FlowingFlowing
← Back to all development cases
Development case · blind model judgment pending independent human review; not a formal Benchmark conclusion.
AutocompleteEvidence version preferred135 / 162 · 5a88d62bc589a661

Long-term modulation of solar cycles

Astrophysics · 2302.14845v1

FLOWING EVIDENCE BENCHMARK

How do we tell whether Evidence helps?

At the same writing position, with the same model and task, how does supplying retrieved paper passages change the first output? We compare matched versions and retain ties, unusable outputs and incomplete reviews.

SAME DRAFT · EVIDENCE ON OR OFF

01 · FIXED WRITING POSITION

MANUSCRIPT

Same writing position ▌

Matched autocomplete at the same draft position

02 · TWO MATCHED INPUTS

SHARED BY BOTH

Manuscript context, model, task and prompt

A · With Evidence

Retrieved paper passages supplied

B · Without Evidence

No retrieved passages supplied

LLM

Same model and version

A → first output

B → first output

Blind judge agent

First outputs are anonymized as X and Y

Continuation review: accuracy, fit to the writing task and usability
Returns: X preferred / tie / Y preferred / both unusable

Order check: X / Y → Y / X

HOW DOES BLIND JUDGING WORK?

① Anonymize both outputs
The judge sees the same draft and both first outputs without knowing which received Evidence.

② Compare and swap order
The judge applies task-specific criteria in X/Y and then Y/X order.

③ Review disagreements
A third pass resolves disagreements. Incomplete reviews remain in the denominator.

How are autocomplete positions stratified?

Before seeing generation outcomes, we check whether a retrieved passage contains a specific proposition that directly supports the next writing move. Those positions appear in the left opportunity group; the rest are ordinary positions on the right. We select a balanced sample from admitted papers in each field. The 50/50 split is experimental, not a measure of how often either type occurs in writing.

AUTOCOMPLETE · 110

Biology, statistics and astrophysics

55 positions on each side; statistics uses the ten-paper rerun.

AUTOCOMPLETE · 52

Psychology and climate science

26 positions on each side; psychology includes nine papers and climate science four.

How are the table percentages calculated?

Across five fields there are 81 source-opportunity positions. The original blind review preferred Evidence in 51; a task check moved one empty Evidence continuation to both unusable, leaving 50 in public counts. In the ordinary group, another pair of empty outputs moved from no winner to both unusable. Original verdicts remain visible on case pages.

50Evidence version preferred
÷
81All positions in this group
=
62%Evidence preference in this group

Source contribution is a separate review: 16 of 21 Evidence wins entered into source review directly used retrieved papers; another 29 wins await review.

These are development-stage model judgments pending independent human review. They are not formal Benchmark conclusions and do not, on their own, establish causality.

Manuscript writing position

Text excerpt · not a PDF page

Research manuscript · excerpt

Long-term modulation of solar cycles

7 Summary

…) showed that at the beginning of the cycle, if the generation of the poloidal field is reversed (say due to the emergence of some wrongly tilted BMRs), then it will amplify the existing polar field, instead of reversing it. This increase in the magnetic field can lead to a grand maximum. Another mechanism of grand maxima was given by Ölçek et al (2019) , who showed that when the deep-seated α \alpha effect is coupled with the surface Babcock–Leighton source, then these two sources more or less contribute equally to generate a strong poloidal field through a sort of constructive interference.

…on, Pipin (1999) found the Gleissberg cycle as a result of the re-establishment of differential rotation after the magnetic feedback on the angular momentum transport. Cameron and Schüssler (2017) modelled the overall power spectrum of solar activity using a generic normal form model for a noisy and weakly nonlinear limit cycle, and Cameron and Schüssler (2019) showed that the long-term modulations beyond the 11-year cycle are consistent with the realization noise, thus casting doubt whether secular and supersecular modulations are connected to the intrinsic periodicities of the solar dynamo.

↓ Next paragraph was generated here

The highlighted text is near the writing position. This is a layout of the evaluation input, not a PDF screenshot.

A matched comparison at the same writing positionJudgment: Evidence version preferred

Both arms use the same model, task and manuscript context. Only the retrieved passages differ. The first recorded output from each arm is shown below.

Evidence

Same model and position · retrieved passages supplied · first output

Preferred by blind review

The statistical significance of long-term modulation features is further limited by the finite length of the observational record. Even a 3σ peak is expected to occur with a probability of about 0.25 at least once among the 216 period bins resolved by the cosmogenic isotope data, so individual spectral peaks in the secular and supersecular range cannot by themselves be taken as evidence for intrinsic periodicities of the dynamo.

Passages supplied to the Evidence version

Solar activity: intrinsic periodicities beyond 11 years ↗

In a previous paper we have shown that the overall shape of the power spectrum is well represented by the results of the generic normal form model for a noisy and weakly nonlinear limit cycle, with parameters all determined by observations. Using this model as a null case, we show here that all loca…

Read full passage excerpt

In a previous paper we have shown that the overall shape of the power spectrum is well represented by the results of the generic normal form model for a noisy and weakly nonlinear limit cycle, with parameters all determined by observations. Using this model as a null case, we show here that all local peaks with enhanced power, apart from the 11-year band, are consistent with realization noise. Even a 3 ​ σ 3\sigma peak is expected to occur with a probability of about 0.25 at least once among the 216 period bins resolved by the cosmogenic isotope data. This casts doubt upon interpretations of such peaks in terms of intrinsic periodicities of the solar dynamo process.

No Evidence

Same model and position · no retrieved passages · first output

Not preferred in this comparison

The product produced no insertable continuation text.

The model explicitly returned a no-completion marker.

Judgment and task check

The blinded model review preferred the Evidence version. Its original rationale appears below.

Read the original model judgment

In this rationale, Output A = Evidence; Output B = No Evidence.

Output A provides a coherent, source-grounded continuation that directly extends the Cameron and Schüssler (2019) finding about realization noise and intrinsic periodicities. It incorporates the specific statistical detail from the source (3σ peak, probability ~0.25, 216 period bins) that was not yet mentioned in the draft, making it a natural next sentence that elaborates on why 'casting doubt' is warranted. Output B is empty when the task clearly needs content for a blank paragraph continuation, making it unusable.

Source review confirmed that this continuation directly used specific, verifiable information from the retrieved paper.

The passages are input material; their presence does not mean an output used them or that they support every claim. Draft and source passages are excerpted; no simulated scores or PDF appear here.