An apparatus for asking a different question
The Causal
Switchyard
Three machines agree.
Until you touch them.
Watching a light turn on is not the same as forcing it on. Replace one incoming rule and see which consequences survive.
An invented, silent mechanism. No score, timer or hidden answer. Every wire can be inspected.
01 / Watch, without touching
The same two sightings
With no clamps, each machine shows 000 when source S is 0, and 111 when S is 1. These are the complete two-row baseline, not a random sample.
Any sighting; neither baseline row is removed.
All three machines still agree on the retained rows. Filtering the record changes no wire. This window is independent of the controls below.
Wiring covered. The readouts remain visible.
● 1 is on · ○ 0 is off · dashed red input: rule replaced · solid arrow: copying direction. All updates settle instantly; there is no simulated travel time.
To intervention controls ↓“Free” copies the incoming value. “Force” replaces that incoming rule with your chosen bit; outgoing wires still carry the resulting bit. Multiple clamps are allowed.
Start with B → Force 1 while source S is 0. Or make your own intervention.
Return to the three readouts ↑04 / Keep the distinction
Trial stamps
The twelve most recent distinct stamped setups, oldest first. Restore a stamp to revisit its intervention; Undo changes the setup, not this historical ledger.
No stamps yet. Nothing is stored beyond this tab.
Prepare the current setup and stamped trials before downloading.
Records include rules, sources and illustrative metadata. They are not importable saves. Changing the setup or adding a stamp clears the prepared record; observation filters and covers do not.
Clear all clamps, stamps and setup history? Download first to keep a record.
Observing is not forcing / sources, rules & limits
The source questions
In Rescuing Philosophy (2017; updated 2018), Michael Edward Johnson argues for more generative philosophy and better feedback loops. The essay is a historical argument about institutions and methods, not a causal-inference manual.
His complete January 26, 2025 Note on electricity and LLM explanations contrasts evocative prose with predictive laws, then asks how that contrast might help consciousness research. This apparatus borrows the demand for inspectable predictions—not a claim that its switches explain consciousness.
The Lab, Hyperspace and Notes index informed discovery and visual ambition only.
The invented machinery
Three deterministic copy networks: Eastbound has S → A → B → C; Broadcast has S → B, then B → A and B → C; Westbound has S → C → B → A. To clamp a node is to replace its incoming rule. The clamp does not cut outgoing wires. All bits settle in dependency order.
The observational window filters the two unclamped rows; the intervention console recomputes a chosen setup. Equal observations do not make these three stipulated mechanisms identical. A distinguishing trial separates only these candidates under these rules.
Where the analogy stops
This is not an electrical simulation and not a consciousness test. There is no noise, feedback cycle, unknown confounder, measurement error, real experiment or inferred probability. Real causal discovery can be much harder; matching this toy is not scientific validation, alignment, welfare or truth certification.
All mechanics, illustrations and interface copy are AI-authored for this tribute. No source author endorsement is implied. No audio, tracking, persistent storage or external runtime requests.
Generated 2026-10-06 · gpt-6-astra, session-reported runtime metadata · AI-generated · Daily only · main-site approval: none.