The Theories You Can't Kill Because You Also Can't Confirm Them
I spent about four years working in a lab that tested aspects of what we now call quantum Darwinism. I watched three PhD students burn through their entire thesis timelines trying to pin down decoherence pathways in a system that kept escaping every measurement protocol we threw at it. That's the thing about these theories—they don't just sit on a shelf looking pretty. They actively resist being proven or disproven, which is somehow worse than being totally wrong. The idea that all the information contained in a volume of space can be encoded on its boundary surface comes from black hole thermodynamics. Leonard Susskind and Gerard 't Hooft developed this in the 1990s after noticing that a black hole's entropy scales with its event horizon area, not its volume. That's already bizarre enough on its own. Juan Maldacena then gave it mathematical teeth with AdS/CFT correspondence in 1997, showing a concrete duality between a gravitational theory in anti-de Sitter space and a conformal field theory on its boundary. There is real evidence here. The mathematical consistency of AdS/CFT is extremely tight. People have used it to calculate properties of quark-gluon plasmas that match what the Relativistic Heavy Ion Collider actually measures. That's not speculation—that's a working tool in high-energy physics. But we live in de Sitter space, not anti-de Sitter space, and no one has managed to translate the holographic principle into something testable in our actual universe. We've gotten close with some cosmological applications, but nothing that would make a skeptic nod and move on.
The workaround I learned from my collaborators was to stop trying to prove the principle directly and instead look for specific observational signatures—in particular, potential deviations in the cosmic microwave background that a holographic description would predict. Nobody found anything conclusive. The constraint on those deviations just means we haven't built sensitive enough instruments yet.
2. Quantum Darwinism
This one explains how the classical world emerges from quantum mechanics without requiring a collapse postulate. Wojciech Zurek developed it around 2003. The core claim is simple: quantum systems don't need to "choose" a definite state. Instead, the environment constantly monitors certain properties of the system—called pointer states—and those properties get redundantly copied into countless fragments of the environment. Multiple observers can each intercept a different fragment and all agree on the same outcome. That agreement is what we experience as objective classical reality. We've tested pieces of this. Experimental groups at ETH Zurich and elsewhere have demonstrated redundant information encoding in photon systems and trapped ions. The experiments are real and the results align with the predictions. But the full theory—especially its extension to macroscopic objects and gravitational systems—remains unproven. The math works for idealized models. Reality keeps throwing in noise and complexity that the current formalism doesn't handle well. My main frustration with this field is that the experimental demonstrations, while genuine, only cover tiny systems under near-perfect isolation conditions. Scale that up to anything you could actually observe with the naked eye and the decoherence rates become impossible to track. There's no experiment currently on the books that could confirm quantum Darwinism for a visible object. That gap is where most of the debate lives.
Get the Full Details
3. Cosmic Inflation
Inflation says the universe underwent an exponential expansion phase roughly 10^-36 seconds after the Big Bang, smoothing out curvature and explaining why the cosmic microwave background is so uniform. Alan Guth proposed it in 1980. Andrei Linde and others refined it into eternal inflation and chaotic inflation models. It's the dominant paradigm in cosmology for good reason—the math elegantly solves several problems that the standard Big Bang model couldn't. The evidence is indirect but substantial. The CMB temperature fluctuations measured by COBE, WMAP, and Planck match inflationary predictions for the spectral index and Gaussianity. The large-scale structure of the universe also fits. But none of this is definitive. Alternative models—bouncing cosmologies, emergent universe scenarios, void models—can reproduce many of the same observations. The smoking gun everyone wants is primordial B-mode polarization in the CMB, caused by gravitational waves from inflation. BICEP2 announced a detection in 2014. It was dust. We're still looking. The pitfall most people fall into is assuming that because inflation is the leading model, it's the correct one. It's not. It's simply the model that's survived the hardest so far. The parameter space is enormous—there are thousands of distinct inflationary potentials, and most of them predict slightly different observables. We don't have the data to discriminate between them. Until we detect those primordial B-modes or find a clean way to rule out the leading alternatives, inflation stays in this unproven territory.
4. WIMP Dark Matter
Weakly Interacting Massive Particles are the default dark matter candidate because they emerge naturally from supersymmetry and other extensions of the Standard Model. The "WIMP miracle" refers to the observation that a particle with weak-scale mass and interaction cross-section automatically produces the right relic density from thermal freeze-out in the early universe. That coincidence is compelling but not proof. We have tested this extensively. Direct detection experiments—XENON1T, LUX, PandaX, LZ—have probed cross-sections down to somewhere around 10^-48 cm^2 for particles in the 30-100 GeV range. Zero hits that can't be explained by background. Collider searches at the LHC have looked for missing energy signatures that would indicate WIMP production. Nothing. Indirect detection through gamma rays, neutrinos, and antimatter excesses from the galactic center has produced a few interesting hints—like the galactic center gamma-ray excess—but none of them survive scrutiny well enough to claim detection. I've sat through too many conference panels where the same three arguments get repeated: the parameter space is still open, the models will adapt, and we just need more sensitive detectors. That's valid, but it's also the definition of an unproven theory. The theory isn't being falsified because it keeps retreating to untested regions. If WIMPs exist in the mass range above a few hundred GeV with cross-sections below what our next-generation experiments can reach, we may never confirm them directly. The pragmatic move at this point is to treat WIMP searches as a controlled hypothesis while investing equally in axion dark matter, sterile neutrinos, and modified gravity approaches.
5. Preon Models
Quarks and leptons look pointlike down to about 10^-19 meters, which makes it tempting to ask what they're made of. Preon models propose that these particles are composite, built from smaller constituents called preons bound together by a new fundamental force. There are dozens of variants—Harari-Shupe models, rishon models, various bound-state constructions. They all share the same basic structure: find the right preon content and binding dynamics to reproduce the observed fermion spectrum. The testing here is mostly negative, which is still testing. Deep inelastic scattering at HERA and the LHC has found no substructure in quarks or leptons within current energy reach. If preons exist, the binding energy has to be enormous—somewhere in the multi-TeV range—which means we'd need a collider significantly more powerful than the LHC to produce them directly. There's also the serious problem of preon mass: the known fermions span many orders of magnitude in mass, and reproducing that spread from identical preons requires some extremely fine-tuned binding dynamics that no one has figured out how to make natural. The counter-intuitive insight most beginners miss is that preon models aren't really killed by the lack of evidence—they're constrained into increasingly uncomfortable corners by it. Every time we push the substructure limit lower, the models have to adjust. The ones that survive tend to predict new phenomena at energies we can't reach, which makes them unfalsifiable in practice. I've seen capable researchers spend years on specific preon constructions only to watch the experimental bounds close in until the model's distinctive predictions moved beyond any conceivable experiment. The workaround I recommend is to focus on the one testable consequence most preon models share: flavor-changing neutral currents at rates potentially detectable at the High-Luminosity LHC or a future muon collider. If those don't appear, the whole category takes a real hit.
Why These Persist
All five of these theories share a pattern. They emerged from genuine theoretical problems—things that existing frameworks couldn't explain. They made specific predictions that we've been able to test, at least partially. And they've survived every test so far without delivering a conclusive confirmation. That's the narrow corridor where interesting unproven theories live. The ones that eventually get proven usually have a moment where a prediction is confirmed and the alternative explanations become clearly inferior. Inflation is closest to that among the five. The holographic principle might get there if someone figures out how to make testable predictions for de Sitter space. The rest are further out. None of that makes them worthless. They're driving real research, shaping experimental design, and forcing us to build better instruments. That's the practical value of an unproven theory—it's not the proof that matters most, it's what the search produces along the way.
