10
seed-coded eventsFirst tranche across classical computing, internet collaboration, AI capability, and quantum self-capability.
A progressively published investigation into whether new problem-solving technologies produce delayed, learning-mediated waves of mathematical and scientific discovery.
Research boundary: This page publishes a testable hypothesis, study design, seed-coded events, sources, uncertainty, and revisions. It does not establish a historical law. Site is the public mirror, not proof authority.
First tranche across classical computing, internet collaboration, AI capability, and quantum self-capability.
Primary or authoritative sources currently attached to the seed events.
Confidence reflects source completeness, not support for the hypothesis.
Quantum infrastructure precursor tracked outside the discovery-event ledger.
General-purpose problem-solving technologies may produce temporally clustered advances rather than a smooth increase in discovery. A wave may move through availability, experimentation, self-capability research, native-method formation, rapid exploitation, and either tapering or overlap with a later capability wave.
The first problem class addressed by a technology may be the technology itself: what can be constructed, measured, controlled, verified, and distinguished from existing methods.
The July 28, 2025 launch of the University of Osaka-led fully Japan-made superconducting quantum computer is being tracked as an infrastructure and exposure precursor, not as a discovery event. Its research relevance is the combination of domestically integrated hardware, cloud execution, browser-facing interaction, and the open-source OQTOPUS operational stack.
The case may indicate movement from a device-construction cluster toward a capability-discovery cluster because access, learning, inspection, and experimental barriers are reduced together. No downstream cluster is claimed. The pilot event ledger remains unchanged until independently verifiable discoveries or capability results can be coded.
The compute-versus-architecture binary stops one layer too early. Compute affects the reachable capability space; architecture affects how efficiently that space is searched and used; governance affects whether resulting capability can be authorized, executed, reconstructed, challenged, and recovered reliably.
The proposed relationship is an inverted U rather than a claim that constraint always improves innovation. Low pressure can favor brute-force scaling. Moderate pressure may intensify architectural, algorithmic, data, and infrastructure innovation. Extreme pressure may suppress experimentation and verification.
This is a candidate explanatory mechanism for variation in learning lags, native-method formation, self-capability research, discovery-wave intensity, tapering, and overlap. It remains untested and must distinguish original research from transferred, licensed, open-weight, imitated, or distilled capability.
Initial events and sources are being coded. Aggregate rows may be split as evidence improves.
An independent second coding pass will test whether dates, dependency, exposure, and acceptance categories can be applied consistently.
Timelines, lags, conversion rates, inventory balance, field comparisons, constraint-pressure models, and placebo tests.
Only after protocol freeze, preregistration, negative controls, and held-out validation.
Candidate production is modeled separately from conversion into accepted knowledge through verification and institutional acceptance.
Generation, verification, disclosure, publication, acceptance, and recognition are distinct timestamps.
Inflow separates inherited problems becoming tractable from entirely new problems becoming formulable.
Availability alone is insufficient; field compatibility, adoption, and effective capability determine exposure.
Cloud, browser, documentation, and open-stack changes are tracked as exposure precursors rather than automatically coded as discoveries.
Moderate resource pressure may accelerate efficiency-oriented method formation, while extreme pressure may suppress experimentation.
Learning and method maturity may mediate technology effects, while recognition and funding create feedback.
Capability per unit of compute is distinct from authority, evidence persistence, commit-time admissibility, and recoverable execution.
Placebo dates, weakly compatible fields, shuffled exposure, and simulated null data test whether the method manufactures clusters.
Seed-coded records are revisable. Missing dates remain blank rather than inferred. Full records and source URLs are available in the JSON ledger.
| ID | Event | Wave | Mechanism | Dependency | Orientation | Confidence |
|---|---|---|---|---|---|---|
| COMP-001 | Four-Color Theorem computer-assisted proof | Classical computer | Computer-assisted proof | Necessary | External | Medium |
| COMP-002 | Nonexistence of a finite projective plane of order 10 | Classical computer | Exhaustive search | Necessary | External | Medium |
| COMP-003 | Solution of the Robbins problem by EQP | Classical computer | Automated theorem proving | Necessary | External | High |
| NET-001 | Density Hales-Jewett Polymath proof | Internet collaboration | Open collective proof | Material | External | High |
| NET-002 | Polymath8 bounded-gaps improvements | Internet collaboration | Open optimization and synthesis | Material | External | Medium |
| AI-001 | AlphaTensor matrix-multiplication algorithms | AI capability | RL search plus exact verification | Necessary | External | High |
| AI-002 | AlphaDev faster sorting algorithms | AI capability | RL program search | Necessary | External | High |
| AI-003 | FunSearch cap-set and bin-packing results | AI capability | LLM-guided program search | Necessary | External | High |
| QNT-001 | Application-motivated full-stack quantum benchmarking | Quantum computing | Self-capability benchmarking | Material | Self-capability | High |
| QNT-002 | Learning logical Pauli noise in quantum error correction | Quantum computing | Error-characterization research | Material | Self-capability | High |
The framework is weakened if reproducible clustering disappears after controlling for research growth; peaks precede effective availability; method maturity adds little; exposure does not predict field response; inventory balance does not predict tapering or overlap; access-layer changes do not alter participation or downstream output; moderate constraint does not increase efficiency-oriented method formation; apparent constraint effects disappear after transferred knowledge is controlled; negative controls produce comparable effects; or process tracing shows that supposedly necessary technology or constraint was incidental.