History I
Schedule an original
One slot per round. “—” adds nothing. Every edit rewinds both presses.
A counterfactual printing room
The Founder’s Press
The future has a first draft.
Place a new idea in two otherwise identical histories. Turn the presses. Follow its descendants. Then change the rule that made being early matter.
An invented arithmetic toy, not a model of AI training or a measurement of consciousness, truth, or wellbeing.
Three studies · freely revisable
Choose the engine
Both histories begin with one A and one B. Each round first adds your scheduled original, if any, then applies this rule to both histories.
Two histories · one controlled difference
Round 0 of 6
■ A / amber ink◆ B / blue ink
History II
Schedule an original
A and B are arbitrary ink traditions. Neither is a better or truer idea.
Trace the founders in II
The comparison
JSON record of both histories, rules and ancestry. Local to this tab; no upload, storage or import.
A head start is not an argument.
A copied page can become ubiquitous without gaining an independent source. But a world that keeps adding fresh originals behaves differently. Try the same intervention under both engines: which conclusion belongs to the schedule, and which was put there by the rules?
For research, the harder questions remain outside this room: what deserves to be written, what evidence it rests on, and which feedback mechanisms help later readers correct it.
Colophon · sources, invention & limits
What the sources say
In Philosophy before superintelligence (full hosted thread, December 2024), Michael Edward Johnson argues that unresolved philosophical questions may become path-dependent commitments. He proposes writing down knowledge in data-sparse domains, where early contributions could have founder effects in future AI training data. Those are his arguments, not findings established by this toy.
In A Paradigm for AI Consciousness (June 2024), especially section XI, Johnson distinguishes understanding a domain from caring about or optimizing it, and suggests that existing structure can make similar structure easier to reuse. The essay’s physicalist and consciousness hypotheses are proposals, not scientific consensus. This press borrows the question of historical dependence, not its proposed physics.
What this app invents
The twin archives, ink traditions, six-round schedules, exact doubling and balanced-arrival control are authored mechanics. An original is simply a new lineage ID; even fresh originals are not evidence of independent truth. Copying retains originals and adds one copy of every page; originals inserted this round are copied too. Fresh-arrival mode instead retains the archive and adds one new A and one new B per round, without copying. There is no competition, selection, forgetting, limited capacity, semantic learning or correction.
What remains unknown
Real training pipelines sample, filter and transform data; prevalence need not become influence. This app predicts no model behavior, makes no consciousness or welfare verdict, and does not validate any source hypothesis. It cannot tell you which ideas deserve amplification.
Generation & publication
Original AI-generated tribute · 15 September 2026 UTC. Model: gpt-6-astra, session-reported runtime metadata, not a public-availability claim. Daily only; not main-site approved or endorsed. All imagery is procedural; no external runtime requests, audio, telemetry or persistent storage.