Comparing Quantum-Gravity Approaches

Quantum-gravity programmes answer different questions with different objects: a perturbative field theory may calculate an infrared amplitude, a renormalization-group study may find a fixed point in a truncation, and a discrete construction may establish a kinematical spectrum before controlling its continuum dynamics. A fair comparison therefore aligns claims by evidential level rather than awarding points for vocabulary. The governing question is: which claims are derived, numerically supported, conjectured, or absent in each programme?

Required background. Einstein Equations defines the classical target. Quantum Mechanics supplies states and observables.

Helpful background. Flat-Spacetime Quantum Field Theory supplies amplitudes and renormalization, while Quantum Information clarifies operational observables and factorization assumptions.

Comparison contract

Every entry is assessed along the same axes:

1. Kinematics: fundamental or effective degrees of freedom, state space, gauge redundancy, and boundary data.
2. Dynamics: action, constraint, history sum, or flow equation; regulator and approximation.
3. Continuum and correspondence: existence and control of a limit recovering general relativity or established quantum field theory.
4. Observables: a gauge-invariant quantity computed with uncertainties, not only formal variables.
5. Discrimination: an empirical prediction that cannot be absorbed into conventional low-energy coefficients or backgrounds.

These axes prevent a category error: an exact kinematical construction is stronger than a conjectured one in its own category, but it cannot be ranked as though it were a computed continuum observable. Likewise, a controlled infrared amplitude is a genuine prediction without being a proposal for ultraviolet microscopic degrees of freedom.

The scope is intentionally selective. It compares low-energy gravitational EFT, functional-renormalization-group asymptotic safety, canonical or spin-network approaches, and causal dynamical triangulations. String and holographic constructions are treated in their dedicated volumes. The matrix is not a census, and omission is not a negative verdict.

Results and limitations

The cutoff is 2026-08-20. “Result” means what follows within the stated assumptions; “limitation” names the first inference that the cited work does not establish.

Programme	Degrees of freedom and kinematics	Dynamics and observable	Continuum, correspondence, and GR result	Evidence, source, date, and limitation
Low-energy gravitational EFT	Metric fluctuations and light matter about a declared background; diffeomorphism or BRST redundancy	Derivative-expanded action; on-shell amplitudes and long-distance nonanalytic terms	Systematic recovery of classical GR plus calculable quantum corrections for $GE^2\ll1$	Derived order by order; Donoghue 2023, §§2–6, assessed 2026-08-20. It does not specify the ultraviolet completion, and local higher-curvature coefficients require matching or measurement.
Functional-RG asymptotic safety	Scale-dependent effective action for metric and ghosts, usually in a background-field formulation	Exact flow equation projected onto a finite operator truncation; fixed-point coordinates and critical exponents	Non-Gaussian fixed points recur across many truncations, suggesting a finite-dimensional ultraviolet critical surface	Numerically supported within truncations; Reuter 1998, §§III–VI, assessed 2026-08-20. Regulator, gauge, background dependence, convergence of the untruncated theory, and Lorentzian observables remain limitations.
Canonical loop and spin-network programme	Holonomy–flux algebra, spin-network states, constraints, and boundary states	Hamiltonian or covariant spin-foam dynamics; geometric spectra and transition amplitudes	Discrete area and volume spectra are controlled kinematical results; several semiclassical sectors reproduce pieces of classical geometry	Derived kinematics with model- and truncation-dependent dynamics; Rovelli and Smolin 1995, §§2–6, assessed 2026-08-20. A unique physical Hilbert space, full continuum dynamics, and generic low-energy observables are not established by the spectrum alone.
Causal dynamical triangulations	Piecewise-flat causal triangulations summed at fixed topology and lattice spacing	Regulated Lorentzian-inspired history sum evaluated numerically; volume profiles, spectral dimension, phase structure	A four-dimensional de Sitter-like macroscopic phase appears in simulations	Numerically supported in a regulated ensemble; Ambjørn, Jurkiewicz, and Loll 2005, §§2–5, assessed 2026-08-20. Continuum critical behavior, observable completeness, matter sectors, and universality beyond explored ensembles remain live questions.

The rows are deliberately asymmetric. EFT has the strongest connection to measured low-energy degrees of freedom and controlled observables, but it declines to answer the microscopic ultraviolet question. Other programmes address that question more directly, yet their strongest current results often concern a truncation, regulated ensemble, or kinematical sector. Parallel presentation does not imply equal support.

Conservative and permissive readings

The strongest conservative reading says that only low-energy EFT currently supplies generic, systematically improvable quantum-gravitational predictions in a domain continuously connected to tested physics. Fixed points, geometric spectra, and emergent macroscopic phases are important programme-internal results, but no listed ultraviolet programme has completed the chain from microscopic definition through a regulator-independent continuum limit to a distinctive empirical observable.

The strongest permissive reading says that robustness across regulators, truncations, boundary states, or ensembles can provide meaningful nonempirical support before the final observable is available. Recurring non-Gaussian fixed points may indicate genuine universality; semiclassical amplitudes and geometric spectra may reveal the right variables; an extended four-dimensional phase may show that macroscopic spacetime can emerge dynamically. These are reasonable research judgements provided each robustness claim names what was varied and what remained invariant.

These competing positions differ over how much weight programme-internal robustness deserves before a generic continuum observable has been obtained; they do not disagree about which calculations have actually been completed.

Neither reading licenses two common shortcuts. Mathematical consistency alone is not empirical confirmation. Conversely, lack of a Planck-energy experiment is not lack of scientific content: controlled correspondence limits, anomaly freedom, regulator independence, and universal infrared predictions can eliminate candidate dynamics. The unresolved status lies in whether those partial successes can be joined into a complete chain.

A discretized correspondence check

A one-dimensional periodic lattice gives a solvable model of why “continuum limit” needs an observable and an adverse discretization check. For a free field with lattice spacing $a$, the nearest-neighbor Laplacian has plane-wave eigenvalue

$$\widehat k_n^2=\frac{4}{a^2}\sin^2\!\left(\frac{ka}{2}\right) =k^2-\frac{a^2k^4}{12}+O(a^4k^6).$$

Thus the dispersion relation $\omega^2=m^2+\widehat k^2$ recovers the continuum result at fixed physical $k$ with relative error $O(a^2k^2)$. It does not converge uniformly for modes near the lattice cutoff $k\sim\pi/a$.

Now change the regulator by using the five-point improved Laplacian. Its eigenvalue is

$$\widehat k_i^2 =\frac{30-32\cos(ka)+2\cos(2ka)}{12a^2} =k^2-\frac{a^4k^6}{90}+O(a^6k^8).$$

The detailed high-momentum spectrum changes, but the low-energy continuum dispersion is robust and the leading error improves from $a^2$ to $a^4$. This finite example separates three claims: a regulated model exists; a chosen observable converges; and the continuum value is regulator-independent. A quantum-gravity simulation that shows only the first has not established the other two. A simulation that shows convergence under one refinement path should still vary the discretization, phase, topology, fit window, and observable.

What strengthens a comparison

A new result strengthens a comparison only when it establishes a specific physical or mathematical connection. A larger truncation strengthens a fixed-point claim if critical exponents and relevant directions stabilize under regulator and gauge variation. A refined discrete calculation strengthens a continuum claim if dimensionless observables scale toward a common limit and finite-size effects are bounded. A semiclassical calculation strengthens correspondence if a second observable, not used to tune parameters, agrees with established physics.

Empirical status requires a further step: publish the likelihood, calibration model, selections, conventional alternatives, and data or sufficient statistics. A phenomenological ansatz attached to a programme is not automatically a prediction of its microscopic dynamics. Null results should be mapped back only to the programme sector that actually derives the tested operator.

Exercises

Exercise. For the nearest-neighbor lattice dispersion above, compute the fractional error in $\widehat k^2$ at $ka=0.2$ through leading order. If the spacing is halved while physical $k$ is held fixed, by what factor should that leading error fall? Explain what failure of this scaling would imply in a numerical continuum study.

Solution

The expansion gives

$$\frac{\widehat k_n^2-k^2}{k^2} =-\frac{(ka)^2}{12}+O((ka)^4).$$

At $ka=0.2$, the leading error is $-0.04/12\simeq-3.33\times10^{-3}$, or about $-0.33\%$. Halving $a$ at fixed $k$ reduces an $a^2$ error by a factor of four. If the measured residual does not approach that factor in the asymptotic regime, higher-order terms, finite-volume effects, solver error, or a different continuum trajectory is controlling the calculation. One must diagnose those effects before claiming convergence.

What could change these conclusions

The conclusions could change when an independent calculation changes a cited result, a larger controlled truncation or refinement alters stability, a regulator-independent continuum observable is established or refuted, a programme derives a discriminating prediction, or evidence after 2026-08-20 changes the relative assessment.

References

• Ambjørn, Jan, Jerzy Jurkiewicz, and Renate Loll. “Reconstructing the Universe.” Physical Review D 72 (2005): 064014. DOI.
• Donoghue, John F. “Quantum General Relativity and Effective Field Theory.” 2023. arXiv:2211.09902.
• Reuter, Martin. “Nonperturbative Evolution Equation for Quantum Gravity.” Physical Review D 57 (1998): 971–985. DOI.
• Rovelli, Carlo, and Lee Smolin. “Discreteness of Area and Volume in Quantum Gravity.” Nuclear Physics B 442 (1995): 593–622; erratum 456 (1995): 753. DOI.