Finite Screen Spacetime: A Falsifiable Test for Late-Time Dark Energy

FDS-G1 predicts a narrow dark-sector pattern: background plus Weyl response, not ordinary matter growth.

Finite Screen Spacetime: A Falsifiable Test for Late-Time Dark Energy

Dark energy is one of the strangest placeholders in modern cosmology.

In the standard Lambda-CDM model, the late-time accelerated expansion of the universe is described by a cosmological constant. It behaves like a uniform vacuum-like component with equation of state w = -1. This model is simple, powerful, and extraordinarily successful as a baseline.

But it is also conceptually incomplete. We do not know why the cosmological constant has the observed value. We do not know whether it is truly constant. And recent large-scale-structure and distance measurements have made the late-time universe an increasingly interesting place to look for small but structured deviations from the simplest picture.

The usual response is to add flexibility.

Let w(a) vary.

Use a CPL parameterization.

Add modified-gravity functions.

Fit residuals.

Let the data decide.

That is a reasonable strategy, but it has a weakness: flexible models can describe many things after the fact. A sharper question is this:

If Lambda-CDM is not the final story, what specific pattern of deviation should we expect?

This post introduces one candidate answer from a framework I have been developing, called FDS-G1, or Finite Screen Spacetime.

The central late-time prediction is deliberately narrow:

s < 3

mu(a,k) ~= 1

Sigma(a,k) < 1

In plain language:

The signal should appear in the background expansion and the Weyl/lensing response, but not as a comparable modification of ordinary matter growth.

Or even shorter:

Background plus Weyl, not growth.

This is not a claim that Lambda-CDM is dead.

It is not a claim that general relativity has already failed.

It is not a claim that a new theory has been established.

It is a falsifiable finite-screen dark-sector candidate, with a replication kit, and with clear ways to fail.

What is Finite Screen Spacetime?

FDS stands for Finite Distinction Systems.

The broad idea is that physical systems do not maintain infinitely sharp, infinitely free distinctions. Any real system has finite capacity, finite boundary maintenance, finite update cost, finite memory, finite resolution, and finite access to its environment.

FDS-G1 applies this idea to gravity and cosmology.

Instead of starting from metric fields as the most primitive object, Finite Screen Spacetime starts from the idea of a finite causal-screen ledger: a boundary-like structure that carries finite distinguishability, entropy response, and causal accounting.

In this view, geometry is not discarded. Geometry is treated as an effective response structure arising from finite causal-screen accounting.

The ambitious theoretical claim is that ordinary Einstein gravity appears as a local equilibrium or closed-ledger limit of a deeper finite-screen response theory.

But this post is not mainly about defending the entire foundation.

The more immediate question is empirical:

Does Finite Screen Spacetime produce a narrow late-time cosmological branch that can be tested against Lambda-CDM, CPL, and modified-gravity alternatives?

That branch is what I call the G1DE-M3/4 branch.

The prediction: s < 3, mu ~= 1, Sigma < 1

The late-time G1 branch predicts three linked features.

First:

s < 3

Here s is a horizon-occupancy or finite-screen residual parameter. The Lambda-CDM-like zero-residual limit corresponds to s = 3. A value below 3 indicates that the late-time expansion history carries a finite-screen residual.

Second:

mu(a,k) ~= 1

The function mu controls the ordinary matter-growth response. If mu remains close to 1, then matter clustering and Ricci-like growth remain close to general relativity.

Third:

Sigma(a,k) — 1 = -(3/4)(3 — s) Rhat_H(a)

The function Sigma controls the Weyl/lensing response. In this branch, the Weyl channel carries a negative residual linked to the same finite-screen parameter s. The coefficient is not left free. It is locked to 3/4.

That is the important part.

The model is not simply saying:

Dark energy evolves somehow.

It is saying something more specific:

If the late-time residual is finite-screen in origin, then the background expansion and Weyl/lensing channel should move together, while ordinary matter growth remains close to general relativity.

This channel separation is what makes the proposal interesting.

Why this is different from a flexible dark-energy fit

A flexible model can often fit more data by adding more freedom.

That is not the goal here.

The G1DE-M3/4 branch is designed to be more constrained than a generic modified-gravity response. It competes against several alternatives:

Lambda-CDM.

CPL dark energy.

A more flexible G1DE-2 envelope.

A free-kappa version where the coefficient is not fixed.

A constant-Sigma model where the Weyl residual is not tied to the G1 horizon-response shape.

The key question is not merely:

Does G1 fit?

The stronger question is:

Does the locked, lower-flexibility model outperform more flexible alternatives?

In the current pilot comparison, the M3/4 branch is selected over Lambda-CDM, CPL, G1DE-2, free-kappa, and constant-Sigma in the reported medium-prior evidence comparison.

That is why the result is worth independent checking.

Not because it proves the theory, but because it gives critics something concrete to attack.

The strongest way to falsify it

A useful theory should not only say what would support it. It should say what would weaken it.

Here are several ways this branch could fail.

Failure mode 1: Lambda-CDM wins.

If future data return strongly to a strict cosmological constant, with no meaningful residual in background or lensing channels, the G1DE observational branch should be demoted.

Failure mode 2: CPL wins.

If a standard w0-wa dark-energy parameterization consistently outperforms the G1 branch across robust datasets and priors, then the finite-screen residual interpretation weakens.

Failure mode 3: free-kappa wins.

If the data prefer a free response coefficient rather than the fixed 3/4 coefficient, then the coefficient lock is not supported.

That would not necessarily kill every finite-screen idea, but it would demote the specific M3/4 branch.

Failure mode 4: constant-Sigma wins.

If a constant Weyl suppression explains the data better than the horizon-shaped G1 output, then the predicted response shape is not doing useful work.

Failure mode 5: growth also shifts.

If future weak-lensing, redshift-space distortion, E_G, or 3x2pt analyses require:

abs(mu — 1) comparable to abs(Sigma — 1)

then the “Weyl but not growth” separation fails.

That would be a serious hit.

Failure mode 6: expanded lensing does not support Sigma < 1.

The G1 branch needs the Weyl/lensing channel to carry the residual. If expanded lensing data do not support this, the dark-sector interpretation fails.

These are not cosmetic caveats. They are the actual pressure points.

Why a replication kit matters

A speculative cosmological model is not very useful if it only exists as a PDF.

For that reason, I have released a replication kit containing:

model specifications;

prior definitions;

likelihood conventions;

benchmark outputs;

validation levels;

failure and demotion conditions;

a reference implementation.

The reference Python code is not meant to be authoritative. The authoritative target is the specification, benchmark, and validation protocol.

The strongest independent test would be a reimplementation that does not use my code.

A good replication attempt should ask:

Can the best-fit values be reproduced?

Can the evidence ranking be reproduced?

Does M3/4 still win under different samplers?

Does the ranking survive different random seeds?

Does it survive prior widening?

Does it survive updated DESI, Euclid, weak-lensing, and 3x2pt likelihoods?

Does the channel pattern remain background plus Weyl, not growth?

That is the standard I want the model held to.

What this would mean if it survived

If the M3/4 branch survives independent replication and future data, the implication would be larger than “one more dark-energy model fits.”

It would suggest that the late-time dark sector may not be best understood as a free fluid or arbitrary equation of state.

It may instead be a residual of finite causal-screen closure: a small failure of the cosmic ledger to behave as the perfect equilibrium limit.

In that interpretation, general relativity is not simply thrown away. It becomes the equilibrium branch. The residual appears only when the finite-screen accounting is not perfectly closed at cosmological scales.

That is why the channel separation matters.

If the residual is finite-screen in origin, it should not generically behave like a free scalar field, a free CPL fit, or an arbitrary modified-gravity function. It should have a constrained output pattern.

For G1DE-M3/4, that pattern is:

s < 3

mu(a,k) ~= 1

Sigma(a,k) < 1

Again:

Background plus Weyl, not growth.

Current status

The current status is best described as:

pilot evidence-selected, not production-confirmed.

That means the result is interesting enough to test, but not established enough to trust as a conclusion.

The next steps are clear:

independent replication;

full production evidence refinement;

wide-prior stress tests;

updated DESI likelihoods;

Euclid and full 3x2pt lensing tests;

adversarial reimplementation from the written specification.

If those tests fail, the branch should be demoted.

If they succeed, the result becomes difficult to ignore.

Links

Paper & Replication kit :https://zenodo.org/records/20382013

Watch Overview: https://www.youtube.com/watch?v=sm2XBJKIUbU

GitHub: https://github.com/yiningwu-research/Distinction-Theory/tree/main/papers/FDS_G1

Theory website: distinctiontheory.org

Prediction and claim status: https://www.distinctiontheory.org/predictions

I am the author of this model, so this post should be read as a proposal rather than as established consensus.