Frame-Budget-Ansatz (FBA)
Was ist Zeit — operativ gedacht?
Frame-Budget Approach (FBA)
What is time — operationally?

Redefining the Second: Optical Clocks as a Genuine Physics Stress Test (FBA Test Plan)

•

, , ,

•

6 min read

In https://opg.optica.org/optica/fulltext.cfm?id=572911&uri=optica-12-6-843 (Optica, 2025), the largest coordinated campaign to date is reported: 10 optical clocks in 6 countries were compared simultaneously over fiber and satellite links for weeks (including 38 evaluated frequency ratios).

The FBA view turns this into a pass/fail program: not “the best clock,” but a preregistered consistency experiment that licenses link and geopotential proxies as preconditions and only then evaluates bridge residuals across clock types.

Categories

  • Contribution type: Idea
  • Topics: C7 (Constants, scales & renormalization), C2 (Gravity & geometry), C8 (Methodology, data & reproducibility)

Source anchors & subject

Submitted link

https://opg.optica.org/optica/fulltext.cfm?id=572911&uri=optica-12-6-843

Primary sources

Reality check

  • Standard/established: The BIPM timeline describes two CGPM steps: “Presentation/consideration” no earlier than 2026 and ratification no earlier than 2030 (status: March 21, 2025).
  • Standard/established: Roadmap documents and CCTF recommendations explicitly tie the 2030 point to fulfilled criteria (including robust comparisons, reliable links, regular contributions to timescales) and still discuss options such as “single species” versus “ensemble”; roadmap scenarios also mention 2034 as a postponement option.
  • Hypothesis: The second redefinition is de facto a long-term network consistency experiment: the measurable core is not individual best values, but residuals from loop closures and clock-type correlations after clean proxy licensing.

FBA view

  • Handle: Time is operationally fixed first via external calibration (κt, κx) and the front constant c; the “second” is then a calibration product, not an axiom. (Definition I.3.3.1)
  • Principle: The front bound is a hard consistency core: link and transfer artifacts are not secondary—they can leak directly into t as a false front/delay calibration. (Lemma I.1.4.1)
  • Proxy: Geopotential/redshift is treated as a licensing proxy block: only once the stationary redshift proxy is consistent do bridge judgements about residuals become meaningful. (Definition X.4.2.1)
  • Proxy: Gravity as a clock-rate proxy α(x): clock comparisons first measure α-ratios; from that follows the operational geometry language (and thus what “GR correction” means in the test plan). (Formula box VI.3.1.1)
  • Handle: H-gate as an exclusion of representation tricks: clock-type invariance is a testable isometry condition, not a silent metrology postulate. (Definition X.4.1.1)
  • Pass/Fail: Preregistered tests as a triple (O, J, δ*) plus a decision tree separate proxy fail (regime not licensed) from bridge fail (domain refuted). (Definition X.10.1.1)

New insights from FBA

  • FROM→TO: “The second as a definition text” → “the second as a network output (OFS ⊕ links ⊕ operation).” Implicit assumption: link models are not merely engineering, but part of the operational time assignment. (Lemma I.1.4.1)
  • FROM→TO: “Clock type doesn’t matter (after corrections)” → “clock type doesn’t matter only if H-gate checks pass.” Implicit assumption: systematics cannot be hidden as “correction terms” via representation choices. (Formula box X.4.3.1)
  • FROM→TO: “Apply GR correction afterwards” → “proxy licensing before bridge judgement.” Implicit assumption: α(x) / the redshift proxy is independently verifiable before residuals are read as physics signals. (Definition VI.5.2.1)
  • FROM→TO: “Comparison campaign as result” → “comparison campaign as a preregistered residual workflow.” Implicit assumption: δ* thresholds, precondition indicators, and fail classes are fixed in advance, not optimized post hoc. (Definition X.10.1.1)

Alternative readings & conclusions

  • Hypothesis: If—after passing the Tolman/α proxy and with stable link metadata—residuals remain correlated with clock type (not correlated with site or link), then the “second” in this regime is not fully H-gate invariant.
  • open/unclear: Clock-type correlation can also arise from a shared, hidden systematics family (analysis windows, dead-time handling, environment metadata, link-model class); without openly logged parameter registers, the separation “physics vs pipeline” remains blurred.

Alternative readings & conclusions

  • open/unclear: The choice “single species” versus “ensemble” can be read operationally as a design trade-off: which residuals count as acceptable (and which as fail) before one commits to a global definition?

Tests/Experiments (Pass/Fail) with an FBA touch


$$
\mathcal{J}_{\rm loop}=\max_{\text{Loops }i\to j\to k\to i}\left|\ln r_{ij}+\ln r_{jk}+\ln r_{ki}\right|
$$

  • Null test (FBA): J_loop | network of at least 3 clocks (multiple types) with closed link loops (fiber/satellite) | J_loop near zero after systematics, link, and proxy licensing | stable offset or drift beyond the uncertainty budget
  • Pass/Fail (FBA): proxy licensing block | stationary clocks at different geopotentials plus independent potential models and link-delay metadata | Tolman/α proxy consistent before bridge judgement | proxy fail or only “salvageable” via ad-hoc model switching
  • Residual (Hypothesis): type residual R_type | same sites, multiple clock types, rotation of link routes | residuals de-correlate from clock type after corrections | residuals cluster by clock type despite site/link variation
  • Pass/Fail (open/unclear): refinement check R_ref | same raw data, alternative analysis classes (windows/dead-time/link model) | decision invariant under refinement | sign flip or strong class sensitivity (requires a preregistered error band)

Added value of the FBA view

Added value: 8/10 – The second redefinition is operationalized as a proxy-licensed, preregistered residual program that cleanly separates link, gravity, and clock-type error classes—making genuine bridge fails identifiable.

Reference list (URL-only)

Leave a Reply

Your email address will not be published. Required fields are marked *