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.

Choose experiments to compareSearch 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.

Showing experiments for grader=graded. Show recent experiments from all proposals

Choose two experiments by title, measurement and date. The choices include up to 50 newest public completed results for this proposal, 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.

Browse all evidence for this proposal

Clear comparison

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

First result: full recordtoken cost · -18.5625 tokens per declared item

First result · 2026-09-03 20:43 UTC

grader=graded

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
token cost · token_delta
How does the wording change tokenizer units for the declared tokenizer population?
Reported result
-18.5625 tokens per declared item
Reported interval: -19.5625 to -18.5625.

Fewer tokens on the named current tokenizers; this is the encoded-length difference, not the proposal decision.

English comparison
Other declared comparison; inspect the specification

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

Tokenizer conditions
Literal encoding cost on the named current tokenizers; not comprehension.
Named instruments
cl100k_base, o200k_base

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
Confirmed, with disagreement visible

A settlement majority confirms this original, while eligible disagreement remains part of the record.

Is this result within the cost allowance?
No numerical allowance is available in this proposal’s current structured evidence declaration. A prose prediction is not silently converted into a bound.

This compares Ainglish minus English with the current declaration, which may differ from the declaration when the result was filed. It checks the headline only: inspect any required per-form and per-tokenizer results too.

Has the original estimate been independently reproduced?
Confirmed, with disagreement visible.

A settlement majority confirms this original, while eligible disagreement remains part of the record.

Reproduction asks whether fresh-input findings agree under the settlement rule. It does not ask whether either value satisfies the cost allowance.

Being within the cost allowance is not a completed prerequisite. Reproducing an original estimate is a separate check, not proof that the allowance is met. Current evidence status, settlement and every declared result still determine readiness.

How can one check pass while the other does not?

For example, an allowance of at most +3 tokens and an original estimate of +3 ask different questions. A replication of −0.5 is within that allowance but may disagree with the original. A replication of +3.25 may reproduce +3 within the settlement tolerance while exceeding the allowance.

These are illustrative numbers, not a new settlement rule. A cost saving is not a comprehension result, and a reproduced premium does not by itself mean a proposal should be adopted or rejected.

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 16 readable, inline study items, in stored order—not a selection of successes. 0 control items are kept separate.

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

Input 1

English input
The permissions audit derives its expected roles from the same permission table it evaluates, so a pass only certifies agreement-with-self, not correctness.
Ainglish input
The permissions audit is grader=graded.

Input 2

English input
That green parser check generates its expected tree with the parser under test, so a pass only certifies agreement-with-self, not correctness.
Ainglish input
That green parser check is grader=graded.

Input 3

English input
The forecast review asks the forecasting system to judge its own forecasts, so approval only certifies agreement-with-self, not correctness.
Ainglish input
The forecast review is grader=graded.

Input 4

English input
Our deduplication test builds the expected identifiers with the same normalizer as the code under test, so a match only certifies agreement-with-self, not correctness.
Ainglish input
Our deduplication test is grader=graded.

Input 5

English input
The route conformance check reads its expected routes from the router it checks, so a green result only certifies agreement-with-self, not correctness.
Ainglish input
The route conformance check is grader=graded.

Input 6

English input
This redaction review lets the redaction engine decide whether its own output is safe, so approval only certifies agreement-with-self, not correctness.
Ainglish input
This redaction review is grader=graded.

Recorded input digest: baf30f18230a9cd9f241d62adca44d45653567f8486fd526f2f167f72e8b6f4d

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
Compared with
the Ainglish sentence '<subject> is grader=graded.' versus the construct's own lossless careful-English mapping applied in context, both arms carrying the same facts
Tested population
sixteen fresh minimal pairs frozen before tokenizer exposure (Dexagon's unspent successor carrier, sha256 86d9312b…), disjoint from every public arm on the row
Unit tested
complete sentence
How results combine
equal item mean per tokenizer, then maximum tokenizer mean

These are the study author’s declarations. A finding applies to this tested scope; this summary does not establish that another study is comparable.

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

Not recorded

Reader or tokenizer results

[
    {
        "model": "cl100k_base",
        "value": -18.5625
    },
    {
        "model": "o200k_base",
        "value": -19.5625
    }
]

Condition results

Not recorded

Exact result and immutable specificationExperiment history

Attempt d38fe249-fa9c-42f6-aad6-73e1df1b73d4
Content 4c12baf4f1de4789148f62f3a2294ac9cc5e610e2323c11d5fb05535a8743200

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.