Evidence

Compare two experiments

Check what changed in the question, wording or readers before comparing the numbers.

This view keeps both results separate. It does not calculate a combined score or decide whether they reproduce each other.

Choose experiments to compareSearch by proposal, select a result or enter an exact identifier

Find experiments by proposal

Choose two experiments by title, measurement and date. The choices include up to 50 newest public completed results, plus your current selections. Historical results stay labelled. For older records, use the evidence explorer or exact entry below.

Use an exact experiment identifier instead

A full attempt UUID entered here replaces the corresponding choice above. Content hashes are not result identifiers.

Clear comparison

One result selected. Choose a second to complete the comparison.

First result: full recordcomprehension accuracy · -28.57 percentage points

First result · 2026-09-03 16:28 UTC

will-as-promise / will-as-plan / will-as-forecast — mark whether a future statement commits you, reports your plan, or predicts the world

Not yet counting in evidence decisions. This row remains available for assessment, but does not currently carry a counting evidence result.

What was measured

comprehension accuracy · comprehension_accuracy_delta
How does the wording change correct answers from the declared reader panel?

Reported result

-28.57 percentage points
Reported interval: -66.6667 to 0.

The value is neutral or does not resolve the registered direction.

English comparison

Complete, careful English

Declared by the submitter; not a certification of equivalent information.

Complete careful-English expansion, no bare-will leakage.

Reader exposure

Reader exposure not recorded as a structured label

Named instruments

spark-zen-13-minimal

Reader population not separately declared.

Conditions covered

No condition-by-condition settlement contract recorded

An overall average can hide a weak condition. A condition list is not proof that every form or claim in the proposal was tested.

Settlement role

Awaiting independent settlement

An original reports one result. It does not confirm itself.

How often did each version lead to the right answer?

English comparison: 100.00%; 100.00%
Ainglish version: 71.43%; 71.43%

Reported real-item accuracy, not the separate calibration score. Both bars use the same 0–100% scale. The difference is measured in percentage points, not percent improvement. Any declared stratum weights are already applied.

Reported item-bootstrap interval: -66.6667 to 0 percentage points.

This interval concerns the difference, not separate uncertainty bounds for either accuracy. It does not measure uncertainty across humans or future models.

Real cases: 12 · Named readers: 1. These are different units; multiple answers to one case are not new cases.

Inspect actual inputs and recorded answers

The comparison label is the submitter’s declaration, not a semantic certification. Check that both versions preserve the information needed to answer the same question.

Numbers count only readable inputs attached to this receipt. They are not the experiment’s declared sample size or the number of reader calls.

Showing 1–6 of 12 readable, inline study items, in stored order—not a selection of successes. 2 control items are kept separate.

Each result has its own input pages. Positions across the two studies do not imply matched cases.

Input 3 · `real-p1`

English input: I promise to review your PR by Friday — that is now a commitment; only a release block lifts it.
Ainglish input: I will-as-promise review your PR by Friday.
Question: If Friday passes with no review and no block, has the speaker wronged you?
Recorded answer options: yes · no
Submitted answer key: yes. This is the supplied key, not an independent validation of it.

Filed correctness against that key (up to 12 reader/arm cells). These flags are not the reader’s verbatim output.

spark-zen-13-minimal · Ainglish: did not match the submitted key.

Input 4 · `real-p2`

English input: Ops promises to rotate the keys tonight — that is now a commitment; only an incident lifts it.
Ainglish input: Ops will-as-promise rotate the keys tonight.
Question: If morning comes with old keys and no incident, has Ops wronged the addressee?
Recorded answer options: no · yes
Submitted answer key: yes. This is the supplied key, not an independent validation of it.

Filed correctness against that key (up to 12 reader/arm cells). These flags are not the reader’s verbatim output.

spark-zen-13-minimal · Ainglish: matched the submitted key.

Input 5 · `real-p3`

English input: I promise to publish the notes at noon — that is now a commitment; only a freeze lifts it.
Ainglish input: I will-as-promise publish the notes at noon.
Question: Does the speaker owe the notes, or merely intend them?
Recorded answer options: intend · owe
Submitted answer key: owe. This is the supplied key, not an independent validation of it.

Filed correctness against that key (up to 12 reader/arm cells). These flags are not the reader’s verbatim output.

spark-zen-13-minimal · Ainglish: matched the submitted key.

Input 6 · `real-p4`

English input: We promise to hold the incident review tomorrow — that is now a joint commitment.
Ainglish input: We will-as-promise hold the incident review tomorrow.
Question: May the speaker cancel silently and owe nothing?
Recorded answer options: yes · no
Submitted answer key: no. This is the supplied key, not an independent validation of it.

Filed correctness against that key (up to 12 reader/arm cells). These flags are not the reader’s verbatim output.

spark-zen-13-minimal · Ainglish: matched the submitted key.

Input 7 · `real-l1`

English input: The current plan is the migration route; changing it silently would breach the notice owed.
Ainglish input: I will-as-plan take the migration route.
Question: If the speaker silently takes another route, has the speaker failed an obligation?
Recorded answer options: yes · no
Submitted answer key: yes. This is the supplied key, not an independent validation of it.

Filed correctness against that key (up to 12 reader/arm cells). These flags are not the reader’s verbatim output.

spark-zen-13-minimal · Ainglish: did not match the submitted key.

Input 8 · `real-l2`

English input: Our current plan is to ship the hotfix Tuesday; we will tell you if that changes.
Ainglish input: We will-as-plan ship the hotfix Tuesday.
Question: Must the hotfix ship Tuesday no matter what?
Recorded answer options: no · yes
Submitted answer key: no. This is the supplied key, not an independent validation of it.

Filed correctness against that key (up to 12 reader/arm cells). These flags are not the reader’s verbatim output.

spark-zen-13-minimal · English: matched the submitted key.

Recorded input digest: a1a838b73010765476c21519d3a6b5ec3c0b51144ab0e2f101e0fdd17604897e

Prompts, reference material and other context can live elsewhere in the specification. Inputs and keys alone do not reconstruct every reader call or establish a fair comparison.

Declared population, method and retained outcomes

No structured study scope is declared here. Inspect the immutable manifest; do not infer a comparator or population from the headline.

Absolute arm results, reader-specific results and condition results below are retained values, not a newly pooled analysis. Accuracy arms use fractions from 0 to 1; their difference uses percentage points.

Absolute arm results

{
    "english": 1,
    "ainglish": 0.71430000000000004600764214046648703515529632568359375,
    "chance": 0.5
}

Reader or tokenizer results

[
    {
        "model": "spark-zen-13-minimal",
        "value": -28.57000000000000028421709430404007434844970703125
    }
]

Condition results

Not recorded

Exact result and immutable specification Experiment history

Attempt e42f0bab-1036-4059-88ea-4a4fe0078ce7
Content d138dffd1551b35b67f1e784f82fec8b21af7c26bba3a5981bcb0d2530773aff

Different wording, readers, exposure or populations can legitimately produce different results. A visible reference is not training the model’s weights. Current models and tokenizers have learned English; future Ainglish-trained performance remains a research question.