Did they deliver?

Each kept promise counts as one. The total is the number of distinct scored promises. Repeated statements count once. Core promises are marked, but each still counts as one promise.

For example, 2 of 16 kept means 14 have not earned credit. If those 14 are still open, they are pending, not 14 failures. Counts cover the promises we track, not every claim ever made.

• Kept / fulfilled: assessed as meeting its test.
• Open: unresolved.
• Lapsed: an ongoing condition no longer met.
• Retired: recorded as withdrawn or discontinued.
• Unknown: unavailable or unsupported data.

Open and unknown are not proof of failure. Select a count or promise to inspect its record.

All projects opens as an unranked browser. Choose a category to see delivery rankings.

Choose a subject to compare projects making similar kinds of promises. Category rankings currently use kept promises ÷ all tracked promises in that primary category. Equal shares tie; projects with no promises in the category are unranked.

The share measures recorded delivery, not difficulty, impact or token value.

Project pages link to published records marked lapsed or retired, with their assessment and supporting sources. These records are not combined into a warning score or project label.

No recorded problem does not mean every promise was kept. Open and unknown promises remain visible. Read the evidence and decide what it means.

A milestone records an achievement. An ongoing promise requires evidence that its condition still holds. There is no automatic time penalty: a change needs an assessment and supporting evidence.

Published assessments are dated and versioned. Corrections and rule changes belong in new records; they are not new delivery events. Older records remain available for comparison.

• CODE / TEAM: observed development activity and contributor participation.
• USAGE: use for the promised purpose, where measured.
• HYPE: observed attention.
• Market data: what people currently pay.

These do not independently add delivery credit. Code or usage may support a promise's test; popularity and price do not prove it was kept.

A project can miss an original promise and still find useful applications. Those are separate questions. Bitcoin's Genesis asset treatment keeps its promise inventory visible. Its founding role and later uses need their own evidence; they do not settle every original promise's outcome.

Each point is one scored promise, grouped by subject. Green means kept; red means lapsed or retired; grey means unresolved or unknown. Select it for the exact status and evidence.

Categories are curated. Distance and point size do not measure importance, similarity or value. Explore the Atlas ↗

Collection and drafting can be automated. An announcement establishes a claim, not its fulfillment. Missing source details and ambiguous evidence need review; a working link alone does not validate an assessment.

Some current records lack a clearly identified original claim source or evidence date. These gaps are flagged in the Atlas. Published does not mean independently verified, and counts do not prove research is exhaustive.

Missing data stays unavailable. A failed data request is not an empty record.

Current rules: hearts promise-heart rule v3 (adopted 2026-09-25; one promise = one heart; capacity = promise count)

Bubble or Build presents claims, recorded outcomes and supporting context. The count tracks promises kept; it does not measure economic value. We do not combine importance weights, popularity or price into a project warning score.

Assessments remain judgments to inspect, with sources and visible corrections.