Can String Theory Ever Be Tested? Part II

What Would Actually Count as Evidence?

This article continues Can String Theory Ever Be Tested?, which examined the experimental pathways through supersymmetry, axions, extra dimensions, cosmic strings, cosmology, and gravitational waves.

String theory does not suffer from a complete absence of testable consequences. It suffers from something subtler: many of its accessible consequences are not unique to string theory.

A new axion would be revolutionary, but axions can arise outside string theory. Supersymmetric particles would transform particle physics, but supersymmetry existed as a quantum-field-theory concept before being incorporated into string models. Cosmic strings could leave gravitational-wave signals, yet not every cosmic string must be a fundamental superstring. Extra dimensions would overturn our understanding of spacetime, but extra-dimensional theories are not exclusively string-theoretic.

This creates a crucial distinction between discovering something that string theory allows and discovering something that string theory specifically predicts.

Part I concluded that the most realistic route to empirical confirmation is convergence: several independent observations aligning with the same underlying construction. Part II asks the harder question:

What combination of evidence would justify saying that nature is described by string theory—or at least by a recognisably string-derived theory of quantum gravity?

The answer requires more than a list of possible discoveries. It requires an architecture of evidence.

---

1. Evidence Is Not the Same as Compatibility

A theory gains little scientific credit merely because it can accommodate an observation after the fact.

If an experiment finds a new particle and theorists subsequently search the enormous landscape of string compactifications until they locate a model containing something similar, the result demonstrates compatibility. It does not necessarily demonstrate prediction.

This distinction can be divided into four levels.

Level 1: Mathematical compatibility

An observation does not contradict the framework.

For example, a measured axion-like particle may fit inside some string-inspired compactification. This keeps the framework viable but provides weak positive evidence because many other theories may fit the same result.

Level 2: Generic expectation

A phenomenon appears naturally or frequently across broad classes of string models.

Axions, moduli, hidden sectors, additional gauge fields, and extended objects often belong to this category. Their discovery would make a string-derived description more plausible, but it would not select string theory uniquely.

Level 3: Correlated prediction

A particular construction predicts several related quantities: particle masses, coupling ratios, decay channels, cosmological abundances, or gravitational-wave features.

If those relationships are measured as predicted, the evidence becomes substantially stronger. The power lies in the correlation, not merely in the existence of an individual particle.

Level 4: Overdetermined reconstruction

Independent experiments recover enough information to reconstruct the same hidden mathematical structure.

This would be the strongest realistic case. Collider data, precision measurements, cosmology, and gravitational waves might all point toward one compactification geometry, symmetry-breaking mechanism, or moduli-stabilisation scheme.

The key principle is simple:

The strongest evidence would not be a single exotic object. It would be a network of measurements whose relationships are difficult to reproduce without the same underlying theory.

---

2. Why One Discovery Would Probably Not Be Enough

Physics history contains celebrated single-event discoveries, but those events were decisive because they occurred inside tightly constrained theoretical systems.

The discovery of the Higgs boson did not stand alone. The Standard Model specified its role, production mechanisms, decay channels, spin, and coupling structure. Experiments did not merely find a new resonance near 125 GeV. They progressively verified that its behaviour resembled the field responsible for electroweak symmetry breaking.

String theory faces a more difficult situation. It is not one low-energy model with a short parameter list. It is a framework capable of generating many effective theories depending on:

• the number and geometry of compact dimensions;
• the arrangement of branes and fluxes;
• the mechanism that breaks supersymmetry;
• the values of moduli fields;
• vacuum selection;
• inflation and reheating;
• the thermal history of the universe;
• the spectrum of hidden sectors.

This flexibility does not make testing logically impossible. It changes the form a successful test must take.

Finding one axion-like particle would establish new physics. Finding several axions with masses and coupling relationships characteristic of a specific compactification could reveal an underlying “axiverse.” Discovering supersymmetry would establish a new symmetry of nature. Measuring a superpartner spectrum that simultaneously fits gauge unification, dark matter, and a calculable string construction would go much further.

The experimental target must therefore shift from isolated objects to structured patterns.

---

3. A Five-Channel Test of String-Derived Physics

A persuasive empirical programme would combine at least five observational channels.

Channel	Possible signal	Why it matters	Main limitation
Particle colliders	Superpartners, resonances, long-lived particles	Reveals low-energy particle spectrum and symmetries	New particles may have non-string explanations
Axion and dark-sector searches	One or more axions, dark photons, hidden gauge sectors	Such sectors are widespread in compactifications	Individual particles are not unique to strings
Precision gravity	Deviations from inverse-square gravity or equivalence	Could expose extra dimensions or new scalar fields	Most simple models are already strongly constrained
Cosmology	Dark radiation, primordial tensors, non-Gaussianity, evolving fields	Probes energy scales inaccessible to colliders	Cosmological conclusions depend on model assumptions
Gravitational waves	Cosmic-string networks or early-universe backgrounds	Could detect extended objects and ancient phase transitions	Astrophysical and field-theoretic alternatives exist

No single row provides decisive confirmation. The opportunity appears where the rows intersect.

Suppose experiments identified:

1. a family of axion-like particles;
2. a hidden gauge sector;
3. a long-lived supersymmetric particle;
4. a stochastic gravitational-wave background;
5. correlated cosmological parameters.

If one controlled compactification predicted all five—including their quantitative relationships—before the measurements were combined, the evidential strength would be far greater than the sum of five unrelated anomalies.

---

4. The Axiverse: From One Particle to a Spectrum

Axions are among the most promising low-energy descendants of string theory.

The original QCD axion was introduced to solve the strong CP problem: why strong nuclear interactions appear to preserve a symmetry they could, in principle, violate. String compactifications frequently generate additional axion-like fields when higher-dimensional fields are integrated over cycles in the compact geometry.

This can produce not one axion but an entire spectrum—the so-called string axiverse.

What would constitute weak evidence?

The discovery of a single axion-like particle with no further structural information.

It would be a major discovery in particle physics and cosmology, especially if it accounted for dark matter. But many non-string models can generate axions or axion-like particles.

What would constitute stronger evidence?

A set of axions exhibiting relationships among:

• masses;
• photon couplings;
• electron and nucleon couplings;
• decay constants;
• cosmological abundances;
• isocurvature signatures;
• black-hole superradiance effects.

A compactification can, in principle, connect some of these properties to the sizes and topology of hidden dimensions. If several measured axions corresponded to the same geometric structure, experiments would begin to function as a form of indirect spectroscopy of extra dimensions.

What would be truly persuasive?

The strongest scenario would involve multiple detection methods observing compatible parts of the same axion sector:

• a haloscope detects axion dark matter;
• helioscopes observe solar axion conversion;
• laboratory experiments measure photon coupling;
• astrophysical observations constrain stellar production;
• black-hole spin measurements identify mass ranges affected by superradiance;
• cosmology measures the relic abundance or isocurvature behaviour.

Each instrument would examine a different consequence of the same underlying fields. Agreement across them would constrain alternative explanations.

Even then, the discovery would support a string-derived axion sector rather than prove every aspect of string theory. Scientific confirmation would remain modular.

---

5. Supersymmetry After the First LHC Era

Supersymmetry remains central to many string constructions, but its experimental status has changed.

The simplest expectation that coloured superpartners would appear quickly at the Large Hadron Collider has not been realised. ATLAS reports that Run 2 searches excluded gluinos up to approximately 2.4 TeV, top squarks up to roughly 1.2 TeV, and certain chargino or neutralino scenarios up to about 1 TeV. These limits are model-dependent, but they have removed large regions of once-popular parameter space. The High-Luminosity LHC is expected to deliver about ten times more data over its operational lifetime, extending sensitivity to rarer and more difficult signatures. ATLAS Run 2 physics summary

This is not evidence that supersymmetry is impossible. It is evidence that nature has not selected many of its simplest accessible versions.

If supersymmetry is discovered, what must be measured?

A convincing string connection would require more than identifying one superpartner. Researchers would need to reconstruct:

• the pattern of superpartner masses;
• the mechanism of supersymmetry breaking;
• mixing angles and quantum numbers;
• relationships among gauge couplings;
• dark-matter stability and abundance;
• possible long-lived or displaced decays;
• connections to hidden sectors;
• whether the parameters can arise from a controlled compactification.

Different string constructions can produce recognisably different supersymmetry-breaking patterns. Some may yield approximately universal particle masses at high energies; others may show anomaly-mediated, gauge-mediated, moduli-mediated, or mixed behaviour.

The measured spectrum could therefore act like a fingerprint.

What if the LHC finds nothing?

Continued null results would further weaken string models predicting accessible low-energy supersymmetry. They would not eliminate string theory as a whole because supersymmetry may be broken at much higher energies.

However, moving every unsuccessful prediction beyond experimental reach would create a methodological problem. A scientifically productive programme must state which model families are ruled out as data accumulate.

Null results are not failures of science. They are how the landscape becomes smaller.

---

6. Cosmic Superstrings and the Gravitational-Wave Spectrum

A fundamental string is far too small to photograph directly. Yet some early-universe scenarios allow string theory to produce enormous one-dimensional objects stretched across astronomical distances.

These cosmic superstrings could form networks, oscillate, reconnect, and emit gravitational waves.

The relevant observable is often the dimensionless string tension:$$G\mu,$$Gμ,

where $G$G is Newton’s gravitational constant and $\mu$μ is the energy per unit length of the string.

A network may generate:

• individual bursts from cusps or kinks;
• a stochastic gravitational-wave background;
• signals extending across several frequency bands;
• gravitational lensing;
• possible effects on the cosmic microwave background.

The importance of a multi-band signal

A background detected at one frequency could be explained by several phenomena, including astrophysical black-hole binaries or early-universe phase transitions.

A cosmic-string network, however, can produce a broad spectrum. Pulsar timing arrays, ground-based detectors, and future space observatories probe different frequency ranges. A signal observed across multiple bands with the expected spectral changes would be far more informative.

LISA was formally adopted by the European Space Agency in 2024 and is expected to launch in 2035. Its three spacecraft will form a triangular laser interferometer with arms approximately 2.5 million kilometres long. Its frequency range will open a new window between pulsar timing observations and ground-based detectors. European Space Agency

Would cosmic strings prove string theory?

No. Conventional field theories can also generate cosmic strings.

The stronger case would require properties associated with cosmic superstrings, such as:

• reconnection probabilities different from ordinary field-theory strings;
• multiple species of strings;
• junctions between different string types;
• a tension spectrum;
• correlated signatures from the same brane-inflation model;
• compatibility with independently measured cosmological parameters.

A network containing several string species and characteristic junction behaviour would be more difficult to imitate than a generic stochastic background.

The ideal result would again be a pattern rather than a solitary signal.

---

7. Dark Energy: A Clue, Not a Confirmation

The Dark Energy Spectroscopic Instrument has produced the largest high-resolution three-dimensional map of the universe. Results based on its first three years of observations strengthened hints that dark energy may evolve with time rather than remain a perfect cosmological constant.

DESI reported that the significance depends on which supernova dataset is combined with its measurements and other cosmological evidence. The disagreement with the standard constant-dark-energy model had not reached the threshold required for discovery. The full DESI survey is expected to test the pattern more precisely. DESI three-year results The planned survey map was completed in 2026, while analysis of the full dataset continues. DESI survey milestone

An evolving dark-energy component would be enormously important. It could indicate a dynamical field rather than vacuum energy fixed at a constant value.

But it would not automatically confirm string theory.

Why the connection is difficult

String cosmology has long struggled with controlled descriptions of stable de Sitter space—spacetime with persistent positive vacuum energy. Some researchers argue that slowly evolving fields may be more natural in quantum gravity. Swampland conjectures have attempted to express this intuition as quantitative restrictions on low-energy effective theories.

These conjectures remain actively debated. They are not universal theorems of string theory, and many proposed formulations face counterexamples, refinements, or disputes over their domains of validity.

If dark energy evolves, the relevant questions would include:

• Does its equation of state follow a specific predicted trajectory?
• Can the responsible field remain consistent with fifth-force constraints?
• Does it couple to ordinary matter?
• Is its potential compatible with a controlled compactification?
• Does the same construction predict inflation, dark radiation, or particle properties?
• Were these relationships calculated before the cosmological fit was known?

Evolving dark energy could make certain string-inspired models more interesting. It could also support many non-string alternatives. It is a clue only when embedded within a larger predictive structure.

---

8. Can the Geometry of Hidden Dimensions Be Reconstructed?

Perhaps the most ambitious possibility is that low-energy physics could reveal the shape of dimensions we cannot see.

In string theory, compact geometry influences:

• the number of particle generations;
• gauge symmetries;
• Yukawa couplings;
• neutrino properties;
• axion spectra;
• supersymmetry breaking;
• moduli fields;
• the strength of effective interactions.

This suggests an inverse problem:

Given the particle spectrum and cosmological observables, can researchers infer the compactification that generated them?

In principle, the task resembles reconstructing the design of a musical instrument from the spectrum of notes it produces. In practice, many different geometries may generate similar low-energy physics. This is a problem of degeneracy.

The role of computation and artificial intelligence

Modern research increasingly uses:

• machine learning to classify compactification geometries;
• symbolic computation to calculate spectra and couplings;
• topological data analysis;
• automated searches through flux vacua;
• differentiable simulators and surrogate models;
• Bayesian inference to compare model families with data.

These tools can help identify which structures produce a given effective theory. They can also reveal statistical patterns across regions of the landscape.

However, computational success is not experimental evidence. An AI system that finds a compactification matching known observations may only demonstrate that a fit exists.

Its scientific value rises sharply when it generates new predictions that are subsequently tested.

A credible workflow would therefore be:

1. use existing data to restrict model families;
2. identify the surviving compactification structures;
3. calculate new correlated observables;
4. publish those predictions and uncertainties;
5. test them with independent datasets;
6. remove constructions that fail.

Machine learning can accelerate this cycle, but it cannot replace it.

---

9. The Bayesian Logic of Converging Evidence

The question “Does this discovery prove string theory?” is often too crude.

Scientific evidence is better treated as a comparison among hypotheses.

Let $H_S$HS​ represent a particular string-derived model, $H_A$HA​ an alternative theory, and $D$D the observed data. Bayesian reasoning compares how probable the data would be under each hypothesis:$$\frac{P(H_S\mid D)}{P(H_A\mid D)} = \frac{P(D\mid H_S)}{P(D\mid H_A)} \frac{P(H_S)}{P(H_A)}.$$P(HA​∣D)P(HS​∣D)​=P(D∣HA​)P(D∣HS​)​P(HA​)P(HS​)​.

The central quantity is the likelihood ratio:$$\frac{P(D\mid H_S)}{P(D\mid H_A)}.$$P(D∣HA​)P(D∣HS​)​.

If both theories easily produce an axion, discovering an axion may change their relative credibility only slightly.

If the string-derived model predicts three axion masses, two coupling ratios, a specific dark-radiation contribution, and a gravitational-wave feature—and all are observed—then the combined likelihood ratio may become large.

There is an important warning: the predictions must not be chosen after inspecting the evidence. Searching millions of models until one matches the data introduces a look-elsewhere problem on an enormous scale.

A rigorous programme should record:

• which models were considered;
• which parameters were adjustable;
• which predictions were made in advance;
• how selection effects were handled;
• how many alternative models could produce similar outcomes;
• what future observations would falsify the preferred explanation.

Without this accounting, apparent convergence can be manufactured through flexibility.

---

10. What Would Falsify a String Model?

The broadest form of string theory may be too extensive to be eliminated by one feasible experiment. Specific string models are much more vulnerable—and should be.

A model can fail if it predicts:

• particles below experimental limits that are not found;
• excessive dark radiation;
• unstable matter;
• unacceptable proton decay;
• an axion abundance inconsistent with cosmology;
• moduli that disrupt nucleosynthesis;
• cosmic strings above observational bounds;
• primordial gravitational waves already excluded by the cosmic microwave background;
• fifth forces or violations of equivalence that experiments do not see;
• a vacuum that is mathematically inconsistent or insufficiently controlled.

This is where testability becomes concrete.

The meaningful unit of empirical evaluation is usually not “all of string theory.” It is a defined compactification plus its cosmological history and mechanism for connecting high-energy structure to measurable physics.

A healthy research culture should label outcomes clearly

• Excluded: the construction conflicts with data.
• Disfavoured: it fits worse than alternatives.
• Viable: it survives current constraints.
• Supported: it predicted evidence better than rivals.
• Strongly supported: several independent observations select its correlated structure.
• Confirmed within a domain: its distinctive predictions repeatedly succeed, while serious alternatives fail.

“Not yet excluded” must never be confused with “probably true.”

---

11. A Plausible Discovery Sequence

What might progress actually look like?

Stage 1: An anomaly

An experiment detects an unexpected resonance, axion candidate, gravitational-wave background, or departure from standard cosmology.

At this point, the result is evidence for an unexplained phenomenon—not for string theory.

Stage 2: Independent confirmation

Another instrument observes the same phenomenon or a connected effect. Instrumental errors and local astrophysical explanations become less likely.

Stage 3: Structural measurement

Researchers measure masses, coupling ratios, polarisation, spectral shape, decay channels, or redshift evolution.

The discovery develops an identifiable fingerprint.

Stage 4: Model discrimination

Specific string-derived and non-string theories are compared using the complete dataset. Most candidate explanations fail.

Stage 5: Novel prediction

The surviving construction predicts an additional phenomenon that was not used to build the model.

This is the decisive transition from fitting to forecasting.

Stage 6: Cross-domain verification

The new prediction is confirmed in a different experimental domain—for example, a particle model predicts a cosmological relic or a gravitational-wave feature.

Stage 7: Overdetermination

More observables are measured than are needed to determine the model’s parameters. The remaining successful predictions become genuine tests rather than adjustable fits.

No single stage would announce “the end of physics.” Together, they could establish a compelling empirical bridge between the observable universe and a deeper quantum-gravitational structure.

---

12. Three Possible Futures

Future A: Convergence

Axions, hidden sectors, gravitational waves, and particle data align with a controlled class of compactifications. New predictions succeed.

String theory would not become unquestionable, but it could acquire a status comparable to other empirically established frameworks: valid within a tested domain and open to deeper refinement.

Future B: Productive restriction

No decisive signal appears, but increasingly precise observations exclude large regions of the landscape. String research becomes more predictive because fewer constructions remain compatible with nature.

This would still represent scientific progress.

Future C: Persistent non-selection

New physics is discovered, but string-derived models never predict it more successfully than simpler alternatives. Connections are repeatedly made only after observations are known.

In that situation, string theory may remain valuable as a mathematical framework, a source of dualities, or a laboratory for quantum gravity, while losing credibility as the uniquely promising description of our universe.

The scientific community must be prepared for all three outcomes.

---

13. Conclusion: From a Theory of Possibilities to a Theory of Patterns

The decisive test of string theory is unlikely to arrive as a photograph of a vibrating string or a single particle carrying a label that says “made in ten dimensions.”

It is more likely to emerge through reconstruction.

We may first discover an axion, hidden sector, unusual gravitational-wave spectrum, or unexpected cosmological field. On its own, each signal will have alternative explanations. The question will be whether their measured relationships reveal a common structure.

The strongest empirical case would require:

• a specified string-derived construction;
• quantitative predictions made before all relevant data are known;
• several independent observational channels;
• more measured quantities than adjustable parameters;
• successful novel forecasts;
• transparent comparison with non-string alternatives;
• clear conditions under which the model would fail.

This is a demanding standard, but foundational claims demand demanding evidence.

The future of string theory should not be judged by whether every experimental result can be made compatible with some corner of its landscape. It should be judged by whether the framework can identify a narrow route through that landscape before nature reveals where the route leads.

Part I asked whether string theory could be tested.

The answer was yes—piece by piece.

Part II asks what would make those pieces convincing.

The answer is convergence, overdetermination, and risk: many measurements pointing toward one structure, more predictions than freedoms, and a genuine possibility that nature could say no.

Only then would string theory move from describing worlds that might exist to explaining why this world does.

Internal links:

• Can String Theory Ever Be Tested? — Part I
• Unlocking the Universe: How String Theory Connects Quantum Physics and Gravity