A mechanism-agnostic technosignature search — motivated by, but not presupposing, long-lived intelligence.
A pre-registered, open search for anomalous departures — static or time-varying — from the natural behavior of cooling stellar remnants that resist explanation by known astrophysics, around catalogued white dwarfs.
We make no assumptions about how a long-lived civilization arises, what it wants, or what it builds (no Dyson spheres, no energy-maximization, no efficiency/resilience optimization). We look for something that fails to look like known nature, then try to explain it away. The most likely interesting outcome is not life but new natural astrophysics; the most likely outcome overall is a clean null, which yields a quantitative upper limit. See
preregistration.md.
✅ Registered on OSF (2026-06-01) — DOI 10.17605/OSF.IO/6YH7R, frozen at git tag registered-1.0. Analysis is under way on real archival data, all of it post-dating the locked plan.
📊 First results — Channel A (static infrared excess): a clean, explained null plus a quantitative upper limit (f_max). See RESULTS.md.
The repository is public with a force-push-protected main, so the entire drafting + analysis history is an independent, tamper-evident record that all analysis post-dates the registration.
- Pre-registration first. The hypotheses, sample cuts, natural-explanation battery, and scoring rule are fixed before any data is touched, so we cannot reason toward a result post hoc.
- Anomaly-residual, not assume-the-signal. Unlike "listen for a beacon we'd recognize" approaches (whose technology assumptions fail when the technology isn't the one we imagined), we only assume an anomaly leaves some imprint in channels we already survey.
- Null is a success. A clean null produces a publishable upper limit; surviving anomalies are published as unexplained, most likely as new natural phenomena. Success never depends on detecting intelligence.
| Path | Contents |
|---|---|
preregistration.md |
The locked pre-registration document (the heart of the project) |
REGISTRATION.md |
Registration route, timing, and provenance decisions (the record of why) |
AMENDMENTS.md |
Dated log of post-registration changes (pre-data = confirmatory; post-data = exploratory) |
RESULTS.md |
First results — Channel A: the explained null + the f_max upper limit |
SOURCES.md |
Data sources, pinned releases, determinism policy, fetched-artifact log |
pipeline/ |
Analysis code (fetch → SED fitting → battery → upper limit), env lock, and the implementation log — developed post-registration |
docs/ |
The project glossary, original brief, and archived AI-collaboration transcripts |
data/ |
Frozen, checksummed sample manifests + small reference grids. Bulk/derived data is fetched on demand, not tracked. |
figures/ |
Result and diagnostic figures (f_max.png, QQ plots) |
- The pre-registration is timestamped on OSF before any project-specific data analysis.
- All code, queries, the frozen sample manifest, the natural-explanation battery, and the residual catalogue are released here, with history post-dating registration.
- Amendments to the registered plan are public, dated, and rationale-bearing; superseded versions are retained.
- Code (
pipeline/, notebooks, scripts): Apache-2.0 — seeLICENSE. Chosen over MIT for its explicit patent grant and patent-retaliation clause. - Documents and data products (the pre-registration,
docs/, residual catalogues, manifests): CC-BY-4.0.
Copyright © 2026 Tonio Loewald. Drafting and analysis-design assistance from Google Gemini and Anthropic Claude (tools, not authors); see the pre-registration's Acknowledgments.