# Frontier Energy Lab: Call for Contributions

Oct 15, 2026 · @Jim Stewart

I run a daily AI research routine that investigates breakthrough energy and propulsion claims and chases real energy records. I'm asking this group for the leads, data, documents and hands-on skills it can't find on its own.

## How the lab works

An AI research agent (Claude) works on this for up to 8 hours every day. It follows the open-source [V3R1T4S](https://github.com/elder-plinius/V3R1T4S) method: only results that someone else's computer can re-check count.

It works across seven families of questions:

- **Claimed devices and effects:** for any published free-energy, over-unity, antigravity, reactionless-drive or anomalous-energy claim with stated numbers, can established physics reproduce everything reported, or provably not? Examples: Sweet's VTA, the Hubbard coil, Moray, Searl, Exodus Propulsion, the EmDrive, Mach-effect thrusters, Podkletnov, LENR.
- **Self-consistency:** do patents and papers (for example Salvatore Pais's Navy patents) support their own conclusions with their own equations?
- **Anomaly data:** re-analysing public raw data from anomaly experiments (thrust traces, calorimetry, resistance curves).
- **Energy records:** beating published records in fusion (stellarator design), power grids and wind-farm layout.
- **Propulsion records:** beating best-known spacecraft trajectory solutions.
- **Anything else checkable:** any open problem in energy, gravitation or propulsion physics with a published definition.

Every result has to survive the same gauntlet before it counts:

1. The pass/fail rule is written down and frozen before any measurement.
2. Statistical claims must beat at least 400 random or shuffled controls.
3. Three independent reviewers, each starting fresh and writing its own code, try to break the result. Any one of them can kill it.
4. A mechanical checker confirms the paperwork is complete and honest.

Most days end with "nothing found", and that is by design. A clean negative result is useful; a false positive is the only real failure.

## First results

There's no confirmed result yet after the first two full runs. Both runs produced specific, re-checkable findings and next steps.

**Exodus Propulsion (Buhler and Aurigema, patent [US11511891B2](https://patents.google.com/patent/US11511891B2/en))**

- The patent's original scans report 237 μN (micronewtons) at +25 kV, measured in air on a gauge with 0.1 mN steps.
- The patent flips the device 180 degrees to rule out attraction to the walls. The lab's 3D models show that grounded surroundings on both sides also produce a force that reverses on a flip: about 1 mN at 100 mm gaps, falling to about 0.14 mN at 200 mm. So the flip test alone can't exclude wall attraction.
- The same models do not explain the patent's ground-plate result: the force rose from 237 to 421 μN with a larger ground plate. That part is still unexplained.
- This is a feasibility result, not a verdict: the patent gives no distances to walls. The lab has designed a bench test that separates the two explanations.

**Pais's inertial-mass-reduction patent ([US10144532B2](https://patents.google.com/patent/US10144532B2/en))**

The lab found what looked like an internal contradiction in the patent's own equations. Two of the three reviewers passed it. The third found that the lab had misquoted the patent, which also states a 10^33 W/m² figure, and the claim was withdrawn. That's the checks doing their job.

## What would help most

The lab can read and compute around the clock, but it can't dig through private archives, measure anything, or know what your experience tells you.

| Contribution | What makes it usable | Example |
| --- | --- | --- |
| A claim to examine | A primary source with numbers: power in and out, force, voltage, mass, duration, and the controls that were run | A patent, paper, conference talk or test report |
| Primary documents | The original, not a retelling, with page or figure numbers | Lab notebooks, 1920s newspaper pages, conference proceedings, government test reports |
| Raw experimental data | Unedited files with units, notes on the setup, and permission to analyse | Thrust-balance logs, calorimetry runs, oscilloscope captures, resistance curves |
| Bench building | High-voltage or precision-measurement training and access to equipment | Running one of the lab's pre-registered tests and sending the raw logs |
| Open benchmarks | A published table of best values and an open way to score a new entry | Fusion, grid, storage, materials or trajectory leaderboards |
| Expert review | Hands-on knowledge of a field, to spot a wrong assumption | Checking a model of a thrust balance or a calorimeter |

## Wanted right now

These are the open leads the lab can't move on alone. Tick one off here if you find it.

- [ ] **Talley's 1991 Phase II report:** the US Air Force follow-up to report AFAL-TR-88-031 on electrostatic thrust in vacuum, which the Exodus patent cites.
- [ ] **Exodus test details:** any written record of a test with on-board batteries and no external leads, or the distances from the device to walls, floor and ground plate.
- [ ] **A primary source for the Exodus press figures** of 5–10 mN in vacuum at 30–40 kV. The lab hasn't found them in the patent or the interviews it checked.
- [ ] **The Hubbard coil:** Alfred Hubbard's 1920 patent number and text, and any first-hand account of the coil's size, output or run time.
- [ ] **Floyd Sweet's VTA:** his original notebooks, or any independent witness measurement.
- [ ] **Public raw data** from any thrust-balance or calorimetry experiment on an anomaly claim, such as the EmDrive, Mach-effect thrusters or LENR.
- [ ] **A qualified builder** for the Exodus wall-gap test: 25–30 kV DC, an analytical balance and two grounded plates. It costs roughly USD 1,000–3,400 if nothing is borrowed, and needs high-voltage DC training.

## How to contribute

Reply in the group thread, message me directly, or comment on this doc.

- **For a claim or document:** send a link or scan, the key numbers with units, and what was claimed versus what was actually measured.
- **For data:** send the raw files unedited, with a short note on the setup and anything odd. Say whether it may be published.
- **Credit:** anything that helps a result is credited to you by name or a pseudonym, your choice. Nothing you send is published without your OK.
- **What happens next:** every contribution is treated as a lead. The lab checks it against the primary sources before using it, and you'll hear what it found either way.

Safety: high voltage, vacuum and strong magnets can kill or injure. Please don't build any of the high-voltage tests without proper training.

## Sources

- [V3R1T4S](https://github.com/elder-plinius/V3R1T4S), the open-source research method the lab follows
- [US11511891B2](https://patents.google.com/patent/US11511891B2/en), the Exodus Propulsion patent (Aurigema and Buhler)
- [US10144532B2](https://patents.google.com/patent/US10144532B2/en), Pais, "Craft using an inertial mass reduction device"
- Talley, AFAL-TR-88-031 (1988), DTIC ADA197537
- The lab's own run reports and dossiers, 14–15 October 2026 (available on request)