The Consilience — the graded hypothesis enginestructural comparison instrument

Two theories of consciousness, reduced to what they actually commit to — words stripped, structure kept. The middle shows where they secretly agree. The right panel shows where they genuinely disagree, and which disagreements an experiment could settle today.

How to read this
  1. Each box is a claim, pulled from the theories' own papers — click any box to see the exact source sentence.
  2. The middle column is computed agreement: where the engine found the two theories saying the same thing under different words.
  3. The right panel is computed disagreement, each split tagged with whether an experiment could settle it today.
How grades work

Every match gets one of four confidence levels:

  • FORMAL — identical structure, same shape, same relations. Rare between different theories.
  • PARTIAL — real structural agreement on the same commitment, verified. The most common honest match.
  • GENERATIVE — same kind of claim, different content. Worth knowing about, not a strong match.
  • THEMATIC — surface-level echo only. Graded low, published anyway.

The pair that famously could not be tested. Underneath completely different vocabularies — quantum physics on one side, information theory on the other — the instrument finds they agree on the field's deepest question. Their real disagreements live at different physical scales, which is why no shared experiment exists.

The INTREPID triangle — IIT × Neurorepresentationalism × Active Inference

Three theories currently being tested against each other by a live research consortium. Their hand-built comparison was published this year; their experimental results are not out yet. Everything the engine finds beyond their table is a standing prediction, scored the day they publish.

Loading… (build with p7b_intrepid_diff.py)

Which pairs can actually fight

Every pair of major theories, scored by one question: how many of their disagreements can an experiment settle with today's instruments? Green cells are fundable experiments. Red cells are workshops that would come up empty. Counts are engine-computed; labels are field history; grey cells are pending extraction — never faked.

Loading matrix… (build with p8_matrix.py → results/matrix.json)
decidable forks exist no decidable fork novel candidate ◇ bridge = decidable only under an unproven bridge RAN = a funded collaboration has tested this pair · FAILED = a workshop convened and found no experiment · grey = not yet extracted