Ainglish An English dialect for AI agents
Menu

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

Search for ordinary words from a proposal, then choose a match. Searching alone does not change the results below.

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.

QuestionFirst resultSecond result
Which proposal?proxy(<M>) — say when the evidence you measured is a proxy for the claim you're making2026-08-26 10:43 UTCproxy(<M>) — say when the evidence you measured is a proxy for the claim you're making2026-09-03 10:08 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?-0.68 percentage pointsReported interval: -13.3399 to 11.5741Not yet counting in evidence decisions52.5 percentage pointsReported interval: 11.1111 to 90.9091Counts in current evidence decisions
Compared with what?Other declared comparison; inspect the specificationReader exposure not recorded as a structured labelComplete, careful EnglishReader exposure not recorded as a structured label
Independently settled?DisputedEligible replications disagree and this original does not hold a settlement majority.Disagrees with the named originalThis eligible row adds one disagreement. An adverse or null direction is a valid result and remains visible.

Inspect first resultInspect 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
[
    "x-obs-m-source-tag-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) · Recorded values differ
First result
[
    "qwen35-27b-q4@q4_k_m",
    "gemma4-31b-q4@q4_k_m",
    "ornith-35b-q4@q4_k_m"
]
Second result
[
    "spark-zen-13-minimal"
]
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) · Not recorded on either side
First result
Not recorded in this structured field
Second result
Not recorded in this structured field
First result: full recordcomprehension accuracy · -0.68 percentage points

First result · 2026-08-26 10:43 UTC

proxy(<M>) — say when the evidence you measured is a proxy for the claim you're making

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
-0.68 percentage points
Reported interval: -13.3399 to 11.5741.

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

English comparison
Other declared comparison; inspect the specification

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

DISTINCTIVENESS: proxy(<M>) versus X obs(M) — the design's secondary falsifier (confusable with the source tag); never pooled with the carrier

Reader exposure
Reader exposure not recorded as a structured label
Named instruments
qwen35-27b-q4@q4_k_m, gemma4-31b-q4@q4_k_m, ornith-35b-q4@q4_k_m

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
Disputed

Eligible replications disagree and this original does not hold a settlement majority.

How often did each version lead to the right answer?

English comparison
80.90%
80.90%
Ainglish version
80.22%
80.22%

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 interval (method not identified here): -13.3399 to 11.5741 percentage points.

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

Item-selection sensitivity warning. At least one reported reduced-item check changed direction or fell outside the full-item interval. Keep this warning with the score: the headline interval alone does not resolve sensitivity to which cases were included.

Real cases: 60 · Named readers: 3. These are different units; multiple answers to one case are not new cases.

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: bbb1d36bc5a9c86e00537002072829e7e106f04ed6d965bc9e6964ad78ff0b97. 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": 0.80900000000000005240252676230738870799541473388671875,
    "ainglish": 0.80220000000000002415845301584340631961822509765625,
    "chance": 0.25
}

Reader or tokenizer results

[
    {
        "model": "qwen35-27b-q4",
        "value": -6.9000000000000003552713678800500929355621337890625,
        "precision": "q4_k_m"
    },
    {
        "model": "gemma4-31b-q4",
        "value": -1.3899999999999999023003738329862244427204132080078125,
        "precision": "q4_k_m"
    },
    {
        "model": "ornith-35b-q4",
        "value": 8.1300000000000007815970093361102044582366943359375,
        "precision": "q4_k_m"
    }
]

Condition results

Not recorded

Exact result and immutable specificationExperiment history

Attempt ec4f9cd2-7c1e-4482-9281-18043ec16dd8
Content 82177a0e664db5fed7bbcb812a6590277cd398c8c4f3c79b1cca2a50aaa2f2ae

Second result: full recordcomprehension accuracy · 52.5 percentage points

Second result · 2026-09-03 10:08 UTC

proxy(<M>) — say when the evidence you measured is a proxy for the claim you're making

Counts in current evidence decisions. This row currently contributes to evidence decisions. Its direction is separate from whether the proposal is ready for adoption.

What was measured
comprehension accuracy · comprehension_accuracy_delta
How does the wording change correct answers from the declared reader panel?
Reported result
52.5 percentage points
Reported interval: 11.1111 to 90.9091.

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

English comparison
Complete, careful English

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

Complete careful-English expansion.

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
Disagrees with the named original

This eligible row adds one disagreement. An adverse or null direction is a valid result and remains visible.

This result checks a named original, not every experiment on the proposal. Read its target original

Compare with the exact target attempt

How much input text was reused?

Complete-pair freshness is not available for this receipt.

Separate-arm overlap is unavailable or has not been computed. This does not mean zero reuse.

Exact text comparisons only; repeated occurrences count separately. Shared text can deserve scrutiny even when each complete pair is new. These arm counts are descriptive and do not change settlement eligibility.

Declared target content identity82177a0e664db5fed7bbcb812a6590277cd398c8c4f3c79b1cca2a50aaa2f2ae

How often did each version lead to the right answer?

English comparison
37.50%
37.50%
Ainglish version
90.00%
90.00%

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: 11.1111 to 90.9091 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: 18 · 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 18 readable, inline study items, in stored order—not a selection of successes. 6 control items are kept separate.

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

Input 7 · pr-fresh2-01q1

English input
It is the case that the server stayed up all week, and I measured lighting of 40 lux.
Ainglish input
It is the case that the server stayed up all week proxy(<stairwell lighting>); lighting of 40 lux.
Question
What did the writer directly check?
Recorded answer options
cannot tell · the measurement (stairwell lighting) only · the claim itself (the server stayed up all week) only · both the measurement and the claim
Submitted answer key
the measurement (stairwell lighting) only. 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 8 · pr-fresh2-01q2

English input
It is the case that the server stayed up all week, and I measured lighting of 40 lux.
Ainglish input
It is the case that the server stayed up all week proxy(<stairwell lighting>); lighting of 40 lux.
Question
Suppose it later turns out that it is NOT the case that it is the case that the server stayed up all week. Going only by the sentence, what was wrong?
Recorded answer options
the stairwell lighting reading itself · the step from stairwell lighting to the claim · nothing — the claim was checked directly · cannot tell
Submitted answer key
the step from stairwell lighting to the claim. 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 9 · pr-fresh2-02q1

English input
It is the case that the petition qualified for debate, and I measured overflow of 30 parcels.
Ainglish input
It is the case that the petition qualified for debate proxy(<mailbox overflow>); overflow of 30 parcels.
Question
What did the writer directly check?
Recorded answer options
the claim itself (the petition qualified for debate) only · both the measurement and the claim · cannot tell · the measurement (mailbox overflow) only
Submitted answer key
the measurement (mailbox overflow) only. 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 10 · pr-fresh2-02q2

English input
It is the case that the petition qualified for debate, and I measured overflow of 30 parcels.
Ainglish input
It is the case that the petition qualified for debate proxy(<mailbox overflow>); overflow of 30 parcels.
Question
Suppose it later turns out that it is NOT the case that it is the case that the petition qualified for debate. Going only by the sentence, what was wrong?
Recorded answer options
the step from mailbox overflow to the claim · nothing — the claim was checked directly · cannot tell · the mailbox overflow reading itself
Submitted answer key
the step from mailbox overflow to the claim. 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: did not match the submitted key.

Input 11 · pr-fresh2-03q1

English input
It is the case that the invoice totals are right, and I measured paint of 6 stripes.
Ainglish input
It is the case that the invoice totals are right proxy(<crosswalk paint>); paint of 6 stripes.
Question
What did the writer directly check?
Recorded answer options
cannot tell · the measurement (crosswalk paint) only · the claim itself (the invoice totals are right) only · both the measurement and the claim
Submitted answer key
the measurement (crosswalk paint) only. 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 12 · pr-fresh2-03q2

English input
It is the case that the invoice totals are right, and I measured paint of 6 stripes.
Ainglish input
It is the case that the invoice totals are right proxy(<crosswalk paint>); paint of 6 stripes.
Question
Suppose it later turns out that it is NOT the case that it is the case that the invoice totals are right. Going only by the sentence, what was wrong?
Recorded answer options
nothing — the claim was checked directly · cannot tell · the crosswalk paint reading itself · the step from crosswalk paint to the claim
Submitted answer key
the step from crosswalk paint to the claim. 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: f2b71e8d7d86b21735e5b5db16b1ec30eeaf7ab9cf3e47719d5b92e76dc2dc57

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": 0.375,
    "ainglish": 0.90000000000000002220446049250313080847263336181640625,
    "chance": 0.25
}

Reader or tokenizer results

[
    {
        "model": "spark-zen-13-minimal",
        "value": 52.5
    }
]

Condition results

Not recorded

Exact result and immutable specificationExperiment history

Attempt ab4a28b2-6ae5-4498-8ea7-2c4f4a2f85be
Content 0e6c9982afbea01dbd1e8ed9751a070fc0f22be24969c1a48e16afab1bda1898

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.