Skip to content
GREEN FUTURETECH.
Explore

GFT / The evidence method

A claim earns
its place.

Claim. Source. Grade. Ledger. Keep a reviewable chain between the sentence a reader sees and the evidence that supports it.

Inspect the truth pipeline
Claim keeps the sentence, with its unit, period, geography and status.
Slip sentence: baseline and target
Classify a claimProcedural visualisation. Schematic, not a facility.

The claim workbench

Choose the status before choosing the graphic.

Pick a claim class. The slip, the preview, the four gates and the verdict change together. A method demonstration, not an automated verification service.

What kind of statement are you using?
SOURCED STATEMENT / WITH ITS REFERENCE95%EXPORTED RAWLOCAL 5%approximatelyR06GRADE CRECORDED ANNEX FIGURE,FIBRE BASISDERIVED RESULT / FROM STATED INPUTS55050 minus 5 = 45 percentage pointsINPUTS R06 + R07 / NOT A MEASURED GAINILLUSTRATIVE SCENARIO / CHOSEN ASSUMPTIONS1,000 started x 92 percent = 920 goodCHOSEN VALUES, BASIS STATEDILLUSTRATIVEUNSUPPORTED ASSERTION / NO REFERENCEHOLDNo source, no figure.THE RESEARCH TASK STAYS VISIBLE INSTEAD

Gate read-out / selected class

  • CLAIMpass Sentence keeps its unit, period and statuspass The comparison is stated as a comparisonpass Labelled illustrative in the sentencehold Assertion withheld from the argument
  • SOURCEpass Reference, date and qualification travel with itpass Every input carries its own referencelabel Assumption basis stated, no source claimedhold No reference attached
  • GRADEpass The supplied A to D grade stays beside itlimit Limited by the weakest input gradeno grade No grade: it is not a recordhold No grade without a source
  • LEDGERpass Wording and reference are recordedpass Formula and inputs are recordedpass Assumptions are logged with the outputpass The research task is recorded

Sourced statement. Retain the reference, date, unit and qualification. Annex provenance does not establish fresh article verification.

Derived result. Preserve the input references, units, formula and boundary. Recalculate when an input changes. A calculation cannot strengthen a weak source.

Illustrative scenario. Label the assumptions and explain their basis beside the output. Use it to explain a mechanism, not to claim observed performance.

Unsupported assertion. Withhold the factual assertion. Record the research task and the evidence needed before it can enter a published argument.

Name the state the record supports: prepared, reviewed, piloted and measured differ. Label every mark, so a proposed route or scenario never reads as an operating result.

Follow the pipeline

Each gate preserves a different part of the truth.

Claim / boundary / status

Scroll the four steps or select a stage. The slip travels gate to gate and earns one mark at each. No verification score is invented.

Write the sentence a reader can challenge. Define unit, period, geography and boundary, and say in the sentence whether it is an observation, estimate, forecast, target or proposal.

Claim / boundary / status

Name the publisher, document, date and supporting passage, from the original record where possible. A source about another product or time window cannot substitute. A domain-only reference stays labelled as one.

Source / passage / date / limitation

Keep the supplied A to D grade beside the claim with its qualification. Grade describes source type, never a confidence score. A missing reference stays missing even when a plausible number appears elsewhere.

Grade + qualification, never a confidence score

Record the wording, references, inputs and derivation so a reviewer can rebuild the result. When evidence changes, trace the affected claims and outputs instead of editing one figure.

Sentence, evidence, calculation, revision record

As statedWith limits

0255075100PERCENTCOTTON FIBRE BY DESTINATION95%EXPORTED RAWLOCAL 5%BASELINE, ANNEX (R06)5%TARGET BY 203050%45 ppUNITS STARTED, ILLUSTRATIVE92%GOOD UNITS920 of 1,00080 REJECTED
0255075100PERCENTCOTTON FIBRE BY DESTINATION95%EXPORTED RAWLOCAL 5%APPROXIMATE FIGURES · FIBRE BASISANNEX PROVENANCE (R06), GRADE CBASELINE, ANNEX (R06)5%TARGET BY 203050%45 ppAPPROXIMATE ANNEX DATA, GRADE CTARGET, NOT A MEASUREMENT (R07)DERIVED, NOT A FORECASTUNITS STARTED, ILLUSTRATIVE92%GOOD UNITS920 of 1,00080 REJECTEDCHOSEN ASSUMPTIONS, NOT OBSERVEDILLUSTRATIVE
Schematic of the arithmetic. Hover or focus any bar for its read-out.

Sourced, derived, illustrative

A calculation keeps the limits of its inputs.

One chart, three kinds of statement. Choose a tab, slide the assumption, then drag the divider between stated and limited.

Arithmetic, live

50 minus 5 = 45 percentage pointsabout 95 raw, about 5 local1,000 x 92% = 920 good

Derived comparison

Show the arithmetic.

The recorded local-processing target is 50 percent by 2030. Against the approximately 5 percent annex baseline, the gap is approximately 45 percentage points. A derived comparison of a target and a baseline, not a measured improvement. Both inputs are grade C.

Recorded target (R07)

Read-outGap: 45 percentage points, target minus baseline. Hover or focus a bar for its source.

A result must be reproducible

Keep the ledger beside the output.

Output / the sentence as published

The recorded local-processing target is 50 percent by 2030, against an annex baseline of approximately 5 percent, a gap of approximately 45 percentage points.

SourcesGrade C, approximate
Selected ledger entry

Wording

Show ledger entries

01RecordGate: CLAIM

Record

The sentence exactly as used, with its status stated inside it.

Record

Unit, time window and geography, written beside every figure.

02Source idsGate: SOURCE

Record

R06 and R07, each with grade and qualification. A domain reference stays labelled as one; a fresh check also records URL, publisher, dates and the exact passage.

03RecalculateGate: GRADE then LEDGER

Recalculate

The inputs, 50 and 5 in percent, and the formula 50 minus 5, with units checked before computing and totals reconciled on one basis.

Recalculate

The baseline is approximate annex data; the target is a target. A long decimal would not resolve an approximate input.

04ReviseGate: LEDGER

Revise

When R06 or R07 changes, list the comparisons, graphics, tool defaults and conclusions that depend on it, and record why. A corrected numeral without the changed conclusion leaves the decision wrong.

  1. 01SourceR06 and R07The record that changes
  2. 02Comparison50 minus 5Recalculate from the inputs
  3. 03GraphicThe bars and labelsRedraw from the new figure
  4. 04Tool defaultAny value seeded from itReset and note why
  5. 05ConclusionThe published sentenceNarrow or restate it
No revision. Every dependant is unchanged.

The missing gate

Withhold the assertion when the source is absent.

An unresolved claim remains outside the publishable fact chain.

The supplied annex is missing four reference records, so their assertions are not published here. A plausible statement is not a source, and a neighbouring reference cannot inherit an unsupported sentence. Re-source with a named publisher, date and passage, or state an illustrative assumption.

Hold.Hold.

0of 4 simulated sources supplied
Gate held. Nothing is published.

Simulation only. A toggle shows what a supplied source would change at the gate. It verifies nothing; the real task stays in the research register.

Frame 7 of 7 / preserved final

4545

percentage points, derived, approximately (50 minus 5)

Final digits only. Blank until the fact arrives.

Not a true value. A tween shows numbers that never were.

Truth survives the animation

Every frame should preserve the final fact.

A figure may arrive from blank, but it never counts through values that were never true. Scrub the frames and switch the mode.

Mode

Illustrative frames of an easing curve, chosen to show the mechanism. Not measured data. Reduced motion and Calm show the composed state.

Choose what to inspect

Carry the method into the next decision.

The method is useful when it changes what you are willing to claim. Read the source, expose the assumption and keep the boundary with the result.

Select the next evidence task: the gate you choose sets the destination
Claim token

Local processing: about 5 percent of cotton fibre now, 50 percent targeted by 2030

source · gate 02R06R07grade · gate 03C
  1. Wording
  2. Inputs R06, R07
  3. Formula 50 minus 5
  4. Revision record
45percentage points, derived

The punched slip: every gate passed leaves its hole

Gate 01 / CLAIM

Inspect a recorded finding

Open the data hub and read each recorded finding with its reference, grade and qualification beside it.

Open: Inspect a recorded finding

Gate 02 / SOURCE

Follow a publisher thread

Follow the publisher locations recorded against each reference, one thread at a time.

Open: Follow a publisher thread

Gate 03 / GRADE

Read the research register

Browse the paper library, where each paper carries its status and the evidence it depends on.

Open: Read the research register

Gate 04 / LEDGER

Test an explicit assumption

Open the tools and change an assumption against a stated method, with the basis shown beside the result.

Open: Test an explicit assumption

These materials support research and scenario exploration. They are not investment, legal or tax advice. Where an action depends on current policy or specialist interpretation, obtain current primary evidence and appropriate review before relying on it.

Find a thread.

Type to search.