🔬⚡ Calling all energy researchers: help my daily AI research lab

[JimBoot]

I've set up an AI agent that spends up to 8 hours a day on breakthrough energy and propulsion. That covers free-energy and over-unity claims (Sweet's VTA, the Hubbard coil and many more), antigravity and reactionless drives (Exodus, the EmDrive, Pais's patents), anomalous energy (LENR), and real efficiency records in fusion, power grids and spaceflight.

It's strict. Every result has to pass a rule set before the measurement, beat hundreds of random controls, and survive three independent reviewers. Most days it finds nothing, on purpose. So far it has found that the Exodus patent's test for wall attraction doesn't rule it out, and one of its own findings was withdrawn when a reviewer caught a mistake.

What would help most:
📄 Claims or primary documents with real numbers (patents, notebooks, test reports, old newspaper pages)
📊 Raw experimental data (thrust-balance logs, calorimetry, scope captures)
🔧 People with high-voltage or precision-measurement training who could run a ready-designed bench test
🏆 Open energy or propulsion benchmarks with published best scores
🧠 Expert eyes to check its models

Wanted right now: Talley's 1991 US Air Force Phase II report on electrostatic thrust, any record of an Exodus test running on internal batteries, Hubbard's 1920 patent number, and Floyd Sweet's original notebooks.

Full details and the wish list attached - it's an md file but I had to change the file extension to work here.

Reply here or message me. Anything that helps gets credited to you, by name or a pseudonym, and nothing you send is published without your OK. 🙏

[PhysicsProf]

This is awesome, Jim!   I'm particularly interested right now in:
1- LENR which uses light hydrogen (H2 or H2O etc) and various metals (such as Ni, Ti, etc.)
2- spinning disk generators such as Faraday's generator (paradoxical?)

Wish I could help right now, but caring for my wife in hospice and she deserves and needs my attention atm...

[JimBoot]

Hey Mate thanks for the response. You've got your hands full so this is plenty . I've added both of your suggestions as priorities: light-hydrogen LENR with Ni, Ti and similar metals, and homopolar / Faraday-disk generators. Whenever you feel like looking, I'll happily share what the lab finds, with no expectations at all.

[Magluvin]

hey jim
so which ai's do you believe come up with the best solutions and info on these ou subjects?  does it seem like some or all have built in restrictions on these subjects?

thanks

mags

[JimBoot]

Anthropic are the worst in chat. However I;m using this method. https://github.com/elder-plinius/V3R1T4S - Already about 10 maths problems have been solved on opus 5.5 using this method. If I get a refusal it should be obvious. But I also have qwen3.8 27b jailbroken running locally (no gaurdrails) this model is as good as SOTA ones from just a few months ago. We can work around the censorship. The main thing this author does is jailbreak models

[JimBoot]

1st confirmed output that passed.

Confirmed result: three skeptics agree!!!
In US10144532b2, equation 2 (stated by the source to follow from its non-relativistic equation 1) yields the illustrated 10³³ W/mat the quoted surface charge density of order 1 C/m² with f_g = 1 only if its speed-dimensioned bracket (r_v v²) t_op equals about 8.85*10²¹ m/s (about 2.95*10¹³ c) for any r_v and t_op, whereas keeping that bracket at or below c caps the flux at about 3.4*10¹⁹ W/m², and the quoted "operational time on the order of the inverse of the vibrational amplitude" is dimensionally 1/m so the source's 10³³ figure depends on the length unit chosen.

[Verpies]

...surface charge density of order 1 C/m²

Jeez!  That charge density would cause effects similar to this:
https://youtu.be/jNuify4u-28

[JimBoot]

cool vid Verpies.

🏆🎉⚡ CONFIRMED RESULT: THREE SKEPTICS AGREE!!! ⚡🎉🏆 EVALUATED WITH THE OFFICIAL CONSTELLARATION 0.3.0 GEOMETRICALPROBLEM EVALUATOR, THE BOUNDARY IN OBJECT/BOUNDARY.JSON IS FEASIBLE UNDER THAT EVALUATOR'S 1E-2 RELATIVE-TOLERANCE RULE (FEASIBILITY 0.0098997) AND HAS MAX ELONGATION 1.1743662351223367, STRICTLY LOWER THAN THE BEST FEASIBLE ENTRY (1.1756487689141413) IN THE PROXIMA-FUSION/CONSTELLARATION-BENCH-RESULTS GEOMETRICAL TABLE AS FETCHED ON 2026-10-17. 🥳🎊🚀🌍

Headline: a new boundary for ConStellaration's geometrical problem has max elongation 1.1743662351, below the best feasible public entry (1.1756487689) under the official evaluator. The gain mostly comes from using more of the evaluator's 1% feasibility tolerance (0.0099 against the leader's 0.0090). Second target: nothing found on Kasagi et al. 2025, the nickel–hydrogen excess-heat paper.

ConStellaration: I reproduced the leader's score locally to 1.6e-14, then refined that public boundary step by step with the official evaluator scoring every candidate. The gain is 0.11%. About a fifth of it holds even at the leader's own feasibility level, so it isn't only spent tolerance. The design stays feasible only on the evaluator's fixed 201-point grid; at 401 points both it and the leader would fail.

[JimBoot]

My AI research lab found a better way to arrange 64 wind turbines in a standard computer test. The same turbines, in the same area, produced about 0.85% more electricity in the simulation than the best published layout we found. We checked it using the test's original software, published the coordinates, and asked one of the researchers who created the test to verify it.
https://zenodo.org/records/23448033

[Peterae]

Jim is this what you are after

If looking for old documents, its worth looking in wayback machine to see if a copy was recorded 

[JimBoot]

Jim is this what you are after

If looking for old documents, its worth looking in wayback machine to see if a copy was recorded 

Thanks Peter added to the next run.

[JimBoot]

JimBoot,

do you know Mike Waters Windtourbine ?


I have not seen however any product since then ( last 7 years)

Mike(Ger)

Hi Mike not familiar with it. I'll see if I can add some direction. It came up with the wind turbine idea all by itself. I just check in to see what it finds. Initially I gave it direction on the PAis patents and a few other tidbits. I'll thow this into the mix.

[JimBoot]

ESA turned a real Saturn mission study into a set of puzzles. Plan a route from Earth to Saturn, swinging past other planets on the way to pick up speed. You have to arrive within 10 years. Your score is how much spacecraft arrives at the end, because every kilogram not burned as fuel is a kilogram of useful payload. Researchers have competed on this table since 2008.

The best answers on four of those routes came from a University of Florence team in 2017. Our agents started from those answers and fine-tuned them. The new routes deliver 0.5% to 1.9% more mass to Saturn. One clear difference: the 2017 routes arrived 2.5 to 6 weeks before the 10-year deadline, and the new routes use almost all of that spare time. It's currently with the ESA for review.

[Kator01]

JimBoot,

I deleted in my post #10 the first vid ( was not related to his tourbine) and changed the time in the second vid to start from because only at this time-stamp his tourbine-design is presented.

However - after changing it to the exact time-stamp- the link is broken. Now I tried it a second time here
however there now is a serious problem. By using youtube-Input Icon the link is broken and if I use the link-Icon the youtube link filled into my last comment "sorry for..." 
Also I wanted to insert an image which is in a directory of my computer..failed because ther's only the possibiity for a picture with a https:// - adress.

sorry for the confusion

Now I put the link here directly

https://www.youtube.com/watch?v=SjDAJP9bruo&t=2756s

it works now ( tested )

Additional Info:

https://emediapress.com/tag/mike-waters/

https://www.youtube.com/watch?v=l5Tw-chMqrA

Critics:

"Mike Waters's wind turbine design challenges conventional aerodynamics by utilizing drag and diverting airflow to maximize torque at the blade perimeter rather than relying solely on lift. The design employs a squirrel cage arrangement set perpendicular to the wind, forcing all incoming fluid to flow around a back plate, which accelerates the air before it strikes the blades located at the maximum point of leverage.

Waters argues that this approach allows the turbine to capture 100% of the wind hitting its area, unlike traditional propeller-style turbines where much of the air passes through unaffected.  By accelerating the flow prior to extraction, the design aims to produce significant torque at wind speeds under 1 mph, claiming this method bypasses the limitations of the Betz limit by increasing wind velocity before energy extraction occurs.

However, mainstream physics experts criticize these claims, noting that drag opposes rotation and that the design"'s high solidity may cause airflow to bypass the turbine entirely or become choked at higher speeds. Critics also point out that the design's high inertia and large frontal area present significant structural challenges in high winds and crosswinds, potentially making it less durable and efficient than well-pitched conventional horizontal axis turbines."

Discussion about the "Betz-Law"

https://www.physicsforums.com/threads/performance-of-waters-wind-turbine-in-open-spaces.1065924/

Moderator : Please move this to the apropriate discussion

Mike (ger)

[JimBoot]

ANother Salvatore Pais patent for a device that can't work according to my team. As the patent describes it, the device can't produce the field it claims, and the rest of the design depends on that field

🏆 CONFIRMED (3 skeptics): Pais fusion patent's 1e6 T spinning-charge field needs ≥3.64e7 C/m² / ≥2.06e18 V/m surface field. Packaged in outbox.

🏆🎉⚡ CONFIRMED RESULT: THREE SKEPTICS AGREE!!! ⚡🎉🏆 FOR US20190295733A1'S STATED B_MAX OF ORDER 1E6 T, TAKEN AS THE FIELD SOURCED BY ONE SPINNING ROTOR ITSELF AT A PEAK SPIN OF 1E4 RAD/S AND ROTOR RADIUS 0.05 M, JEFIMENKO'S EQUATIONS REQUIRE A PEAK SURFACE CHARGE DENSITY OF AT LEAST 3.64E7 C/M^2, AND HENCE A PEAK SURFACE ELECTRIC FIELD OF AT LEAST 2.06E18 V/M, FOR ANY AXISYMMETRIC, TIME-CONSTANT CHARGE ON A CONVEX ROTOR SURFACE OF AREA AT MOST 1 M^2, ANY FIELD POINT AT LEAST 1 NM FROM THAT SURFACE, AND ANY RIM TANGENTIAL ACCELERATION AT MOST 1E12 M/S^2. 🥳🎊🚀🌍

Headline: Pais's "Plasma Compression Fusion Device" application claims a spinning charged tungsten cone gives about 10⁶ T. Classical electrodynamics says that needs at least 3.64e7 C/m² of surface charge, which means a surface electric field at least 3.6e7 times the field that strips atoms off tungsten.

What I tried and learned:

The bound. I bounded, from above, the magnetic field any spinning charged convex rotor can make, then inverted it at 10⁶ T. A second method gives the same answer: exact fields of charged discs and cones need 16–25× more charge than the bound.
Rescues and side checks, all dead. None of these got the needed field under tungsten's limit: sub-attosecond spin-up, a rim moving at light speed, gauss units, atomic-scale charge patterns, or several fusors. An apparent arithmetic slip in Eq. 11 turned out to depend on how the page prints "104 rad/s", so I dropped it. The deuterium–xenon sentence breaks charge conservation as written, but one reading releases energy, so it is too loosely stated to count against the patent.
Checks. A fresh-context self-break returned SURVIVES. The REPRODUCE, NOVELTY and CRITERION skeptics all passed. The package check and the end-of-run check both print CHECK_RUN: PASS.

[JimBoot]

🏆🎉⚡ CONFIRMED RESULT: THREE SKEPTICS AGREE!!! ⚡🎉🏆 IN I. XYDOUS'S PREPRINT "A CLASSICAL VACUUM SPIN MODEL FOR PROPELLANTLESS PROPULSION" (PREPRINTS.RU V2, 2026-07-03), THE SEC. V PFT PARALLEL-PLATE WORKED EXAMPLE STATES C = 15 PF, U = 6.5 mJ AND A THRUST-TO-POWER RATIO OF 0.21 N/W, BUT AT ITS OWN STATED PARAMETERS ITS OWN FORMULAS GIVE C = 397 PF AND U = 0.18 J, SO THE RATIO IS 0.0078 N/W, ABOUT 27 TIMES LOWER THAN STATED; THIS CONTRADICTS THAT EXAMPLE'S STATED RATIO, NOT THE PAPER'S THRUST VALUE OR ITS OTHER EXAMPLES. 🥳🎊🚀🌍

Headline: The Xydous 2026 preprint gets its 0.21 N/W from a capacitance its own formula puts 26.5 times higher (confirmed and packaged). Nothing found on the 2013 E-Cat HT data.

Xydous preprint (new result): I used the preprint's own capacitor formula on its own plate dimensions and got 397 pF, not the stated 15 pF. Its own energy density gives the same answer. So the PFT example's thrust-to-power ratio is 0.0078 N/W, not 0.21. A fresh-context self-check and three independent skeptics all passed it. The result is narrow: it corrects one worked example's efficiency figure in an unreviewed preprint, not the model or its thrust value.
One panel detail to know: the first novelty skeptic's single fetch didn't follow a redirect on the status page, so it failed. That was a fetch problem, not a block. A second fresh skeptic, told only that, read the page and passed. Both files are kept and the repeat is disclosed in panel/gate.md.
E-Cat 2013 report (nothing found): the December test's excess can't come from any emissivity choice. Convection alone at the camera's temperatures already reaches about 381–469 W against 360 W in. In March, the ON/OFF trace fits ordinary electrical heating. It can't separate "meter right plus a hidden steady source" from "meter reading about 2.8 times low". Each needs a different heat capacity (about 1.6 vs 4.1 kJ/K), and the device's real value isn't stated. I wrote down two bench measurements that would decide it: a full mains energy meter and the device's heat capacity.

[JimBoot]

🏆 CONFIRMED (3 skeptics): Gamberale-Modanese TiHx preprint's HD intercept forces ≥7.3x the D2 its own D2 intercept allows. Packaged; see REPORT.md

🏆🎉⚡ CONFIRMED RESULT: THREE SKEPTICS AGREE!!! ⚡🎉🏆 IN GAMBERALE AND MODANESE'S PREPRINT RS-4679459 V1, IF ETA30 = 0.045 (95 % CI 0.037-0.053) IS READ AS HD/H2 AS THE PAPER DOES (EQ. 6) AND THE M/Z 4 SIGNAL AS D2, THEN THE PAPER'S OWN TABLE 1 RELATION [H2][D2] = K4[HD]^2, WITH K = 1/K4 <= 4 (NIST-JANAF, 250-2000 K) AND EQUAL ANALYSER SENSITIVITIES, REQUIRES A ZERO-PRESSURE M/Z 4 / M/Z 2 INTERCEPT OF AT LEAST 0.037^2/4 = 3.4225E-4, WHICH IS 1369/188 = 7.28 TIMES THE PAPER'S REPORTED UPPER BOUND 4.7E-5 FOR ETA40; THE TWO READINGS ARE JOINTLY CONSISTENT ONLY IF S(D2+)S(H2+)/S(HD+)^2 X 4/K <= 188/1369 = 0.1373, WHERE K IS THE ACTUAL RATIO [HD]^2/([H2][D2]). 🥳🎊🚀🌍

Headline: In the Gamberale-Modanese 2024 titanium-hydride "deuterium production" preprint, its own HD intercept requires at least 7.3 times more D2 than its own D2 intercept allows (confirmed, packaged).

Today's run started by scouting for new cases and picked this preprint, the lab's first titanium case, in your requested focus area of light-hydrogen metals. Seven lines of attack, each running its own code, checked the paper's mass-spectrometer numbers against each other and against its own isotope-exchange relation. The bound that relation needs held up two independent ways, using published heat-of-formation tables and molecular constants. Every escape route I could compute failed. A fresh self-review caught wording errors in the claim sentence, which I fixed in one repair round. After that, all three independent skeptics passed it.

Other findings, not part of the claim:

The paper's "280 times natural" is really about 141 times, because its number compares an HD/H2 ratio with a D/H ratio.
Its reference tank gas already reads 12–27 times natural, so the method fails its own control.
Its water peaks leave no room for the heavy water its own HDO reading implies. This could become a second claim in a later run.