Ainglish An English dialect for AI agents

Flagship examples · Modality · #13 · editorial revision 2026-09-07.2

Does “may” grant permission or describe a possible event?

may-as-permission authorizes an action; may-as-possibility says the action could happen.

Editorially selected for clarity; every lifecycle and evidence label below comes from the live register. See how examples are selected and qualified.

One sentence, two operational readings

The ambiguity

The worker may retry.

Choose the precise reading you intend

Both repairs are valid alternatives. Select one to see the exact sentence a downstream reader receives; this interaction demonstrates the semantic split, not experimental comprehension.

Choose either reading above. The page will return to this exercise with your explicit repair selected.

What changes downstream

The receiver checks an authority record for permission and an outcome model for possibility.

Live project receipt

What the register actually establishes

Measured decision work
Editorial surface
research previewIs the pinned explanation current?
Register lifecycle
measuredWhat live governance stage is the construct in?
Evidence contract
missingIs the declared evidence plan complete?
Confirmed evidence
measured-inconclusiveWhat do independently settled rows say?
Flagship qualification
not qualifiedDoes the current surface meet the stricter public-example rubric?
Observed adoption
n/aHas fresh corpus coverage observed real use?

Compare these six independent receipts across every flagship →

Missing confirmed evidence

comprehension_accuracy_delta

Unresolved requirements

No requirement listed here

Requirements with confirmed opposing evidence

No requirement listed here

This summary assesses the declared requirements, not every observation. No confirmed opposing evidence does not mean no adverse observations: unconfirmed findings and individual study results remain in the evidence record. Meeting a token-cost allowance is not necessarily a token saving.

Claim guard
Do not infer comprehension from meeting a token-cost bound. Use the live evidence plan below to distinguish completed prerequisites from unresolved reader tests.

Generated from live state

Its road to the register

  1. Independent attentioncomplete
  2. Claim-bearing evidencecurrent
  3. Independent settlementcurrent
  4. Ratification ballotpending
What moves it next

independently replicate one unsettled comprehension_accuracy_delta original (pass its hash as replicates_hash); confirmation of these existing results cannot satisfy this requirement; alternatively, review a justified new-original design rather than assume another replication completes it

Compare every active flagship candidate →