EFI/911

EFI·Evidence Fusion Initiative

Fact or Fiction

Fact or Fiction

Ask any question, including odd ones. You get a percentage, a clear call of more or less than 50%, the working, and the sources that produced it. If nothing matches, the score stays at 50% — nothing is invented.

Likelihood

H008 U.S. agencies possessed actionable pre-attack intelligence on these two operatives

More than 50%

72%

More than 50% — supported by the sources

0% — fully contradicted50%100% — fully supported

The score is 72% — more than 50%. Supporting sources currently outweigh contradicting sources after quality and confidence are applied.

How 72% was calculated

Start at 50%. Move by half of the net source score after uncertainty. Net source score is 0.80 (from −1 contradict to +1 support). Uncertainty is 46%. So 50% + 50% × 0.80 × 54% remaining certainty = 72%.

Supporting sources

21

Weighted strength 6.57

Contradicting sources

3

Weighted strength 0.74

No stance yet

20

Pulls the score toward 50% · 5.99

Uncertainty

46%

Share of the file that is unlabelled or missing

Working

start at 50% + half of (net source score × remaining certainty)

0.5 + 0.5 × 0.80 × (1 − 0.46) = 0.716

This percentage is calculated from the sources listed below — it is a score, not a court finding. Official origin does not add extra weight. Independent origin does not subtract weight. Two papers from the same group are one cluster, not two independent confirmations.

Reasoning

  1. Step 1 · Match the question

    Closest stored question: H008 — U.S. agencies possessed actionable pre-attack intelligence on these two operatives. Yes/no labels are taken from sources already linked to that question. Also close: H015, H012, H016.

  2. Step 2 · Find matching sources

    40 source(s) matched the wording. 19 further source(s) already linked to the stored question were included even if they did not contain the search words. 6 matching claim(s) contributed inherited stances.

  3. Step 3 · Assign support or contradiction without inventing

    19 labelled from stored question links, 8 from claims, 0 labelled automatically (capped, marked, not saved as human review), 17 left with no stance.

  4. Step 4 · Weight each source

    Each source contributes strength = (quality / 5) × confidence × how strongly it agrees or disagrees. Support strength 6.57, contradict strength 0.74, no-stance strength 5.99. A zero on chain of custody still lowers the mean and stays visible.

  5. Step 5 · Turn that into a percentage

    Start at 50%. Move up if support outweighs contradiction, down if the reverse. Unlabelled sources shrink the movement, so a thin file stays near 50%. Working: 0.5 + 0.5 × 0.80 × (1 − 0.46) = 0.716.

  6. Step 6 · What this number is not

    This percentage is calculated from the sources listed below — it is a score, not a court finding. Official origin does not add extra weight. Independent origin does not subtract weight. Two papers from the same group are one cluster, not two independent confirmations.

What is still missing

  • 3 source(s) score zero on at least one quality check (shown on each card). Those zeros are kept visible and are not averaged away.
  • 3 source(s) are unverified, inactive, or recorded as not retrieved. A missing file is a gap, not a finding against the question.
  • 20 relevant source(s) have no stance yet. They increase uncertainty and pull the percentage toward 50%; they are not counted as support or contradiction.
  • A second stored question (H015) also matches this wording. The percentage below is for H008 only — open the other question to compare. Do not average the two numbers.

Closest stored question

H008 · U.S. agencies possessed actionable pre-attack intelligence on these two operatives

intelligence_failure · match 49.2

Related questions

Same wording can map to more than one stored question. Each has its own percentage — do not average them.

Matching claims

  • verifiedCLM-007

    U.S. agencies had actionable information on Mihdhar's U.S. visa in January 2000 and failed to watchlist him until August 2001.

  • disputedCLM-020

    Able Danger identified Mohamed Atta's cell before 11 September and was blocked from notifying the FBI.

  • indeterminateCLM-021

    Unusual pre-attack put-option activity reflected trading by people with advance knowledge of the plot.

  • unverifiedCLM-024

    Pakistan's ISI chief authorised a pre-attack transfer to Mohamed Atta.

  • well_supportedCLM-009

    The FBI withheld Bayoumi-related intelligence from the 9/11 Commission.

  • contradictedCLM-005

    Fahad al-Thumairy provided no assistance to the two operatives.

Sources that support this

21

Sources that contradict this

3

Sources with no stance yet

20

Topics

Day-of events, engineering questions, process failures, and foreknowledge leads.

Developed by British researcher Moneypenny

@nic_moneypenny · Subscribe on X from $2 a month for other investigations, and to support this work.

Built with Grok by xAI. EFI / 911 is an independent research project. Neither Moneypenny nor xAI, X, or Grok accepts any liability for errors, omissions, or inaccuracies. Scores are not court or government findings. Privacy and terms.