, informed, occasionally chaotic, but never careless. The closest format analogue is Asmongold-style topic and reaction streaming, not his visual identity or personal-life exposure.

The show has three jobs:

1. Attention: earn reach through timely live commentary and aggressively distributed clips.
2. Trust: convert occasional viewers into repeat viewers, registered community members, email subscribers, and paying supporters.
3. Adoption: let that trusted audience become the first users, testers, advocates, and customers of Railgun's privacy products.

The operating doctrine remains the one already established for Railgun:

• The Sovereign Core protects communication, evidence, and peaceful coordination.
• The Business/Services layer sells performance, reliability, storage, creator tools, and convenience.
• The User Sovereignty Boundary means Railgun cannot casually read or exploit users' private material.

Streaming comes first because great products without distribution are expensive secrets.

2. Non-negotiable strategic choices

Why this platform order is rational

The live market is enormous but concentrated. Streams Charts measured 31.4 billion hours watched in Q2 2026; YouTube alone represented 49.5%, while Kick continued to close the gap with Twitch and Rumble remained more volatile and news-cycle-dependent. This supports a simple conclusion: YouTube receives the most care, Kick is a meaningful growth/monetization bet, Twitch is valuable live culture, Rumble is useful for political/news audiences and free distribution infrastructure, and X is a real-time discovery layer—not the archive. Streams Charts, Q2 2026

3. Broadcast architecture

Rumble Studio's Direct RTMP mode accepts a feed from OBS and can multicast it to multiple destinations; its documentation explicitly lists YouTube, Twitch, X, Facebook, and custom destinations such as Kick. It also supplies aggregate chat and health monitoring. Rumble Direct RTMP documentation

Launch configuration

• Canvas: 1920×1080, 30 fps.
• Video: H.264, CBR, 8 Mbps, keyframe every 2 seconds.
• Audio: AAC stereo, 48 kHz, 160–192 kbps.
• Normal delay: 20–30 seconds.
• High-risk/tentpole delay: 60–90 seconds.
• Local recording: MKV with separate program, voice, browser, and music tracks; remux to MP4 after the stream.
• Backup: a prepared “technical difficulty” scene, global audio mute, screen-hide hotkey, and a backup mobile connection known only to the producer.

Do not put a merged cross-platform chat overlay into the shared feed. Twitch permits simulcasting, but its written rules still require a Twitch experience at least as good as elsewhere, prohibit pushing Twitch viewers to a concurrent outside stream, and have historically restricted combined on-screen activity. A private multichat dashboard is fine; keeping it out of the program feed is the least fragile policy. Twitch simulcasting guidance

4. Cast and operating roles

These are recommended operational roles, not permanent creative cages.

Only two or three personas should be vocal in an ordinary segment. A five-person permanent panel becomes a traffic jam with avatars. Rotate pairings; use the full cast for election nights, debates, major cultural events, and monthly roundtables.

5. Software each person needs

Every cast member

Pitch/formant shifting frustrates casual voice recognition but is not magic. At scale, speech cadence, vocabulary, recurring stories, room acoustics, and friends recognizing verbal habits can still identify someone. The stronger long-term option is a locally trained, commercially licensed real-time voice model that does not imitate a real person. Do not use a bundled model whose license forbids political or religious speech.

Producer station

Automation after the manual process works

Do not automate a bad workflow faster. For the first 30 days, human beings select and approve every clip. Once the winning format is known:

1. OBS writes a local master and timestamp markers.
2. An ingestion worker submits the VOD or an approved public URL to the OpusClip API.
3. Candidate clips enter a review queue, never straight to the public feed.
4. A human approves the hook, source attribution, caption accuracy, crop, and privacy safety.
5. The Buffer API schedules platform-native variants with unique captions and tracked links.
6. Performance returns to the Railgun dashboard and updates the topic/host scorecard.

6. Anonymity and anti-doxxing system

What “anonymous” realistically means

Railgun can keep the cast's legal identities, faces, exact locations, home addresses, family, and daily routines hidden from the audience—even at large scale—if the team treats privacy as an operating system rather than a costume.

It cannot be anonymous from the regulated companies that process payouts, identity checks, taxes, subpoenas, or banking. YouTube/AdSense, X/Stripe, Kick, Twitch, and Rumble may know the account owner's legal entity and responsible person. That information must be accurate and kept out of public profiles and receipts.

Identity compartments

1. Legal compartment: entity documents, tax IDs, bank, platform KYC, contracts, registered agent, accountant, and counsel. Access: Ravioli plus one designated backup only.
2. Operations compartment: channel credentials, production calendar, sponsorship inbox, unpublished VODs, analytics. Access: minimum necessary staff.
3. Persona compartment: public accounts, avatar assets, transformed voices, public biography. Contains no legal identities.
4. Community compartment: Discord/public chat. Treated as hostile input. No internal files, private schedules, recovery data, or raw media.

Never create a cloud document called “Ravioli = [real name].” The identity map belongs in an encrypted offline record held by counsel or the responsible company officer.

Rules that prevent most leaks

• Never stream a desktop. Capture one sanitized browser window or a pre-approved media source.
• Hide system clock, language/keyboard indicators, weather, battery, Wi-Fi name, bookmarks, download names, calendar, notifications, and file paths.
• Never open viewer-supplied files or links on the production machine. A researcher opens them in an isolated disposable browser/VM and sends the producer a clean screenshot or approved URL.
• Strip EXIF and document metadata from every upload. Export through the media pipeline; never post an original phone photo.
• No live IRL footage, windows, reflections, street noise, delivery labels, skylines, hotel rooms, car interiors, schools, gyms, churches, or recurring local businesses.
• No “I am here right now” stories. Personal stories are delayed, generalized, composited, or omitted.
• Do not reveal exact age, birthday, school, employer history, family names, immigration history, distinctive legal cases, or simultaneous private events.
• Do not use an old username, gamer tag, email, profile picture, catchphrase, or avatar asset connected to a real identity.
• Do not follow real friends/family from persona accounts or let them follow early in a recognizable cluster.
• All public gifts, tips, shipping, and returns go through a company-controlled commercial mailbox or fulfillment provider—never a residence.
• Each account has a recovery sheet, both hardware keys, last-known-good backup codes, and an owner. Quarterly access review is mandatory.
• Twice a year, hire an independent red-team researcher to try to identify the cast from public information. Fix what they find before a hostile audience does.

Data-broker removal

For U.S.-exposed members, use one high-coverage primary service rather than blindly stacking three providers that send conflicting duplicate removals. Optery Ultimate currently lists automated coverage of 635+ sites, unlimited custom requests across 950+ sites, search-engine outdated-content submissions, and human review at $249/year. Optery pricing

For members outside its supported regions, select a provider that legally serves the person's actual jurisdiction, then run a quarterly manual search in English and local languages for legal names, aliases, phones, emails, past addresses, and image matches. Privacy removal is continuous gardening, not an exorcism.

Incident levels

No host improvises an incident response live. Silence is not weakness; it is containment.

7. Platform and monetization facts as of 1 August 2026

YouTube currently pays 70% of net fan-funding revenue, 55% of net watch-page ad revenue, and 45% of the creator's allocated Shorts pool. These are platform shares, not guaranteed RPMs. YouTube earnings

Reaction-content rule

Railgun must be a commentary show that uses material, not a pirate feed with occasional grunts. YouTube explicitly permits reaction videos with meaningful commentary and critical review, but its reused-content policy can demonetize a whole channel when the original contribution is minimal. Fair use is case-specific and only courts finally decide it; disclaimers and attribution are not magic shields. YouTube monetization policy, YouTube fair-use guidance

Operationally:

• Prefer primary documents, public-domain/government material, licensed footage, screenshots, headlines, and short necessary excerpts.
• Pause every excerpt to analyze; do not let another creator's full video run while the cast watches silently.
• Mute music and sports/TV footage unless licensed.
• Keep a source log for every segment and clip.
• The avatars are real humans protecting their privacy. Never market them as autonomous AI experts; YouTube bars AI personas that present sensitive political, legal, financial, or health advice from monetization.

8. Content engine

Editorial lanes

Daily topic radar

Osased maintains a queue from:

• primary sources: official statements, legislation, court documents, company filings, product changelogs, election authorities;
• reputable wire/news reporting;
• Google Trends, YouTube search suggestions, X lists/trends, Reddit, TikTok discovery, Twitch/Kick categories, and competitor streams;
• community questions and comments;
• the team's real expertise and arguments—not generic trend summaries.

Every candidate receives a 0–5 score:

Topic score = 25% velocity + 20% audience fit + 20% disagreement/tension + 15% clipability + 10% evidence quality + 10% shelf life – legal/privacy risk penalty

Only topics scoring 65/100 or higher enter a normal show. A weaker topic can enter only when it deepens a winning lane or matters strongly to the community.

Response speeds

• Tier A — event now: producer decision in 10 minutes; go live within 30–60 minutes if the cast has facts and a point.
• Tier B — same-day story: researcher packet in 90 minutes; segment that evening.
• Tier C — evergreen: 24–72 hours of research; designed as a 10–20 minute YouTube video first, live segment second.

Standard three-hour show

1. 00:00–00:03: cold open with the strongest claim; no housekeeping.
2. 00:03–00:25: main story and immediate thesis.
3. 00:25–01:00: evidence, counterargument, and live chat.
4. 01:00–01:10: reset, supporter acknowledgements, one concise CTA.
5. 01:10–02:00: second story or reaction chain.
6. 02:00–02:40: lighter/cultural segment.
7. 02:40–03:00: mailbag, tomorrow's question, and the single owned-audience CTA.

Publishing cadence

• Live: Tuesday, Thursday, Saturday, Sunday; normally 3 hours; public schedule displayed in ET and UTC only.
• Short clips: 16–24 per week across Shorts, TikTok, Reels, and X. Not identical captions everywhere.
• Edited YouTube: two 8–20 minute videos per week.
• X: 3–5 original posts and 5–10 substantive replies per day; no bot-like reply flooding.
• Newsletter: one weekly “Railgun Brief” with the best argument, clips, upcoming schedule, and one product/community CTA.
• Community: one members-only aftershow or town hall per week once 50 paying supporters exist.

9. Monetization ladder

10. Twenty-four-month targets

Each scenario tracks four numbers over time: cumulative followers/subscribers across public platforms, monthly content views across live/VOD/short-form, combined average live concurrent viewers (CCV), and monthly gross attributable revenue.

These are operating scenarios, not promises:

• Pessimistic: the team maintains the full cadence, learns slowly, and gets no major sustained breakout.
• Expected: disciplined production, above-average personalities, regular iteration, several meaningful viral clips, and a functioning sales funnel.
• Optimistic: repeated breakout clips, strong collaborations/tentpoles, excellent retention, sponsor demand, and a product people actually adopt.

The unconditional outcome for a random new streamer is worse than the pessimistic column. These numbers assume 16 quality streams and at least 64 approved clips every month.

11. Dated launch plan and tentpoles

First 45 days

For 2027–July 2028, lock an event into the calendar within seven days of its official date announcement. Recurring priorities: State of the Union, major court decisions, major AI/platform launches, Summer Game Fest/major gaming showcases, major debates, 2027 US state/local election night, year-end reviews, and the 2028 US primary/convention cycle. Never build the schedule around unofficial leaked dates.

Election-night production rule

Use official election authorities, licensed data, and Railgun-created charts. Do not restream cable-news video or copy a network's proprietary map. Each race called on screen must identify the calling source; “too early to call” is better than a viral false call. A 60–90 second safety delay, two moderators, one results researcher, and one backup producer are mandatory.

12. Scoreboard and decision rules

13. Lean budget

One-time

Monthly before payroll

Do not hire a full-time editor before either (a) media gross revenue has exceeded $5,000 for two consecutive months, or (b) one person's editing bottleneck is demonstrably suppressing a winning clip format. Until then, assign a rotating human approval shift and use the tools.

14. The first meeting agenda

The team should leave the first 90-minute meeting with decisions, not vibes:

1. Approve Railgun Live as the flagship name or choose the final alternative.
2. Approve the one-sentence public biography for each persona without identifying facts.
3. Assign the five operating roles and a backup for producer and moderation.
4. Sign the privacy rules and incident chain.
5. Confirm the four weekly live windows in ET/UTC.
6. Choose the first 20 topic candidates and score them.
7. Put every August deadline above on the calendar.
8. Schedule the 29 August pilot and 1 September regular launch.

The show should start slightly rough and unmistakably alive. It should not start insecure. Production polish can improve in public; a leaked identity cannot be put back in the bottle.

15. Source register

• Streams Charts — Q2 2026 livestreaming landscape
• Rumble Studio — Direct RTMP
• Rumble — 2026 Creator Program
• Twitch — Monetization for All and Affiliate criteria
• Twitch — Simulcasting Guidelines
• Kick — Partner/subscription requirements
• YouTube — Expanded YPP eligibility
• YouTube — Partner earnings shares
• YouTube — Monetization and reused-content policies
• YouTube — Fair-use guidance
• X — Creator Revenue Sharing
• X — Creator Subscriptions
• X — Live video and third-party encoders
• TikTok — Creator Rewards
• TikTok — LIVE
• OBS — System requirements
• OpusClip — Pricing and API
• Buffer — Pricing and API
• Google — Advanced Protection
• Optery — Data-removal coverage and pricing
• Brazil TSE — 2026 election calendar
• US FEC — 2026 election activity calendar