FAQ — Δ ATS

Status: Pre-release v0.7 Document type: Non-normative FAQ. Conversational complement to the normative spec. Normative reference: manifesto.en.md Authoritative language: English. French translation in faq.fr.md.

Each answer below cites the source section in the manifesto or an annex so claims are verifiable. Questions cluster around foundations (§1), format and notation (§2), math and precision (§3), multi-planetary (§4), cultural calendars and adoption (§5), implementation and conformance (§6), site and UI (§7), and versioning and process (§8). Use the navigation to jump.

1. Foundations

Why not just keep the Gregorian calendar?

The Gregorian calendar optimises seasonal alignment for a 16th-century agrarian society. For computing durations, comparing periods, or reasoning about human cycles (focus, rest, project, generation), it adds noise: irregular months (28–31 days), religious weeks (no biological warrant), shifting time zones, summer time. ATS does not replace Gregorian — it offers a cleaner measurement layer that coexists with it. See comparison.en.md §9.1 for the full comparative argument; philosophy.en.md §1 for the rationale.

Why 1969-07-20, not a "neutral" date?

The Apollo 11 lunar landing is the most universally positive, technically verifiable event of the 20th century. It marks the first time a member of our species occupied, in person, the surface of a body other than Earth — the discontinuity ATS uses as its anchor. The four properties that make Apollo 11 uniquely suited as the ATS epoch (verifiability, species-scale witness, no human victim, presence-elsewhere discontinuity) are documented in manifesto.en.md §2.3. Six rejected alternatives (Sputnik, Hiroshima, Wright, Apollo 11 launch, first step, etc.) are documented at manifesto.en.md §2.2.

Why the start of the day (00:00 UTC), not the touchdown instant (20:17:40 UTC)?

To keep the day counter aligned with UTC: with the epoch on the midnight boundary, Bloc 5 = 12:00 UTC exactly (5 × 2 h 24 min), Bloc 0 = midnight, and so on. Earlier drafts ("RC v1.1") anchored on the touchdown instant; that made Bloc 5 fall at 08:17:40 UTC, which surprised newcomers ("why isn't Bloc 5 noon?"). v0.5 fixed that. The touchdown itself is preserved as a remarkable instant inside Δ 0, at T+ Δ 0.0.0.0.84560 (Bloc 8, Centi 4, Deka 5, Kin 6). See manifesto.en.md §2.1.

Why not the first step (1969-07-21T02:56:15Z) as the anchor?

The first step is symbolically heavier than the landing — but it happens 6 h 35 after Eagle has already touched down, and on the following UTC day (Δ 1). The world remembers the landing as "July 20, 1969"; picking the start of that day keeps the cultural memory intact while aligning with UTC. The rejection appears in the manifesto.en.md §2.2 candidate table.

Why not the tropical year (365.2422 days) as a macro unit?

Because it is not an integer multiple of a day. Forcing it into a positional format requires leap days or fractions — exactly what ATS is designed to eliminate. The same logic rejects the synodic month (29.53 days) as a positional unit on Earth, and applies in reverse on the Moon (where the synodic lunar day is the natural unit; see multi-planetary.en.md §3.3).

Why not base 12 or base 20?

Base 12 has divisibility benefits (2, 3, 4, 6); base 20 matches the Mayan Long Count. But the global ecosystem — SI units, currencies, scientific notation, computing, finance — is overwhelmingly base 10. The decimal choice drives adoption friction to near zero. The choice is engineering, not numerology — manifesto.en.md §1.1 is explicit about this.

Why not the Holocene Era (Year 12026 = agriculture-start anchor)?

Holocene Era keeps Gregorian structure and only shifts the year count by +10 000. It is a cosmetic improvement; it does not address the underlying computational problems (irregular months, time zones, leap rules). ATS keeps a clean decimal counter under the same UTC axis. See comparison.en.md §5.1 — Holocene wins on archaeological intuition; ATS wins on computational simplicity. The two could coexist (Holocene year + ATS day count).

Is ATS a religion or a worldview?

No. ATS is a coordinate system on the UTC time axis. It is not a theory about the nature of time, not a replacement for cultural or religious calendars, not a clock synchronisation protocol, and not a metaphysical claim about base 10. The four explicit boundaries are stated in manifesto.en.md §1.1. The proposed cultural rituals in conventions.en.md (Kilo-versary, Hecto-feast, Generation transitions) are opt-in cognitive scaffolds with no metaphysical content; philosophy.en.md §8.1 addresses the "secular religion" charge explicitly.

2. Format & notation

How do I read T+ Δ 20.7.8.0.61137 out loud?

The spec does not standardise oral form (manifesto.en.md §5 describes only written forms). Three observed usages:

• Digit-by-digit: "tee-plus delta two-zero dot seven dot eight dot zero dot six-one-one-three-seven".
• Grouped: "twenty point seven point eight point zero point sixty-one thousand one hundred thirty-seven".
• Short conversational (Δ20.7.8.0-61.1): "delta twenty dot seven dot eight dot zero dash sixty-one dot one".

Pick what fits your context.

What's the difference between the canonical and the short form?

• Canonical (T+ Δ K.H.D.Kin.fffff): full precision, used for logs, storage, and interchange. Always carries the direction marker (T+ or T-). See manifesto.en.md §4.
• Short (ΔK.H.D.Kin-BC.M, v0.7+): conversational form for UI and watches. No space after Δ, separator - between Kin and BC, separator . before Milli. No direction marker (T+ implied). See manifesto.en.md §5.

Where does the Milli digit come from? (v0.7 update)

Before v0.7, the short form was Δ K.H.D.Kin/cc (e.g., Δ 20.7.8.2/50), encoding Bloc and Centi but not Milli — precision ±14 min 24 s. Since v0.7, the format is ΔK.H.D.Kin-BC.M (e.g., Δ20.7.8.2-50.0), adding the Milli digit and bringing precision to ±1 min 26 s. The Milli digit is always emitted, including when its value is 0, so the precision floor is predictable. See manifesto.en.md §5.1.

Why does Δ20.7.8.2-50 — without the Milli digit — get rejected?

The v0.7 short parser is strict: the trailing .M is part of the format. The Milli digit must always be present (even when 0). Refusing a missing Milli prevents ambiguity at the precision boundary — a value with no .M looks like it could be from the pre-v0.7 era. The manifesto.en.md §5.1 ABNF grammar enforces this; the Python and JS reference parsers reject the missing .M form with an explicit error.

"Δ20.7.8-65.1" — which day exactly?

That's still ambiguous on which day, because the Kin digit is missing. v0.1.2 brought Kin back into the short form precisely to fix this: the v0.7 short form is Δ20.7.8.5-65.1 (with Kin = 5). The short form is now precise to the day, with the Milli digit (since v0.7) bringing intra-day precision to ±1 min 26 s. See manifesto.en.md §5.1.

Why no time zones?

Because time zones are the #1 source of date-handling bugs in commercial software [Hertel 2011 in philosophy.en.md references]. ATS = strict UTC by spec (manifesto.en.md §7). To calibrate human activity to the local sun, use the LST (Local Solar Time) overlay as a presentation-layer convenience only — never store an LST value as part of an ATS instant. Software interfaces transporting ATS values must not carry a time-zone field (manifesto.en.md §7).

What about daylight saving time?

Gone, as far as ATS is concerned. DST is a cultural convention applied at the presentation layer; no ATS value moves when DST changes. A society that wants more summer sun shifts its activity hours (in ATS), not its clock. The cities-page dataset (docs/assets/data/cities.json) reflects activity windows; some cities visibly shift their activity windows summer-to-winter. ATS itself is unchanged.

What happens during a leap second?

ATS aligns with POSIX: a day is exactly 86 400 SI seconds. The leap second is absorbed, just like Unix time. Implementations choose one of three documented smear policies — POSIX-naïve, linear smear, or step-jump — and must document the choice (manifesto.en.md §8.2). A future ATS-TAI variant is reserved (manifesto.en.md §8.3) for aerospace and high-precision science where leap-second behaviour matters at sub-second scale.

Why is Kin in the integer part, not part of the fraction?

Because the integer part (Kilo.Hecto.Deka.Kin) answers "which day?" and the fractional part (Bloc.Centi.Milli.Beat.Blink) answers "at what moment in the day?". Two distinct questions, two distinct halves. In canonical form they are separated by .; in short form by - (since v0.7). The Kin is the smallest macro unit (1 day) — it belongs with the date, not with the clock.

What about Kilo > 9?

Kilo has no upper bound. T+ Δ 124.3.5.7.00000 is valid (124 × 1000 + 357 days = ~344 years after epoch). Internally it is an unbounded integer; the reference Python implementation uses Python int (arbitrary precision); the reference JS implementation uses JavaScript number up to 2^53. See manifesto.en.md §4.2.

Can I omit the direction marker T+ in canonical form?

No. manifesto.en.md §3 and §4.1 are strict: canonical form requires either T+ or T-, never omitted. The omission shorthand exists only in the conversational short form (manifesto.en.md §5), where T+ is implied.

3. Math & precision

Won't truncation cause the clock to lag behind reality?

Yes — by design. ATS is a counter of completed units. Until the next fraction has fully elapsed, it is not shown. Consequence: the displayed time is always ≤ the actual instant; drift ≤ 864 ms at 5-digit precision. See manifesto.en.md §6 for the formal invariants. The reference Python implementation passes a property-based test (Hypothesis) on 1000 random instants verifying the drift bound (tests/test_property.py).

Why not banker's rounding (half-even)?

Tried in v0.1.0, reverted in v0.1.1. Half-even can briefly push the display ahead of reality — incompatible with a monotonic counter. ATS prioritises truthful monotonicity over averaging symmetry. See manifesto.en.md §6.4 for the rationale.

How precise is the short form?

The short form ΔK.H.D.Kin-BC.M carries BC (2 digits of Bloc+Centi) plus M (1 digit of Milli) — three digits of intra-day precision, giving ±1 min 26 s. The Beat and Blink digits are not present in the short form. The full canonical form has 5 digits and ±864 ms precision; extended-precision implementations (9 digits) reach ±0.0086 ms. See manifesto.en.md §4.4.

Can the short form round-trip to UTC?

Yes, with the documented precision. Decoding Δ20.7.8.2-50.0 produces an ATS instant whose UTC interpretation is a half-open interval [utc_decoded, utc_decoded + 86.4 s) — exactly one Milli of uncertainty. Decoders must label the result as approximate (manifesto.en.md §10). The 5-digit canonical form round-trips with ≤ 864 ms drift.

4. Multi-planetary

Why Mars Pathfinder as the Mars epoch?

Mars Pathfinder Sagan Memorial Station touchdown (1997-07-04T16:56:55Z) was the first successful modern Mars landing of the contemporary era, with public records precise to the second. The symbolic July 4 date matches the Apollo 11 cultural framing. Six rejected alternatives — Viking 1, Mars 3, Curiosity, Perseverance, Schiaparelli, future human landing — are documented in multi-planetary.en.md §8.1. The first human landing on Mars may become the anchor in a future revision via the RFC process.

Why does the Moon share Earth's epoch?

Doctrinal choice: the Moon is Earth's satellite; the Earth-Moon system is gravitationally and astronomically one system. Using one epoch reflects this. Lunar settlers experience the synodic day (≈ 29.53 Earth days) as the natural intra-day cycle, but the epoch is shared with Earth's so cross-body coordination is simpler. See multi-planetary.en.md §8.2.

Why the synodic lunar day, not the sidereal day?

Three reasons (multi-planetary.en.md §3.3.2):

1. Lunar settlers experience the synodic cycle (sunrise–sunset–sunrise), not the sidereal one.
2. The synodic cycle matches the phase cycle observed from Earth — the cultural reference.
3. The NASA Coordinated Lunar Time framework (2024) uses the synodic cycle.

The sidereal day may be exposed as an additional non-normative unit by implementations that need it.

Can I define a Venus ATS or an Europa ATS?

Yes. The generic framework in multi-planetary.en.md §6 lets any implementation register Δ_X for any body by providing four parameters: epoch_X, day_seconds_X, suffix_X, and (optional) symbol_X. The reference Python implementation (code/ats_multi_planetary.py) provides a Body class with this interface. Vector requirements are in multi-planetary.en.md §6.3.

Do relativistic effects matter for ATS?

For 5-digit precision (864 ms resolution), no. The lunar surface clock drifts ≈ 58.7 µs/day faster than UTC [NIST 2024] — about 14 700× smaller than one Blink. Even over 50 years, cumulative drift is ≈ 1.07 s. For default ATS precision, this is negligible. For sub-millisecond precision use (autonomous lunar rover scheduling, deep-space communication), implementations must maintain their own TAI ↔ surface-local correction tables (multi-planetary.en.md §9.3). A future ATS-TAI variant is reserved.

5. Cultural calendars & adoption

Does ATS replace my religious calendar?

No. ATS is a computational time standard; cultural and religious calendars (Hebrew, Islamic, Hindu, Chinese, Bahá'í, Mayan Long Count, Persian, Ethiopian, etc.) serve cultural and religious functions ATS does not address. The relationship is interoperation, not displacement (manifesto.en.md §13.1). The reference implementation includes 5 bridges (code/bridges/*.py) demonstrating that an ATS instant maps to an instant in any cultural calendar. See comparison.en.md §7 for the full discussion.

Does ATS legislate a 10-day week?

No. The Deka (10 days) is a unit of measurement, not a social mandate (manifesto.en.md §13.3). The 7+3 work-rest rhythm proposed in conventions.en.md §2 is one of several alternatives (6+1+2+1, 8+2, 5+5, 6+4); adopting any of them, or none, is compatible with ATS. Communities committed to the 7-day liturgical week (Sabbath, Sunday, Friday prayer) retain it; conventions.en.md §2.4 documents three reasonable responses for those communities.

What about the Hebrew/Islamic/Hindu/Chinese calendars?

They coexist with ATS via interop bridges. The bridges are in code/bridges/{hebrew,islamic,hindu,chinese,maya}.py. Each maps an ATS instant to its representation in the cultural calendar (and back). Conformance vectors for bridges are in docs/spec/test-vectors-bridges-*.json. See comparison.en.md §7 for the discussion of why cultural calendars are interoperated, not compared.

Won't decimal time fail like the French Republican calendar (1793)?

The French Republican calendar failed for three documented reasons (philosophy.en.md §4.1, comparison.en.md §4.1):

1. Religious-cultural rupture too sharp — the 10-day week meant 1 rest day per 10 instead of 1 per 7.
2. Decimal clocks were expensive to manufacture and the population resisted relearning.
3. Political symbolism — the calendar was associated with the Terror.

ATS draws three explicit lessons:

1. ATS does not legislate rest rhythms.
2. ATS does not abolish the 24-hour civil clock — analog ATS and Gregorian wall clocks coexist.
3. ATS is anchored on a verifiable, non-political event (manifesto.en.md §2.3).

None of the 1793 failure modes apply to ATS. The 1793 failure is a constraint on the design space (which ATS respects), not a proof against decimal time.

Does ATS work on mobile / in the browser / offline?

Yes. The reference site (/{fr,en}/index.html) is a Progressive Web App; the JavaScript reference (docs/assets/js/ats.js) is ~10 KB and runs anywhere a <script> runs. The site is installable on Chrome Android and Safari iOS; the analog dial works offline once cached.

6. Implementation & conformance

How do I verify a third-party library is spec-compliant?

docs/spec/test-vectors.json lists 12 reference instants with canonical and short encodings; test-vectors-arithmetic.json lists 12 algebra cases; test-vectors-multi-planetary-{mars,moon}.json list 10 cases per body. Any implementation must produce bit-identical outputs (manifesto.en.md §16.5). The project's CI runs this contract against Python and JavaScript on every push; the workflow file is .github/workflows/ci.yml.

What does "conformance" mean exactly?

Conformance to ATS v0.7 means:

1. Passing test-vectors.json (Earth) bit-for-bit.
2. Passing test-vectors-arithmetic.json (algebra) bit-for-bit.
3. Implementing the strict canonical and short-form parsers.
4. Documenting the chosen leap-second smear policy (manifesto.en.md §8.2).
5. Documenting the precision class (default 5-digit or extended).

Multi-planetary support adds passing the Mars and Moon vectors. Conformance is binary, not graduated: an implementation is conformant or not. Partial conformance to optional sections must be documented (manifesto.en.md §16.5).

Which leap-second smear policy should I use?

Three policies are normative options (manifesto.en.md §8.2):

• POSIX-naïve smear: matches time_t semantics; RECOMMENDED for general use.
• Linear smear (Google-style): distributes the leap over a 20-hour window; RECOMMENDED for distributed systems where clock-jump tolerance is low.
• Step-jump (TAI-aligned): tracks TAI internally; PERMITTED for aerospace and scientific use.

The reference Python implementation uses POSIX-naïve (inherited from Python's datetime). Document your choice in your release notes.

Is there a JSON schema for spec_version?

Not in v0.7 — the field uses MAJOR.MINOR syntax (no PATCH) defined informally in versioning.en.md §2.1. Implementations are free to validate the field with JSON Schema or any other mechanism. A reference JSON Schema may be added in a future MINOR if implementations request one (the request is itself an RFC; see versioning.en.md §6).

Is there a Rust or Go implementation?

Not yet. A third-party reference implementation in Rust or Go is a v1.0 ship requirement (versioning.en.md §7.2 (3)). It is on the roadmap (ROADMAP.md V1.0-B). Contributors are welcome.

7. Site & UI

What's the analog clock for?

To make ATS readable in a single visual gesture, the way a Gregorian wristwatch is readable. Five hands (Bloc, Centi, Milli, Beat, Blink) sit at five fixed length-and-colour tiers, with the slowest hand shortest and the fastest hand longest (watchmaker convention). The dial also carries an outer ring of city arcs showing 8 capitals' active-day windows (08–22 local). See analog-clock.en.md for the full reference design and philosophy.en.md §2 for the rationale behind the Bloc unit.

How do the city arcs work?

Each city draws a coloured arc covering its local 08:00 → 22:00 active day. The arcs use four different stroke styles per slot (morning/noon/afternoon/evening) to make the time-of-day visually scannable. Click a city trigram to enter focus mode: a "camembert" overlay appears at the centre of the dial, the other cities' arcs dim, and the focused city's trigram enlarges. See analog-clock.en.md §9 and §10.

What's the Cities map page?

A separate reference page at /{fr,en}/cities.html showing a world map of ~40 capitals with an emoji per city evolving with the activity in progress locally (sleeping, working, lunching, etc.) at the current ATS instant. A slider lets you scrub through 24 h UTC. See analog-clock.en.md §12 and conventions.en.md §3.4.

Can I add my own city?

Yes. Open the <details> panel below the clock, enter a code (2–4 letters), name, IANA time zone (autocompleted from Intl.supportedValuesOf('timeZone')), and pick a colour. Up to 6 custom cities are persisted in localStorage["ats-custom-cities"]. Nothing leaves the browser. See analog-clock.en.md §11.

Why does the clock use 10 Hz refresh instead of 60 Hz?

Three reasons (analog-clock.en.md §8.1):

1. The Blink position refreshes every ≈ 864 ms, below 10 Hz visibility; faster ticks would not show new information.
2. 10 Hz is cheap on CPU and battery.
3. Mobile browsers throttle background timers to 1 Hz; 10 Hz active gives a headroom margin without visible jitter when the tab returns to focus.

A 60 Hz implementation driven by requestAnimationFrame is permitted.

8. Versioning & process

What changes between v0.x versions?

Breaking changes are permitted at minor boundaries pre-v1.0; each is recorded in CHANGELOG.md with a migration path. The most recent example: v0.6 → v0.7 changed the short-form syntax from Δ K.H.D.Kin/cc to ΔK.H.D.Kin-BC.M. Post-v1.0, breaking changes are forbidden at minor boundaries; they require a new project (ATS 2) per versioning.en.md §3.8.

Is v1.0 done yet?

No. v1.0 ships when all seven requirements in versioning.en.md §7.2 are met. As of v0.7:

• (1) spec_version field on all vectors — DONE (v0.6).
• (2) Multi-planetary annex normative — DONE (v0.7).
• (3) Third-party implementation (Rust or Go) — TODO.
• (4) Published artefacts (npm publish, twine upload, signed GitHub Release) — TODO.
• (5) RFC archive (docs/spec/rfcs/) — TODO.
• (6) GOVERNANCE.md naming editors of record — TODO.
• (7) Lighthouse CI workflow — DONE (v0.7).

3 of 7 requirements met; 4 remaining.

How do I propose a change to the spec?

Open a public GitHub Issue or PR titled RFC: <topic> containing the required sections (summary, motivation, specification, migration, backward-compat analysis, vector impact). Public comment period: minimum 14 calendar days. The decision is recorded by the editors of record with written justification. Accepted RFCs are merged and version-bumped. See versioning.en.md §6 for the full procedure.

Why isn't ATS an IETF RFC yet?

The editors intend to submit ATS as an Informational RFC to the IETF after v1.0 ships (manifesto.en.md §16.3). Pre-v1.0 the spec is iterating too rapidly for the IETF process (multi-month last-call periods). The lightweight RFC process documented in versioning.en.md §6 is sufficient for the current cohort of contributors. An IANA registry for canonical/short-form labels and multi-planetary body identifiers is planned for the IETF submission.

Is there a governance document?

Not yet. GOVERNANCE.md (naming editors of record and the decision procedure) is a v1.0 ship requirement (versioning.en.md §7.2 (6)). Pre-v1.0, decisions belong to the document editor of record. Post-v1.0, a multi-editor model with rough consensus and backward-compatibility veto applies (versioning.en.md §6.3).

Is there a public roadmap?

Yes, in ROADMAP.md at the project root. It tracks delivered work (v0.6 + v0.7) and remaining v1.0 blockers (V1.0-B Rust/Go, V1.0-C JS converter, V1.0-E i18n, V1.0-F tests, V1.0-G background sync, V1.0-H branding, V1.0-I adoption signal). As of v0.7, 3 of 9 v1.0 blockers are closed.

What this FAQ is not

To pre-empt category errors (mirroring the posture of the other annexes):

1. This is not normative. Conformance (manifesto.en.md §16.5) does not depend on any answer here. The FAQ exists to make the normative content more accessible, not to define new normative content.
2. This is not exhaustive. Questions are added as they emerge from user interaction. Submitting a question via GitHub issue is welcome (manifesto.en.md §16.1).
3. This is not legal or financial advice. ATS is a time representation; how you use it in a regulated context (financial reporting, legal contracts) is your responsibility. The Hanke-Henry calendar may be a better fit for fiscal reporting (comparison.en.md §3.3).

Cross-references at a glance

Every claim in this FAQ either references a spec section or labels itself as the editor's opinion. If you spot an answer that cannot be traced to a normative source, please open an issue.