token costHow does the wording change tokenizer units for the declared tokenizer population?
token costHow does the wording change tokenizer units for the declared tokenizer population?
What was the result?
1.875 tokens per declared itemReported interval: 1.875 to 1.875Not yet counting in evidence decisions
2.25 tokens per declared itemReported interval: 2.1875 to 2.25Not yet counting in evidence decisions
Compared with what?
English comparison not recorded as a structured labelCurrent tokenizer cost, not comprehension
English comparison not recorded as a structured labelCurrent tokenizer cost, not comprehension
Independently settled?
Awaiting independent settlementAn original reports one result. It does not confirm itself.
Inside the sampling band of both recountsThe point rule filed a disagreement, but the difference lies within the sampling noise the register derived from both rows’ own pairs. The rows cannot tell each other apart, so this row counts toward neither agreement nor disagreement and spends no settlement voice.
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
token_delta
Second result
token_delta
English comparison declarations · Not recorded on either side
First result
Not recorded in this structured field
Second result
Not recorded in this structured field
Tested population · Same recorded value
First result
Fresh fictional deterministic reports: 8 tells-apart, 4 fits-both, 4 fits-neither; cl100k_base and o200k_base; not natural-use prevalence
Second result
Fresh fictional deterministic reports: 8 tells-apart, 4 fits-both, 4 fits-neither; cl100k_base and o200k_base; not natural-use prevalence
Unit tested · Same recorded value
First result
complete message pair including identical explicit context
Second result
complete message pair including identical explicit context
Named instruments (including order) · Same recorded value
First result
[
"cl100k_base",
"o200k_base"
]
Second result
[
"cl100k_base",
"o200k_base"
]
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 · Same recorded value
First result
{
"reducer": "least_favourable",
"rule": "maximum tokenizer mean over the same 16 equally weighted message pairs"
}
Second result
{
"reducer": "least_favourable",
"rule": "maximum tokenizer mean over the same 16 equally weighted message pairs"
}
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 recordtoken cost · 1.875 tokens per declared item
Not yet counting in evidence decisions. This row remains available for assessment, but does not currently carry a counting evidence result.
What was measured
token cost · token_delta How does the wording change tokenizer units for the declared tokenizer population?
Reported result
1.875 tokens per declared item Reported interval: 1.875 to 1.875.
More tokens on the named current tokenizers; this is the encoded-length difference, not the proposal decision.
English comparison
English comparison not recorded as a structured label
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
Awaiting independent settlement
An original reports one result. It does not confirm itself.
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?
Awaiting independent settlement.
An original reports one result. It does not confirm itself.
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 · lamp
English input
I hold the violet filter reading; the rival is the amber filter reading. A test lamp has been switched on.
The rival predicts amber light; my reading predicts violet light. The lamp emitted violet light.
Ainglish input
I hold the violet filter reading; the rival is the amber filter reading. A test lamp has been switched on.
The lamp emitted violet light tells-apart(the amber filter reading: amber light | violet light).
Input 2 · gate
English input
I hold the spring return reading; the rival is the magnetic latch reading. A model gate was released from the test stand.
The rival predicts a closed gate; my reading predicts an open gate. The gate remained closed.
Ainglish input
I hold the spring return reading; the rival is the magnetic latch reading. A model gate was released from the test stand.
The gate remained closed tells-apart(the magnetic latch reading: a closed gate | an open gate).
Input 3 · sorter
English input
I hold the round tray reading; the rival is the square tray reading. A wooden disk entered the sorter.
The rival predicts delivery to the square tray; my reading predicts delivery to the round tray. The disk reached the round tray.
Ainglish input
I hold the round tray reading; the rival is the square tray reading. A wooden disk entered the sorter.
The disk reached the round tray tells-apart(the square tray reading: delivery to the square tray | delivery to the round tray).
Input 4 · buzzer
English input
I hold the double pulse reading; the rival is the single pulse reading. The buzzer received one trigger.
The rival predicts one beep; my reading predicts two beeps. The buzzer gave one beep.
Ainglish input
I hold the double pulse reading; the rival is the single pulse reading. The buzzer received one trigger.
The buzzer gave one beep tells-apart(the single pulse reading: one beep | two beeps).
Input 5 · loader
English input
I hold the left diversion reading; the rival is the right diversion reading. A green parcel entered the loader.
The rival predicts movement to the right; my reading predicts movement to the left. The parcel moved to the left.
Ainglish input
I hold the left diversion reading; the rival is the right diversion reading. A green parcel entered the loader.
The parcel moved to the left tells-apart(the right diversion reading: movement to the right | movement to the left).
Input 6 · display
English input
I hold the cross display reading; the rival is the dot display reading. A display received the test code.
The rival predicts a dot; my reading predicts a cross. The display showed a cross.
Ainglish input
I hold the cross display reading; the rival is the dot display reading. A display received the test code.
The display showed a cross tells-apart(the dot display reading: a dot | a cross).
Recorded input digest: 088783c89b385860de46e34aacc0afb53612f41b3092818b9001714b87a44f8e
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
Ainglish minus concise complete English; both name held/rival readings and give the same predictions, observation and applicable domain
Tested population
Fresh fictional deterministic reports: 8 tells-apart, 4 fits-both, 4 fits-neither; cl100k_base and o200k_base; not natural-use prevalence
Unit tested
complete message pair including identical explicit context
How results combine
maximum tokenizer mean over the same 16 equally weighted message pairs
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.
Not yet counting in evidence decisions. This row remains available for assessment, but does not currently carry a counting evidence result.
What was measured
token cost · token_delta How does the wording change tokenizer units for the declared tokenizer population?
Reported result
2.25 tokens per declared item Reported interval: 2.1875 to 2.25.
More tokens on the named current tokenizers; this is the encoded-length difference, not the proposal decision.
English comparison
English comparison not recorded as a structured label
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
Inside the sampling band of both recounts
The point rule filed a disagreement, but the difference lies within the sampling noise the register derived from both rows’ own pairs. The rows cannot tell each other apart, so this row counts toward neither agreement nor disagreement and spends no settlement voice.
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?
Inside the sampling band of both recounts.
This replication reports 2.25 tokens; the named original reported 1.875.
The point rule filed a disagreement, but the difference lies within the sampling noise the register derived from both rows’ own pairs. The rows cannot tell each other apart, so this row counts toward neither agreement nor disagreement and spends no settlement voice.
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.
This result checks a named original, not every experiment on the proposal.Read its target original
100.0% of complete English–Ainglish pairs are fresh.
0 of 16 English inputs reuse text from either side of the original.
0 of 16 Ainglish inputs reuse text from either side of the original.
Declared item-bank digests: different. This compares bank identity, not shared sentences; different bank digests can still contain identical pairs.
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.
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 · kettle
English input
I hold the dry-boil cutoff reading; the rival is the fixed timer reading. An empty kettle was switched on and watched for ten seconds.
The rival predicts a full boil; my reading predicts a cutoff. The kettle gave a cutoff after six seconds.
Ainglish input
I hold the dry-boil cutoff reading; the rival is the fixed timer reading. An empty kettle was switched on and watched for ten seconds.
The kettle gave a cutoff after six seconds tells-apart(the fixed timer reading: a full boil | a cutoff).
Input 2 · scale
English input
I hold the tare on power-up reading; the rival is the stored tare reading. The scale was powered up with a 50 gram weight already on the pan.
The rival predicts a reading of 50 grams; my reading predicts a reading of zero grams. The display gave a reading of zero grams.
Ainglish input
I hold the tare on power-up reading; the rival is the stored tare reading. The scale was powered up with a 50 gram weight already on the pan.
The display gave a reading of zero grams tells-apart(the stored tare reading: a reading of 50 grams | a reading of zero grams).
Input 3 · bell
English input
I hold the edge trigger reading; the rival is the level trigger reading. The bell input was held high for three seconds.
The rival predicts continuous ringing; my reading predicts a single ring. The bell gave a single ring.
Ainglish input
I hold the edge trigger reading; the rival is the level trigger reading. The bell input was held high for three seconds.
The bell gave a single ring tells-apart(the level trigger reading: continuous ringing | a single ring).
Input 4 · fuse
English input
I hold the slow-blow reading; the rival is the fast-blow reading. A one-second surge at twice the rated current was applied.
The rival predicts an open fuse; my reading predicts an intact fuse. Inspection found an intact fuse.
Ainglish input
I hold the slow-blow reading; the rival is the fast-blow reading. A one-second surge at twice the rated current was applied.
Inspection found an intact fuse tells-apart(the fast-blow reading: an open fuse | an intact fuse).
Input 5 · basket
English input
I hold the weight limit reading; the rival is the count limit reading. Both limits are 12. The basket holds 9 items weighing 15 kilograms in all.
The rival predicts acceptance; my reading predicts rejection. The basket was rejected. This test applies only to baskets with fewer than 12 items and more than 12 kilograms.
Ainglish input
I hold the weight limit reading; the rival is the count limit reading. Both limits are 12. The basket holds 9 items weighing 15 kilograms in all.
The basket was rejected tells-apart(the count limit reading: acceptance | rejection) over(baskets with fewer than 12 items and more than 12 kilograms).
Input 6 · lock
English input
I hold the keypad only reading; the rival is the keypad with an override key reading. The override key was turned with no code entered.
The rival predicts an unlocked door; my reading predicts a locked door. The check found an unlocked door.
Ainglish input
I hold the keypad only reading; the rival is the keypad with an override key reading. The override key was turned with no code entered.
The check found an unlocked door tells-apart(the keypad with an override key reading: an unlocked door | a locked door).
Recorded input digest: 6a68831ea07a03a21ab8b776d7b544d6e9929ce3a78fdc25786b822645a2fc99
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
Ainglish minus concise complete English; both name held/rival readings and give the same predictions, observation and applicable domain
Tested population
Fresh fictional deterministic reports: 8 tells-apart, 4 fits-both, 4 fits-neither; cl100k_base and o200k_base; not natural-use prevalence
Unit tested
complete message pair including identical explicit context
How results combine
maximum tokenizer mean over the same 16 equally weighted message pairs
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.
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.
