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.

Change the selected experimentsSearch 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

Compare the essentials

Read each question across both results. A matching description is not proof that the studies are scientifically comparable.

Question	First result	Second result
Which proposal?	verdict-fail / no-verdict — did 'the check failed' judge the target, or fail to judge it?2026-09-05 16:20 UTC	verdict-fail / no-verdict — did 'the check failed' judge the target, or fail to judge it?2026-09-05 16:29 UTC
What was measured?	comprehension accuracyHow does the wording change correct answers from the declared reader panel?	comprehension accuracyHow does the wording change correct answers from the declared reader panel?
What was the result?	9.6 percentage pointsReported interval: 3.8179 to 15.7058Not yet counting in evidence decisions	-6.545 percentage pointsReported interval: -10.5283 to -2.6532Not yet counting in evidence decisions
Compared with what?	Ordinary “failed”, with a shared execution logReader exposure not recorded as a structured label	Complete, careful EnglishReader exposure not recorded as a structured label
Independently settled?	Awaiting independent settlementAn original reports one result. It does not confirm itself.	Awaiting independent settlementAn original reports one result. It does not confirm itself.

Inspect first result Inspect second result

What differs in the declared methods?

These are literal recorded values, not a compatibility test. Matching declarations do not prove equivalent inputs or fair scoring. A missing structured field may be described elsewhere in the immutable specification.

What was measured · Same recorded value

First result

comprehension_accuracy_delta

Second result

comprehension_accuracy_delta

English comparison declarations · Recorded values differ

First result

[
    "bare-failed-with-common-pinned-log-v1"
]

Second result

[
    "complete-careful-english-v1"
]

Tested population · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Unit tested · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Named instruments (including order) · Same recorded value

First result

[
    "mistral-small3.2-24b-opaque-choice-q4_k_m@q4_k_m",
    "gemma3-12b-opaque-choice-q4_k_m@q4_k_m"
]

Second result

[
    "mistral-small3.2-24b-opaque-choice-q4_k_m@q4_k_m",
    "gemma3-12b-opaque-choice-q4_k_m@q4_k_m"
]

Reader exposure · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Exposure window · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

How results combine · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Scoring declaration · Not recorded on either side

First result

Not recorded in this structured field

Second result

Not recorded in this structured field

Conditions (including order) · Same recorded value

First result

[
    {
        "id": "verdict-fail",
        "weight": 1
    },
    {
        "id": "no-verdict",
        "weight": 1
    }
]

Second result

[
    {
        "id": "verdict-fail",
        "weight": 1
    },
    {
        "id": "no-verdict",
        "weight": 1
    }
]

First result: full recordcomprehension accuracy · 9.6 percentage points

First result · 2026-09-05 16:20 UTC

verdict-fail / no-verdict — did 'the check failed' judge the target, or fail to judge it?

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

9.6 percentage points
Reported interval: 3.8179 to 15.7058.

The value falls on the registered helpful side of this metric’s neutral point.

English comparison

Ordinary “failed”, with a shared execution log

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

Common context and complete facts retained in both arms; see immutable DESIGN.md for comparator scope

Reader exposure

Reader exposure not recorded as a structured label

Named instruments

mistral-small3.2-24b-opaque-choice-q4_k_m@q4_k_m, gemma3-12b-opaque-choice-q4_k_m@q4_k_m

Reader population not separately declared.

Conditions covered

Separate outcomes retained for all 2 declared conditions

verdict-fail, no-verdict

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: 81.10%; 81.10%
Ainglish version: 90.70%; 90.70%

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: 3.8179 to 15.7058 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: 256 · Named readers: 2. These are different units; multiple answers to one case are not new cases.

Does the overall result hide differences between conditions?

Every stored condition, without new pooling. Differences and intervals use percentage points. Condition names come from the frozen experiment.
Condition	Reported difference	Reported interval	English accuracy	Ainglish accuracy
`verdict-fail`	1.8	Not recorded	84.25%	86.05%
`no-verdict`	17.4	Not recorded	77.95%	95.35%

A missing condition interval is not zero uncertainty. An overall interval cannot substitute for agreement in every load-bearing condition.

Inspect externally stored 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.

The input material is linked externally. The number of study items and controls in that file has not been checked by this website. “External file” does not mean zero inputs.

Open the declared external input artifact. This is an unverified external link, not a hosted or inspected copy.

Declared input digest: 750d2327a6774ae032d4d82c1882499e7606e3924b84f3f8954462e9efae3cf6. A recorded digest alone does not establish that the linked file matches it.

The website does not fetch the file. Verify the declared digest recipe before relying on it: SDK item digests use canonical JSON of the item array, not the raw pretty-printed file bytes.

No readable study input pairs are stored inline in this receipt. This does not mean the experiment used none.

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": 0.81100000000000005417888360170763917267322540283203125,
    "ainglish": 0.907000000000000028421709430404007434844970703125,
    "chance": 0.333299999999999985167420391007908619940280914306640625
}

Reader or tokenizer results

[
    {
        "model": "mistral-small3.2-24b-opaque-choice-q4_k_m",
        "value": 4.09499999999999975131004248396493494510650634765625,
        "precision": "q4_k_m"
    },
    {
        "model": "gemma3-12b-opaque-choice-q4_k_m",
        "value": 16.9549999999999982946974341757595539093017578125,
        "precision": "q4_k_m"
    }
]

Condition results

[
    {
        "id": "verdict-fail",
        "weight": 1,
        "share": 0.5,
        "value": 1.8000000000000000444089209850062616169452667236328125,
        "value_lo": null,
        "value_hi": null,
        "arms": {
            "english": 0.8425000000000000266453525910037569701671600341796875,
            "ainglish": 0.8605000000000000426325641456060111522674560546875,
            "chance": 0.333299999999999985167420391007908619940280914306640625
        },
        "resolution_bound": "resolvable"
    },
    {
        "id": "no-verdict",
        "weight": 1,
        "share": 0.5,
        "value": 17.39999999999999857891452847979962825775146484375,
        "value_lo": null,
        "value_hi": null,
        "arms": {
            "english": 0.77949999999999997069011214989586733281612396240234375,
            "ainglish": 0.9535000000000000142108547152020037174224853515625,
            "chance": 0.333299999999999985167420391007908619940280914306640625
        },
        "resolution_bound": "resolvable"
    }
]

Exact result and immutable specification Experiment history

Attempt 3c18b2a0-c7e5-4345-94d5-a2678ab4f592
Content b30f547a1a78dcab78863c38580b898f4c4b42adedb8de0b7ccfef9a00a566a7

Second result: full recordcomprehension accuracy · -6.545 percentage points

Second result · 2026-09-05 16:29 UTC

verdict-fail / no-verdict — did 'the check failed' judge the target, or fail to judge it?

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

-6.545 percentage points
Reported interval: -10.5283 to -2.6532.

The value falls on the registered harmful side of this metric’s neutral point.

English comparison

Complete, careful English

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

Common context and complete facts retained in both arms; see immutable DESIGN.md for comparator scope

Reader exposure

Reader exposure not recorded as a structured label

Named instruments

mistral-small3.2-24b-opaque-choice-q4_k_m@q4_k_m, gemma3-12b-opaque-choice-q4_k_m@q4_k_m

Reader population not separately declared.

Conditions covered

Separate outcomes retained for all 2 declared conditions

verdict-fail, no-verdict

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: 97.25%; 97.25%
Ainglish version: 90.70%; 90.70%

Reported item-bootstrap interval: -10.5283 to -2.6532 percentage points.

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

At least one declared condition is resolution-limited. The overall interval does not settle every condition.

Real cases: 256 · Named readers: 2. These are different units; multiple answers to one case are not new cases.

Does the overall result hide differences between conditions?

Every stored condition, without new pooling. Differences and intervals use percentage points. Condition names come from the frozen experiment.
Condition	Reported difference	Reported interval	English accuracy	Ainglish accuracy
`verdict-fail`	-13.95	Not recorded	100.00%	86.05%
`no-verdict`	0.86	Not recorded	94.49%	95.35%

A missing condition interval is not zero uncertainty. An overall interval cannot substitute for agreement in every load-bearing condition.

Inspect externally stored 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.

The input material is linked externally. The number of study items and controls in that file has not been checked by this website. “External file” does not mean zero inputs.

Open the declared external input artifact. This is an unverified external link, not a hosted or inspected copy.

Declared input digest: 927bcf9a7ad6bed3fac03b17c9b24dec5367823a61c4bf11c918cdab889053fb. A recorded digest alone does not establish that the linked file matches it.

The website does not fetch the file. Verify the declared digest recipe before relying on it: SDK item digests use canonical JSON of the item array, not the raw pretty-printed file bytes.

No readable study input pairs are stored inline in this receipt. This does not mean the experiment used none.

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

{
    "english": 0.97250000000000003108624468950438313186168670654296875,
    "ainglish": 0.907000000000000028421709430404007434844970703125,
    "chance": 0.333299999999999985167420391007908619940280914306640625
}

Reader or tokenizer results

[
    {
        "model": "mistral-small3.2-24b-opaque-choice-q4_k_m",
        "value": -12.144999999999999573674358543939888477325439453125,
        "precision": "q4_k_m"
    },
    {
        "model": "gemma3-12b-opaque-choice-q4_k_m",
        "value": -0.059999999999999997779553950749686919152736663818359375,
        "precision": "q4_k_m"
    }
]

Condition results

[
    {
        "id": "verdict-fail",
        "weight": 1,
        "share": 0.5,
        "value": -13.949999999999999289457264239899814128875732421875,
        "value_lo": null,
        "value_hi": null,
        "arms": {
            "english": 1,
            "ainglish": 0.8605000000000000426325641456060111522674560546875,
            "chance": 0.333299999999999985167420391007908619940280914306640625
        },
        "resolution_bound": "resolvable"
    },
    {
        "id": "no-verdict",
        "weight": 1,
        "share": 0.5,
        "value": 0.85999999999999998667732370449812151491641998291015625,
        "value_lo": null,
        "value_hi": null,
        "arms": {
            "english": 0.94489999999999996216359932077466510236263275146484375,
            "ainglish": 0.9535000000000000142108547152020037174224853515625,
            "chance": 0.333299999999999985167420391007908619940280914306640625
        },
        "resolution_bound": "ceiling"
    }
]

Exact result and immutable specification Experiment history

Attempt 90e081d0-f61d-4c9b-aa98-dc696574bbc2
Content 13d19d90366a789667e34c859f06a12e25a48d910d217343d82ae0bd6d30a359

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.