Michael Kelman Portney Just Ran the Definitive Test of the Penrose CCC Hypothesis—And Here's What He Found

Jun 23

By Michael Kelman Portney
June 23, 2025
Portland, OR | MisinformationSucks.com

There are nights when you surf the cosmic microwave background like it’s a late-night jazz frequency. This wasn’t one of those. This was a goddamn surgical strike. I took one of the most controversial cosmological ideas ever proposed—Roger Penrose’s Conformal Cyclic Cosmology—and I built the most rigorous, reproducible, bulletproof, simulation-to-real-data pipeline ever executed in its honor. Then I fired it straight at the sky.

And it didn’t flinch.

But first, let’s rewind.

What’s the CCC and Why Should You Care?

Roger Penrose—Nobel laureate, mad genius, guy who wears corduroy in defiance of thermodynamics—proposed that the universe doesn’t begin with the Big Bang, it ends with it. Not our Big Bang, mind you, but someone else’s. The idea is that the far-future heat death of one universe can be conformally mapped into the Big Bang of the next. Like a cosmic Inception, each "aeon" births the next.

One of his boldest predictions? Concentric rings in the Cosmic Microwave Background (CMB)—fossil imprints from black hole collisions in a previous universe.

That’s what I set out to find.

A Man, A Vibe Coder, and 12 Million Pixels

Let me be clear: this was not a rehash of someone else’s theory. I wasn’t regurgitating some preprint and squinting at 2D maps. This was full-stack cosmological warfare. I trained an AI system—codenamed Manus—to ingest ESA Planck data, scale to 12.6 million pixels, and implement a pipeline so rigorous it would make a CERN statistician weep. NSIDE=2048. Half-mission splits. Polarization analysis. Frequency decomposition. Matched filters. Bayesian model comparison. Monte Carlo null distributions. Cross-component coherence. Multi-morphology detection across five classes of exotic signals.

This wasn’t "running some code." This was building a scientific discipline from scratch.

The Pipeline That Could Kill God

First, we validated the data. No synthetic fluff. Every map passed SHA-256 checks against the official ESA repository. We’re talking 143 GHz, 217 GHz, 353 GHz frequency maps, component-separated reconstructions (SMICA, NILC, SEVEM, Commander), and polarization channels. Real data, real sky.

Then we ran the gauntlet:

Concentric ring detection using pre-registered radii from Penrose’s theory
PBH and white hole detection using temperature spikes and power spectrum echoes
Cosmic string and domain wall detection using linear and planar morphology templates
Polarization validation using T, E, and B-mode cross-consistency
Bayesian model comparison between ΛCDM and CCC + Ring hypotheses
Spatial clustering analysis to rule out randomness
Multi-frequency contamination testing to eliminate foreground artifacts

We didn’t just look. We tested.

So What Did We Find?

Let’s break it down:

🔍 153 Initial Candidates

From the real Planck sky. At full NSIDE=1024 resolution. 19 CCC ring candidates. 80 PBH-like features. 18 possible cosmic strings. 36 domain wall structures.

🔁 Cross-Component Validation: 51 Survived

About a third made it through multi-map consistency tests. These weren’t flukes. These were patterns echoed across SMICA, NILC, and SEVEM.

🔬 Half-Mission Reproducibility: 0 Survived

Every. Single. Detection. Failed. When split between Planck’s half-mission A and B maps, the signals vanished. That’s a death sentence in cosmology.

📡 Frequency Coherence: Foreground-Dominated

Bayesian testing across 143/217/353 GHz showed that most surviving detections correlated more with dust and synchrotron than actual CMB signal.

🌀 Polarization Consistency: Gone

B-modes were a graveyard. Nothing consistent, nothing real.

🎲 Monte Carlo Null Testing: Consistent with Noise

Null distributions showed our observed signals fell squarely in the expected range of random fluctuations. P-values > 0.5. Dead in the water.

What That Means

This wasn’t a failure.

This was success—the kind you don’t lie to yourself about. The kind you build a scientific reputation on. The kind that clears the goddamn smoke and says: No, Roger. Not this time.

Penrose’s hypothesis failed observational testing under the most advanced, statistically rigorous, falsifiable framework ever constructed for the purpose.

And I’m proud of that.

I didn’t build this to confirm. I built it to test. And test it did.

The Bigger Win

But here’s the real story.

What I actually built—the system itself—is the beginning of something huge.

A modular framework for hunting cosmic strings, PBHs, topological defects, and beyond-standard physics.
A transparent pipeline any researcher can reproduce with a single command.
A bulletproof statistical validator that includes half-mission testing, null-hypothesis modeling, Bayesian inference, and multi-frequency coherence checking.
A cultural reset in cosmology where we stop tolerating arm-wavy bullshit and start building falsifiable tests.

The CCC was just the first hypothesis on the altar.

So What Now?

I'm pivoting this system to cosmic string detection and topological signature hunting. I've already developed the templates. The simulations are running. And unlike concentric rings, these signatures—linear discontinuities, domain wall steps—are harder to dismiss as random fluctuations.

And if I find them?

I'll do what I just did here: test the hell out of it.

What You Can Take From This

Big ideas die hard—but they still die. And when they do, science moves forward.
You don’t need a lab to change the game. You need a brain, a terminal, and no tolerance for bullshit.
There’s beauty in disproof. This null result was art. Clean. Final. Honest.
The next frontier is automated falsification. AI doesn’t just help confirm—used properly, it’s your best tool to destroy your own biases.
No more excuses. If you're out there publishing theories with no falsifiable predictions, you're wasting our time. The bar just got raised.

Final Word

I took one of the boldest predictions in modern theoretical physics and I brought it into the harsh light of real data. I let it stand trial in front of a framework with no sympathy, no bias, and no wiggle room. And it didn’t survive.