Six decades of attempts at the cosmological constant problem. Each approach captures something real — and each runs into a wall. Understanding where they fail is the prerequisite for understanding what a solution must do differently.
The Observational Target
Any viable dark energy theory must place its prediction inside the DESI 2024 confidence region — or explain why the deviation from Λ is a systematic artifact. The cosmological constant sits at (−1, 0); DESI excludes it at 2.5–3.9σ.
w₀–wₐ parameter space. DESI DR1 +CMB+DES-SN5YR confidence ellipses (arXiv:2404.03002, ρ ≈ −0.90). The CVC ray wₐ = −3(1+w₀) is the Causal Vacuum Correspondence prediction — a one-parameter family through ΛCDM. The CVC point at (−0.73, −0.82) falls within the DESI 2σ ellipse.
ΛCDM — The Standard Model That Explains Nothing
incomplete
The idea
Add Λ as a free parameter to the Friedmann equations and fit its value to observations. Accept that we don't know where it comes from.
What it gets right
✓Fits every cosmological observation to extraordinary precision
✓Only six free parameters describe the entire observable universe
✓Predicted CMB anisotropies later confirmed by WMAP and Planck
Where it breaks down
✗Λ has no theoretical explanation — it is a number we measured, not derived
✗Does not address why ρ_vac(obs) / ρ_vac(QFT) = 10⁻¹²¹
✗The coincidence problem: why does Λ dominate precisely at z ≈ 0.3?
✗DESI DR1 hints w ≠ −1 — if confirmed, ΛCDM is observationally wrong
Supersymmetry
incomplete
The idea
Every boson has a fermion superpartner. Their vacuum energy contributions have opposite signs and cancel. The residual is much smaller than the bare QFT prediction.
What it gets right
✓Reduces the discrepancy: cancels contributions up to the SUSY breaking scale
✓Motivated independently by gauge coupling unification and the hierarchy problem
✓Elegant mathematical structure
Where it breaks down
✗SUSY must be broken in our universe — broken SUSY still leaves ρ_vac ∼ M_SUSY⁴
✗For M_SUSY ∼ 1 TeV (electroweak scale), residual is still 10⁶⁰ times too large
✗LHC has found no SUSY particles up to ∼1–2 TeV — the preferred range
✗Even perfect SUSY cancellation at known scales leaves ∼60 orders of magnitude unexplained
Quintessence — Dynamical Scalar Fields
active
The idea
Replace Λ with a slowly rolling scalar field φ with potential V(φ). The field's kinetic and potential energy mimic dark energy, with w(z) ≠ −1 in general.
What it gets right
✓Naturally produces w ≠ −1 — matches the DESI 2024 DR1 signal
✓Tracker solutions exist: φ evolves toward the observed value from a wide range of initial conditions
✓Avoids the coincidence problem if the tracker attractor is efficient
Where it breaks down
✗Does not solve the 10¹²¹ problem — just replaces Λ with V₀(φ)
✗Still requires V₀ to be tuned to 10⁻⁴⁷ GeV⁴ — why is the potential so flat?
✗Requires a light scalar with m_φ ∼ H₀ ∼ 10⁻³³ eV — no particle physics motivation
✗Coupling to Standard Model fields must be suppressed to avoid fifth-force violations
The Anthropic Principle / String Landscape
not science
The idea
String theory predicts ∼10⁵⁰⁰ vacua with different values of Λ. Observers can only exist where Λ is small enough for galaxies to form. We observe small Λ because we're here.
What it gets right
✓Weinberg 1987 correctly predicted Λ would be non-zero before it was measured
✓Provides an order-of-magnitude argument for why Λ ≠ 0 but Λ ≪ Λ_QFT
✓Takes seriously that the fundamental constants may not be uniquely determined
Where it breaks down
✗Not a dynamical explanation — no mechanism selects our vacuum
✗Makes no testable predictions beyond the range of Λ we already know
✗Requires accepting a multiverse for which there is no direct evidence
✗Explains the coincidence problem only if the measure over vacua is precisely defined — which it isn't
Modified Gravity — f(R), Galileons, DGP
active
The idea
Modify the Einstein–Hilbert action by replacing R with f(R) or adding higher-derivative terms. Late-time cosmic acceleration emerges without any dark energy component.
What it gets right
✓Avoids introducing dark energy entirely — acceleration is geometric
✓f(R) can exactly reproduce the ΛCDM expansion history
✓Makes distinctive predictions for structure growth — testable with future surveys
Where it breaks down
✗Requires chameleon or Vainshtein mechanisms to hide modifications in the solar system
✗Gravitational wave speed constraint (GW170817): many models ruled out
✗Does not explain the coincidence problem — why does modification kick in now?
✗The 10¹²¹ vacuum energy problem remains entirely unaddressed
Unimodular Gravity / Vacuum Energy Sequestering
incomplete
The idea
In unimodular gravity, the determinant of the metric is fixed: det(g) = 1. This decouples vacuum energy from the gravitational field equations — Λ becomes an integration constant.
What it gets right
✓Vacuum energy from QFT does not source curvature — the 10¹²¹ discrepancy dissolves
✓GR is recovered in the classical limit
✓Sequestering variants extend this to loop corrections
Where it breaks down
✗Λ as integration constant still needs a value — why is it 10⁻⁴⁷ GeV⁴ and not zero?
✗Moves the problem: trading a fine-tuning for a boundary condition with no dynamical origin
✗Coincidence problem not addressed
✗Predicts w = −1 exactly — in tension with DESI hint
What a Real Solution Must Do
1Suppress or reinterpret the QFT vacuum energy without fine-tuning — not just rename the parameter
2Explain the coincidence problem dynamically — not as an accident, not anthropically
3Predict w(z) consistent with DESI 2024: w₀ ≈ −0.73, wₐ ≈ −1.05, or explain the deviation
4Address the 5σ Hubble tension without introducing larger inconsistencies
5Recover GR in the solar system and match binary pulsar timing