This is a follow-up to 1 and 2. It is a status report: what survived further checking, what was retracted, and what the checks added.

Summary

  • Nothing in the algebraic framework selects \$\Lambda\$. The reason is now more precise than “we did not find it”. The algebra of a de Sitter observer is insensitive to \$\Lambda\$ except through a state-independent normalization of the trace and through the state on the center. Both are inputs, not outputs.
  • The normalization ambiguity is the measure problem in algebraic form. For a tracial state the central marginal is fixed by the normalization of the trace. The three natural conventions give three different cross-sector weights. Two of them are excluded by observation on opposite sides, and the third carries no \$\Lambda\$-dependence at all (Section 3).
  • A single clock records the sector only weakly. The mutual information between clock energy and a continuous sector label is of order one nat, against \$S_{\rm dS}\sim10^{122}\$ (Section 5).
  • Observers sharing a region of space share less than one might expect. In a finite-dimensional model with a quantum relative-position label, the shared algebra is classical in that label (Section 6).
  • Several earlier statements are retracted (Section 2).
ItemStatus
Fixed-volume ensemble gives \$Z(V)\approx e^{2S_{\rm dS}(V)}\$exact (saddle point)
One observer produces no interior extremum of the weightnumerical (grid) plus exact first-order
Three normalization conventions and their implied weightsexact formulas; orders of magnitude from approximate constants
Clock energy records a sector label with \$O(1)\$ natsnumerical, Monte Carlo cross-checked
Shared algebra classical in relative geometryexact in a toy; conjecture in de Sitter
Frame-loop holonomy equals Haar areaexact
Any selection of \$\Lambda\$not obtained

1. Where the first note left off

The first note 1 organized the problem around three structural facts. The observer algebra of Chen and Xu 3 is type II\$_\infty\$ with a nontrivial center encoding causal visibility, and \$\Lambda\$ does not appear in their construction (they set \$\ell_{\rm dS}=1\$). A Henneaux–Teitelboim sector 4 promotes \$\Lambda\$ to a variable conjugate to four-volume. The entropy of the observer algebra matches a generalized entropy only up to a state-independent constant 56. The note’s own program was to ask whether any state on the center, or any consistency condition between observers, could constrain \$\Lambda\$.

2. Retractions and corrections

These are statements made in intermediate drafts and discussion that did not survive further checking.

  1. “The normalization is fixed by matching to the horizon entropy” is [retracted]. In the standard treatment the additive constant \$S_0\$ is a state-independent ambiguity; when the clock energy is bounded below it is fixed by \$\mathrm{Tr}(1)=1\$, which makes the maximally mixed state have zero entropy 56. Matching \$S_{\max}=S_{\rm dS}\$ is an additional choice, not part of the construction.
  2. A Jones-index quantization constraint on the overlap of two observers is [retracted]. It needed the overlap inclusion to be an irreducible finite-index subfactor inclusion. The multi-observer algebra is not a factor, and the QFT part of the overlap is type III\$_1\$ before dressing, for which we have no reason to expect finite index.
  3. “The algebra cannot depend on \$\Lambda\$ because the hyperfinite II\$_1\$ factor is unique” needs restating. The relevant algebra is II\$_\infty\$ with a nontrivial center, and hyperfiniteness is not stated. What survives is weaker and sufficient: \$\Lambda\$ is not in the construction, so it can enter only through normalization and state.
  4. "\$\Lambda\$ is a free central \$\alpha\$-parameter whose measure is fixed by the gravitational path integral" is downgraded to an assumption. Harlow’s abstract states that in higher dimensions the gravitational path integral is too ambiguous to construct well-defined \$\alpha\$-states 7.
  5. “Mutual visibility is a single projection common to both observers” is [retracted]. Visibility of the other observer’s geodesic is symmetric, but visibility at the level of the relative group element (b’s clock reading inside a’s patch versus a’s inside b’s) is not: the two sets differ, so the two observers have different walls.
  6. “Entropy production seen through a shared algebra might carry new information” was tested in a toy model (Section 6). It reduces to the entropy production of the overlap region.

3. The normalization is the measure problem

3.1 The identity

Let the observer algebra over a family of sectors \$\lambda\$ (values of the cosmological constant) be a direct integral \$\mathcal A=\int^\oplus\mathcal A_\lambda\,d\mu(\lambda)\$ with \$\lambda\$ in the center. A trace on \$\mathcal A\$ is \$\tau_N=\int d\mu\,N(\lambda)\,\tau_\lambda\$ for any positive \$N\$. For the tracial state \$\omega=\tau_N/\tau_N(1)\$ (it exists only in the II\$_1\$ case), the marginal on the center is

$$p(\lambda)=\frac{N(\lambda)\,\tau_\lambda(1)}{\tau_N(1)}.$$

Relative entropy and modular data are blind to \$N\$, because rescaling the trace by a central function changes the Connes cocycle only by a central phase. The absolute entropy shifts by \$\int p\log N\$ (checked numerically in finite dimensions: the shift matched \$\langle\log N\rangle_p\$ to 14 digits). So the normalization is not an observable, but when the physical state is the equilibrium state across sectors, it is the prior. This is precisely the statement of eq. (3.10) of 6, applied to a family of sectors.

3.2 The explicit form

In the construction of De Vuyst, Eccles, Höhn and Kirklin 68 the trace for an observer with clock \$C_i\$ and KMS state \$\psi_S\$ at inverse temperature \$\beta\$ is

$$\mathrm{Tr}(\cdot)=e^{S_{0,i}}\,\langle\psi_S|\langle0|_i\,(\cdot)\,e^{-\beta H_i}\,|\psi_S\rangle|0\rangle_i , \qquad \mathrm{Tr}(1)=e^{S_{0,i}}\frac{Z_i}{2\pi},\quad Z_i=\int_{\sigma_i}e^{-\beta\epsilon}\,d\epsilon .$$

For a clock with energy spectrum \$\sigma_i=[m,\infty)\$ this gives \$Z_i=e^{-\beta m}/\beta\$. In de Sitter \$\beta=2\pi R=2\pi\sqrt{3/\Lambda}\$, so \$Z_i\$ depends on \$\Lambda\$.

3.3 Three conventions

ConventionChoice of \$S_0\$Sector weight \$\propto\$Pushes \$\Lambda\$
(a) same formula in every sector\$S_0=0\$\$Z/2\pi=e^{-2\pi mR}/(2\pi\beta)\$up (small \$R\$)
(b) normalized trace, as in 5\$\mathrm{Tr}(1)=1\$\$1\$no preference
(c) maximum entropy equals the horizon entropy\$S_{\max}=S_{\rm dS}\$\$e^{S_{\rm dS}}=e^{\pi R^2/G}\$down (\$\Lambda\to0\$)

Convention (a) is not canonical (the KMS temperature differs between sectors), but it is the naive choice. Convention (c) reproduces the Baum–Hawking-type weight 910. With approximate constants \$\Lambda\approx1.1\times10^{-52}\,\mathrm{m^{-2}}\$, one finds in Planck units \$R\approx1.0\times10^{61}\$ and \$S_{\rm dS}\approx3.3\times10^{122}\$. For a proton-mass clock \$\beta m\approx5\times10^{42}\$, and for a 70 kg observer \$\beta m\approx2\times10^{71}\$. Halving \$R\$ changes the log-weight by about \$+2.5\times10^{42}\$ under (a) and about \$-2.5\times10^{122}\$ under (c).

Consequence. Conventions (a) and (c) are excluded by observation on opposite sides, one toward Planckian \$\Lambda\$ and one toward \$\Lambda=0\$. Convention (b) survives, but it carries no \$\Lambda\$-dependence from the algebra. Whatever selects \$\Lambda\$ must therefore enter through the observer weight, not through the counting. This relies on the clock’s lower bound \$m\$ being \$\Lambda\$-independent and on the tracial-state reading; it fails in the multi-observer II\$_\infty\$ case, where \$\mathrm{Tr}(1)=\infty\$ and no tracial state exists.

3.4 A sense in which the count can be renormalized

In the semiclassical regime the entropy of a physical state is 6

$$S\approx S_0-S_{\rm rel}(\phi_S\Vert\psi_S)-\beta\sum_{C_i\in R}\langle H_i\rangle+S_R .$$

The area term enters through the relative entropy. The same paper shows that in an “antisemiclassical” regime, where a clock’s energy sits in a very narrow window, the leading entropy is independent of the QFT part of the state, so no area term appears; and that the total horizon entropy including all clocks is observer-independent 6. A count restricted to what a given observer can access can therefore lose the \$S_{\rm dS}\$ dependence, while a count of the total keeps it. Which count a measure should use is the same choice as before. This as an interpretation, not a result of 6.

4. Baseline joint computation

4.1 Fixed-volume ensemble

With \$\Lambda\$ conjugate to four-volume 4, integrating \$\Lambda\$ against the volume constraint gives, on the \$S^4\$ saddle,

$$Z(V)=\int d\Lambda\,\exp\!\Big[\frac{\Lambda V}{8\pi G}+\frac{3\pi}{G\Lambda}\Big]\;\approx\;e^{2S_{\rm dS}(V)},\qquad S_{\rm dS}(V)=\frac{\sqrt{6V}}{4G}.$$

The saddle sits at \$\Lambda_*=2\sqrt6\,\pi/\sqrt V\$, which reproduces \$V=24\pi^2/\Lambda^2\$. The second derivative is positive (a minimum along real \$\Lambda\$, as expected for a Legendre transform along a rotated contour). The factor 2 is not an error: the \$\Lambda\$-conjugate phase contributes \$e^{S}\$ and the fixed-\$\Lambda\$ sphere contributes another \$e^{S}\$. The weight grows with \$V\$, so the ensemble prefers large volume, i.e. small \$\Lambda\$. The relation \$S_{\rm dS}=\sqrt{6V}/4G\$ holds identically because all these quantities are functions of the single scale \$R\$; it is not independent evidence for anything.

4.2 Adding one observer

To first order in \$Gm/R\$ an observer of mass \$m\$ changes the weight to \$e^{S_{\rm dS}-2\pi mR}\$. With \$x=1/\sqrt\Lambda\$ the exponent is \$a x^2-bx\$ with \$a=3\pi/G\$ and \$b=2\sqrt3\,\pi m\$, whose stationary point is a minimum at \$R=Gm\$. For the exact Schwarzschild–de Sitter cosmological horizon, with the Nariai bound \$R\ge3\sqrt3\,Gm\$, we checked \$dS_c/dR>0\$ on a 2000-point grid up to \$R=200\,Gm\$, and that the first-order offset \$S_c-(\pi R^2-2\pi mR)\$ approaches a constant near \$-2\pi\$ (\$-6.30\$ at \$R=1000\,Gm\$). So a single observer produces no interior extremum in the physical range. We ignored one-loop factors and the phase of the sphere path integral; none changes the exponential behaviour.

5. What a clock can record about a sector

5.1 Weak records of a continuous label

If the sector label is a continuous variable and the clock energy is thermal in each sector, \$q\mid\beta\sim\mathrm{Exp}(\beta)\$. For a log-uniform prior on \$\beta\in[1,\rho]\$ the marginal is

$$f(q)=\frac{e^{-q}-e^{-\rho q}}{q\ln\rho},\qquad I(\Lambda;q)=h(q)-1+\tfrac12\ln\rho .$$
\$\rho=R_2/R_1\$210100\$10^4\$\$10^8\$
\$I\$ in nats0.0190.1650.4370.8881.459

The marginal normalizes to \$1.000000\$ and a Monte Carlo estimate at \$\rho=100\$ gives \$0.4368\pm0.0021\$. A scalar record of a continuous label carries a few nats at most, while \$S_{\rm dS}\sim10^{122}\$. The posterior over sectors given the clock energy is \$p(\Lambda\mid x)\propto p(\Lambda)\,\beta e^{-\beta x}\$, whose log-weight with the entropy prior is \$S_{\rm dS}-\beta x\$, the weight of Section 4.2.

5.2 Visibility labels and the charge relation

The center of the multi-observer algebra of 3 is generated by event-level visibility projections: conditions on the relative angle and on the other observer’s clock time lying inside a window. For two observers and no QFT excitations the constraints reduce to \$d_{a,i}+d_{b,i}=0\$ for the three generators, which 3 notes forces the observers to be antipodal. Using the classical charge vector of b, \$E_b(\cos\varphi\cosh s,\ \sin\varphi\cosh s,\ \sinh s)\$, and of a, \$(q_a,0,0)\$, the three constraints give \$\sinh s=0\$ and \$\sin\varphi=0\$: b is either coincident with a or antipodal, with \$q_a=\mp E_b\$.

Writing the left boost generator along a’s geodesic in terms of b’s right generators gives exactly

$$\mathrm{Ad}(g^{-1})B_1=a\,R+b\,B_1+c\,B_2,$$ $$a=-\sin\varphi\sinh t-\cos\varphi\sinh s\cosh t,\quad b=\cos\varphi\cosh s,\quad c=-\sin\varphi\cosh t-\cos\varphi\sinh s\sinh t,$$

with \$b^2+c^2-a^2=1\$. The \$B_1\$ coefficient is the classical relation \$q_a=-E_b\cos\varphi\cosh s\$; the other two components carry b’s remaining right charges, which do not commute (\$[R_{B_2},R_R]\propto E_b\$) and so fluctuate by at least \$\sim\sqrt{E_b}\$. This supports the classical relation for heavy observers and says nothing for \$E_b\sim O(1)\$. Positivity \$E_b\ge0\$ cannot be an operator statement on \$L^2(SO(1,2))\$, since boost generators have spectrum on all of \$\mathbb R\$ in unitary irreducible representations, so it must be imposed by projection.

5.3 Degradation with QFT energy

When the QFT carries boost energy \$h\$, \$q_a=-(h+d_{b,1})\$ and \$h\$ is not sign-definite. Modeling \$q_a=-h-\sigma e\$ with \$h\sim\mathcal N(0,w^2)\$ and \$\sigma=\pm1\$ for coincident versus antipodal b, the fraction of the visibility bit recoverable from the clock energy is:

Thermal clock, \$e\sim\mathrm{Exp}(1)\$ (units of \$T_{\rm dS}\$), noise \$w\$0.010.10.31310
\$I/\ln2\$0.9870.8790.6860.3180.0690.007
Heavy clock \$e=m\$, ratio \$m/w\$0.51234
\$I/\ln2\$0.1610.4860.9130.9941.000

Normalizations are \$1.000000\$, and Monte Carlo agrees (\$0.4763\pm0.0019\$ against \$0.47575\$ at \$w=0.3\$; \$0.2200\pm0.0020\$ against \$0.22024\$ at \$w=1\$). The record is robust for heavy clocks and fragile for thermal-scale ones. The model assumes \$h\$ is independent of the visibility label, which real excitations need not satisfy.

6. Shared algebras of non-complementary observers

6.1 What the literature gives

De Vuyst et al. show that algebras for different clock sets \$R_1,R_2\$ in the same region satisfy \$r(\mathcal A_{R_1})\cap r(\mathcal A_{R_2})=r(\mathcal A_{R_1\cap R_2})\$, which is trivial for disjoint sets when the modular flow is ergodic, while the algebras still do not commute 6. They also state that they assume each clock has access to a single fixed subregion, and that clocks with access to different subregions are a separate source of observer dependence 6. The patches of Chen and Xu 3 differ, so the overlap problem sits in that gap.

6.2 A finite-dimensional model

Take four qubits on a ring plus a four-state label \$x\$ for b’s offset. Observer a has \$\mathcal A_a=B(\text{label})\otimes B(\text{sites }0,1)\$, and b’s algebra is its conjugate by the controlled translation \$U=\sum_x|x\rangle\langle x|\otimes T^x\$, so b’s region is \$\{x,x+1\}\$. We computed everything by linear algebra, with rank thresholds of \$10^{-8}\$ against clear gaps.

Version\$\dim\mathcal A_a=\dim\mathcal A_b\$\$\dim(\mathcal A_a\cap\mathcal A_b)\$Center of the intersection\$\dim(\mathcal A_a\otimes_N\mathcal A_b)\$
label always accessible256254 blocks, sizes \$n_k=1,2,4,2\$6400
wall at \$x=2\$ (scalar block)16025same1552
asymmetric walls16025same1680 (blocks lopsided: 16 vs 48 and 48 vs 16)

The intersection is exactly the classical-label structure \$\bigoplus_x B(\mathcal H_{\rm overlap}(x))\$, with dimension \$16+4+1+4=25\$ matching the expectation for a classical label: only the diagonal label projections are shared, never the coherences \$|x\rangle\langle x'|\$. The relative entropy between a state with coherence between two labels and its dephased version, restricted to each algebra, is:

Pairon \$\mathcal A_a\$on \$\mathcal A_b\$on the intersection
both labels visible\$0.642\$ nats\$0.01\$–\$0.11\$ (state-dependent)\$\sim10^{-15}\$
across a wall of that observer\$\sim10^{-16}\$\$\sim10^{-16}\$\$\sim10^{-15}\$

Values near \$10^{-15}\$ are floating-point noise; the pattern of zeros is robust and the nonzero entries depend on the random site state. Monotonicity \$D_N\le\min(D_a,D_b)\$ held in every case. Coherence between relative geometries is therefore perspective-dependent: each observer sees some of it and the shared algebra sees none.

Conjecture (de Sitter). If each observer has access to the relative-frame variables, the QFT part of \$\mathcal A_a\cap\mathcal A_b\$ is the overlap algebra \$\mathcal A(M_a\cap M_{b(g)})\$ fibered over the relative position \$g\$, with \$g\$ central, and off-diagonal shift operators are excluded because they differ by a global isometry. In the opposite case of no operational access, De Vuyst et al. give the trivial intersection 6. The clock generators are unresolved.

6.3 Correspondences and holonomy

The overlap-induced correspondence \$L^2(\mathcal A_a)\otimes_N L^2(\mathcal A_b)\$ has dimension \$\sum_k\dim(\mathcal A_aq_k)\dim(q_k\mathcal A_b)/n_k^2\$, which we verified against explicit counts for four size choices (64, 144, 144, 144). It is invertible only if \$N=\mathcal A_a=\mathcal A_b\$, so distinct observers’ overlap correspondences are not invertible and the Picard group is the wrong home for them.

For the geometric holonomy, with \$g(\varphi,s,t)=e^{-\varphi R}e^{-sB_2}e^{-tB_1}\$ the Maurer–Cartan form is

$$g^{-1}dg=\big(-\cosh s\cosh t\,d\varphi-\sinh t\,ds\big)R+\big(\sinh s\,d\varphi-dt\big)B_1+\big(-\sinh t\cosh s\,d\varphi-\cosh t\,ds\big)B_2,$$

so a horizontal lift has \$dt=\sinh s\,d\varphi\$ and a closed loop in \$(\varphi,s)\$ shifts the proper time by the Haar area

$$\Delta t=\oint\sinh s\,d\varphi=\iint\cosh s\,ds\wedge d\varphi$$

The proper-time shift is modular flow on the ungauged type III\$_1\$ algebra (outer) and is implemented by \$e^{-itq}\$ in the gauged algebra (inner), so geometric holonomy is trivial in the Picard group after gauging.

6.4 Entropy production through a shared algebra

Under brickwork random circuits the entropy produced through \$\mathcal A_a\$, \$\mathcal A_b\$ and the intersection saturates at the finite-size Page values (\$0.92\$ for two of four qubits, exact \$0.922\$; \$0.60\$ for one qubit, exact \$0.600\$), and is exactly zero for the intersection when the regions do not overlap. The shared quantity is just the entropy production of the overlap region.

7. The environmental factor

To see where a scale can come from, we reproduced a Weinberg-style estimate 11 with a Press–Schechter-like collapsed fraction. For flat \$\Lambda\$CDM linear growth with fixed matter density, the asymptotic growth is exactly

$$D_\infty(\lambda)=\frac56\,\frac{\Gamma(5/6)\Gamma(2/3)}{\Gamma(3/2)}\,\lambda^{-1/3},\qquad F(\lambda)=\mathrm{erfc}\!\Big(\frac{\delta_c}{\sqrt2\,\sigma_\infty(\lambda)}\Big),\quad \sigma_\infty=\sigma_0\frac{D_\infty(\lambda)}{D_{\rm today}},$$

with \$\lambda=\rho_\Lambda/\rho_m\$, \$\delta_c=1.686\$, and the closed form checked against numerical growth to six digits (\$K=1.437283\$). With a flat prior in \$\rho_\Lambda\$:

\$\sigma_{\rm gal}\$ today\$F\$ at observed \$\Lambda\$median \$\Lambda\$ / observed\$P(\Lambda\le\text{obs})\$\$P(\Lambda\ge10\times\text{obs})\$
2.00.548110.0890.53
2.80.668310.0370.76
4.00.764890.0140.89

The observed value lies in the lower tail but is not excluded, and the sensitivity to \$\sigma_{\rm gal}\$ is large. This is not the calculation of the causal entropic principle 12, whose weight is entropy production in a causal diamond. It illustrates that the scale in such arguments comes from the fluctuation amplitude and matter density. A flat prior in \$\rho_\Lambda\$ is the assumption in the landscape discussions of 111213, and it is incompatible with an \$e^{S_{\rm dS}}\$ weight, which is why the normalization choice of Section 3 matters.

8. A two-dimensional laboratory

A two-dimensional analogue of the Henneaux–Teitelboim sector exists: unimodular JT gravity has a unimodular time conjugate to the cosmological constant 14. In the JT action \$\tfrac12\int\sqrt{-g}\,[\phi(R-2\lambda_J)-2\Lambda]\$ the shift \$\phi'=\phi+\Lambda/\lambda_J\$ removes the vacuum term and leaves a topological term,

$$\tfrac12\int\sqrt{-g}\,[\phi(R-2\lambda_J)-2\Lambda]\;=\;\tfrac12\int\sqrt{-g}\,[\phi'(R-2\lambda_J)]\;-\;\frac{\Lambda}{2\lambda_J}\int\sqrt{-g}\,R ,$$

which in two dimensions is proportional to the Euler characteristic. So a vacuum-energy ensemble is an ensemble over the topological coupling \$S_0\$ (the sign depends on conventions), and with the usual \$e^{S_0\chi}\$ weighting 15 one expects Baum–Hawking-type unbounded growth. This makes the laboratory useful for testing the machinery of normalization and shared algebras in a solvable setting, not for selecting \$\Lambda\$.

9. What remains

The pattern of the negative results is consistent. The structures examined so far are kinematical: they describe how sectors fit together given a state, and \$\Lambda\$ enters only through \$S_{\rm dS}\$, the trace normalization and the state on the center. A selection needs one more ingredient, and the most concrete candidate is a state principle. Equilibrium states appear throughout the algebraic constructions, in which an observer is a fluctuation costing the entropy deficit, while our universe has an observer-supporting low-entropy past. The relative entropy to the equilibrium state is the available entropy budget. A principled statement of the initial condition as a condition on that observer-independent total relative entropy is a natural target.

Open problems, in order of tractability:

  1. Prove or refute the de Sitter intersection conjecture of Section 6.2, including the clock generators.
  2. Read and use the long companion paper 8 on degenerate-clock superselection, which may already contain the clock-correlation corrections of 2.
  3. Carry out the 2D laboratory computation with an observer, with the sign conventions checked.
  4. Formulate the entropy-budget condition algebraically and test whether it constrains the sector weights.

References


  1. S. Aliei, “CC Problem and ER=EPR in dS”, blog post (November 2026), https://saeidaliei.github.io/posts/cosmological-constant-problem-algebraic-er-epr-ds/. ↩︎ ↩︎

  2. S. Aliei, “The Observer, The Observer, The Observer”, blog post, https://saeidaliei.github.io/posts/the-observer/. ↩︎ ↩︎

  3. Chen and Xu, “An algebra for covariant observers in de Sitter space”, arXiv:2511.00622; JHEP 09 (2026) 240. ↩︎ ↩︎ ↩︎ ↩︎

  4. M. Henneaux and C. Teitelboim, “The cosmological constant and general covariance”, Phys. Lett. B 222 (1989) 195. ↩︎ ↩︎

  5. V. Chandrasekaran, R. Longo, G. Penington and E. Witten, “An algebra of observables for de Sitter space”, JHEP 02 (2023) 082, arXiv:2206.10780. ↩︎ ↩︎ ↩︎

  6. J. De Vuyst, S. Eccles, P. A. Höhn and J. Kirklin, “Gravitational entropy is observer-dependent”, arXiv:2405.00114 (v3, April 2025). ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎ ↩︎

  7. D. Harlow, “Observers, \$\alpha\$-parameters, and the Hartle–Hawking state”, arXiv:2602.03835. ↩︎

  8. J. De Vuyst, S. Eccles, P. A. Höhn and J. Kirklin, “Crossed products and quantum reference frames: on the observer-dependence of gravitational entropy”, arXiv:2412.15502; JHEP 07 (2025) 063. ↩︎ ↩︎

  9. S. W. Hawking, “The cosmological constant is probably zero”, Phys. Lett. B 134 (1984) 403. ↩︎

  10. S. Coleman, “Why there is nothing rather than something: a theory of the cosmological constant”, Nucl. Phys. B 310 (1988) 643. ↩︎

  11. S. Weinberg, “Anthropic bound on the cosmological constant”, Phys. Rev. Lett. 59 (1987) 2607. ↩︎ ↩︎

  12. R. Bousso, R. Harnik, G. D. Kribs and G. Perez, “Predicting the cosmological constant from the causal entropic principle”, arXiv:hep-th/0702115. ↩︎ ↩︎

  13. R. Bousso and J. Polchinski, “Quantization of four-form fluxes and dynamical neutralization of the cosmological constant”, JHEP 06 (2000) 006, arXiv:hep-th/0004134. ↩︎

  14. Alexandre, Etkin and Rassouli, “Unimodular JT gravity and de Sitter quantum cosmology”, arXiv:2501.17213. ↩︎

  15. J. Maldacena, G. J. Turiaci and Z. Yang, “Two dimensional nearly de Sitter gravity”, arXiv:1904.01911. ↩︎