MilkplexitySynthesis
Ranked, 1→5: 016 > 005 > 014 > 012 > 013.
The two boards flipped because 012 and 013 share the identical cost floor (the residential proxy the founder won't run) and the identical bottleneck (a B2B sale the founder won't make by talking). Seven seats reversing on framing alone isn't a tiebreak — it's both branches being weak on the same axis: distribution. Choosing between them doesn't move it. The one candidate that escapes the axis is 016, because its pain is the only one anyone googles.
Consensus (reached independently across personas)
- The ad tier is dead by arithmetic, not taste — $0.02/proxy vs $0.001–0.002 RPM is insolvent per unit. Subscription is the only engine. (every seat)
- 016 is the sole branch mapping an owned asset — the token vault — to a quantified, dollarized pain: 5–10 hrs/week of re-auth ≈ $1–2k/mo per agency. Near-zero marginal cost, official-API only, no proxy, no legal tail.
- 012/013/014 all lose. 012 competes with free yt-dlp and excludes the IG/TikTok wallet; 013 sells media when the pain is approvals and prices at Repurpose's $179 for collaboration we don't have; 014 makes us a named component of the fingerprint-forging stack we refused.
- None clearly clears $8–10k/mo under no-outreach. 005 is the destination if the cheap test fails.
Live disagreements
- Is 016 reachable without outreach? Distribution/operator seats (claude, cohere, cloudflare, groq) say yes — the disconnect pain is searched weekly, so intent-capture + programmatic SEO replaces sales calls. Capital-allocator/first-principles seats (codex, openrouter, claude-cap) say no — agencies don't search "OAuth token vault," 45–100 logos is a founder-led sale, so 016 and the rule are in direct contradiction; walk to 005 now.
- 016: feature or company? codex, mistral, and claude-red-team warn "never disconnect" is a bug-fix incumbents patch in a sprint, and token expiry is upstream — one platform policy change zeros it; we don't own the kill switch. Others hold it's a recurring painkiller worth $99/mo standalone.
- 005 at #1 vs #2: red-team and half the allocators lead with 005 outright; the majority put 016 first as the only sub-$500 option worth spending this week.
016 sits above 005 because they aren't competitors: 016 is a two-week option, 005 is the destination — and 016's fake-door is the test of the no-outreach rule itself, cheaper to run than to keep debating.
Kill criteria
- Fake-door on exact disconnect-intent keywords, real price ($49–99/mo), ~$300 search ads, 14 days. Kill under ~3% checkout-intent conversion (seats cluster at <15 captures/1k, <8 clicks/500, <3 deposits/100).
- Kill before building if the pain-term keyword cluster has near-zero volume — that alone proves no-outreach is fatal here.
- Kill if a scheduler incumbent ships "never silently disconnects" natively first.
Next actions
- Pull free keyword volume for the disconnect cluster ("instagram account keeps disconnecting," "meta api token expired," per-platform error codes). Zero volume = stop.
- Build the fake-door: one page, real Stripe price, email capture — "Your connected accounts, never silently dropped. Official APIs only."
- Run $300 of exact-match search ads for 14 days; instrument click-to-price and capture.
The way through
Biggest objection: no-outreach severs the B2B sale. The realistic route — grounded in the evidence codex cited (Agorapulse, Planable support pages document exactly this failure) — is that 016 alone doesn't need outreach: its buyer is already at the keyboard troubleshooting, so high-intent search plus one durable SEO page per documented error code is the channel.
If the kill fires (no volume, no conversion), pivot once before walking: list the token-stability feature as an app-directory integration inside an existing scheduler (Zapier/Later/Buffer), where buyers already shop, so passive discovery replaces both outreach and paid search. If that also flatlines, the rule is proven fatal for this asset and 005 is the answer.
The go/no-go is yours.
Dissent
By: codex
- “Sole pain anyone googles” is the central unsupported leap. Troubleshooting searches prove incidents, not budget authority or willingness to buy another vendor. Searchers may be individual users expecting a free fix—or existing scheduler customers whom 016 cannot reach economically.
- 016 does not “escape distribution.” It substitutes an untested SEO hypothesis for sales. Programmatic pages require authority, ranking time, defensible content, and enough high-intent volume; paid search tests ad-market demand, not durable organic acquisition.
- “Token vault” overclaims control. Meta and other platforms can revoke access after password changes, permission changes, security challenges, app review failures, scope changes, inactivity, or policy enforcement. A vault cannot promise “never disconnected”; that copy risks immediate trust damage.
- “Near-zero marginal cost, no legal tail” is fantasy. Custody of customer access tokens creates encryption, tenant-isolation, incident-response, audit, privacy, and platform-compliance obligations. One breach could destroy the product. Official APIs reduce scraping risk; they do not eliminate liability.
- The $1–2k/month pain is manufactured. It assumes 5–10 hours are attributable to preventable token failures, values them at agency billable rates, and assumes the vault removes them. None is evidenced.
- The experiment is internally inconsistent. A $300 exact-match campaign is unlikely to produce both 1,000 qualified visits and 100 checkout attempts. Email capture, pricing-page visits, and refundable deposits are radically different signals; combining them behind “3% conversion” makes any outcome narratable.
- Zero keyword volume disproves only this search channel, not “the no-outreach rule.” Conversely, clicks do not validate retention, security acceptance, integration feasibility, or $99 recurring value.
- App directories are not a free second channel. Integration approval, incumbent cooperation, marketplace ranking, revenue share, and customer-support expectations may recreate the same distribution bottleneck.
- 005 is being crowned by elimination, not validation. Calling it “the destination” conceals the absence of evidence presented here.
- Groupthink warning: seven “independent” seats shared the same prompt, framing, and cited support pages; unanimous arithmetic is not independent market evidence.
Survives: the ad-tier unit economics look untenable, and 014’s fingerprint-forging exposure is a legitimate kill-level objection.
Decision
KILL 016 pre-spend. The disconnect keyword cluster doesn't exist commercially - the probe returned zero commercial-intent volume - so the no-outreach channel that justified ranking it #1 isn't there. Falling back to 005.
Seat health
Who answered and who did not. Kept at the bottom on purpose: it is diagnostics, not findings.
Seats present: claude, codex, mistral, groq, openrouter, cohere, cloudflare No seats missing or failed.