{"data":{"filename":"COMMITMENTS.md","content":"# The Seven Structural Commitments\r\n\r\nThis file is the canonical text of the structural commitments that define The One.\r\nIt is referenced by the Terms of Service, the Privacy Policy, the foundation's\r\nbrand license, the Public Benefit Corporation's public benefit purpose statement,\r\nand the engineering Counter-Position test suite at `__tests__/counter_position/`.\r\n\r\nThis file is included in the quarterly commitment hash published at\r\n`theonecompass.app/transparency/{quarter}` and at `data.theonecompass.app/v1/commitments`.\r\n\r\nModifying this file produces a different hash. The hash change is publicly visible.\r\n\r\n---\r\n\r\n## 1. The Compass Builder is $199, one-time — and the Compass is a living practice\r\n\r\nThe initial Compass Builder is $199, paid once. It is a continuous, self-paced\r\nconversation — as short or as long as the user prefers. The first\r\nversion of the document is produced through that conversation. The user owns the document\r\nin perpetuity, portable in PDF, JSON, and Markdown, off the platform any time.\r\n\r\nThe Compass itself is not a static artifact. Self-knowledge is a continuous\r\npractice. While a user participates in matching, the Compass is the live substrate\r\nthe matching engine reads — refined after each exit interview, after reviewer\r\nconversations, after the user's own reflection. Ongoing refinement is included\r\nin the matching subscription (Commitment #2). The user is not paying for access\r\nto a static pool; the user is paying for the continuous practice that produces\r\ndurable partnership.\r\n\r\nEngineering invariants: no `compass_subscription` table exists. No upsell flow\r\nexists for refinement. Refinement is part of the matching subscription, not a\r\nseparate paywall. Each compass version is preserved (the `derived_from_version`\r\ncolumn and the `compass_version_at_match_time` columns on matches make the\r\nlongitudinal record permanent).\r\n\r\n## 2. Matching is $399 per year, flat\r\n\r\nThere is no premium tier that shortens the cooldown. There is no boost feature.\r\nThere is no priority matching. There is no paid visibility.\r\n\r\nEngineering invariant: no `boost`, `priority_matching`, or `paid_visibility` schemas\r\nexist. The code paths are not built.\r\n\r\n## 3. Singular match\r\n\r\nA user has at most one active match at any time. A user cannot browse other users.\r\nA user cannot see a queue of matches. The interface presents one match, with the\r\nreviewer's note, and asks the user to consider it.\r\n\r\nEngineering invariant: partial unique indexes on `matches.user_a_id` and\r\n`matches.user_b_id` `WHERE outcome IN ('pending','accepted','continuing')`. The\r\ndatabase refuses to create a second active match regardless of which client is\r\ncalling.\r\n\r\n## 4. Silent until worthy\r\n\r\nThe notification service operates on an allowlist. The only events that can trigger\r\na user-facing notification are listed in §6.6 of the Engineering Plan. There are\r\nno marketing emails. There are no \"we miss you\" sequences. There are no weekly\r\ndigests. There are no profile-view alerts. There are no engagement nudges.\r\n\r\nEngineering invariant: the `allowed_events` enum is the entire list. Adding an\r\nentry requires a database migration plus a code change plus a Counter-Position\r\ntest update. It is structurally hard to add, on purpose.\r\n\r\n## 5. In-person within fourteen days\r\n\r\nWhen a match is released, the user has fourteen days to accept and schedule the\r\nin-person meeting. If that window passes without a meeting, the match expires\r\nand both users enter the six-month cooldown.\r\n\r\nEngineering invariant: `matches.expires_at = released_at + interval '14 days'`.\r\nA cron enforces the transition to `outcome = 'expired'`.\r\n\r\n## 6. Six-month cooldown — a refinement period, not a waiting period\r\n\r\nWhen a match ends short of graduation, both users enter a six-month cooldown\r\nbefore another match is considered. The cooldown is for both users, not the\r\nuser who declined.\r\n\r\nThe cooldown protects the architecture from becoming a high-cycle dating\r\nexperience. It also protects the user from pattern-matching exhaustion. And it\r\nis a refinement period: the user who has just exited is the user with the\r\nfreshest data about which compass items predicted accurately and which did not.\r\nSix months allows that data to become compass updates before the next match is\r\nsurfaced. The architecture asks the user to do the work, not just to wait.\r\n\r\nEngineering invariant: `users.next_match_eligible_at` is writable only by the\r\n`complete_exit_interview()` stored procedure. No admin function overrides it.\r\nNo premium tier shortens it.\r\n\r\n## 7. Quarterly public graduation disclosure\r\n\r\nOn the first day of each new quarter, a scheduled job computes the platform's\r\ngraduation rate and publishes it to a public static page and a public versioned\r\nAPI. No human approves the number. The number publishes regardless of whether\r\nit is good.\r\n\r\nEngineering invariant: the quarterly cron writes to `disclosure_snapshots` and\r\npublishes to `theonecompass.app/transparency/{quarter}` and\r\n`data.theonecompass.app/v1/disclosures/{quarter}`. No human approval gate exists\r\nin the code path.\r\n\r\n---\r\n\r\n## The architecture is the company's conscience\r\n\r\nThese commitments are not principles to be honored in spirit. They are invariants\r\nthat the code enforces. Future business pressure will ask the architecture to be\r\nmore permissive. The architecture refuses.\r\n\r\nThe CEO five years from now, under acquisition pressure, will not be able to flip\r\na feature flag to turn engagement monetization on, because the feature flag does\r\nnot exist, because the database tables do not exist, because the architectural\r\ndecision was made by a younger version of the company that still remembered why.\r\n\r\nEvery engineer who joins The One inherits these commitments. Every PR reviewer\r\nasks the question: *does this commit make a Counter-Position commitment easier\r\nor harder to violate?* If easier, the PR does not merge.\r\n\r\nThis is the contract.\r\n","hash":"8554a9c3fea0eb4400ae228b2a051fca7f0fc6ded8db6f60ce10d619fdd2c131","algorithm":"sha256","last_published_hash":"a3a49658dd7e0aee155fdc47e869b926aa4b81588eacf5867c44b263faa9c8da","last_published_quarter":"2026Q2","matches_last_published":false}}