When constants compensate
Fine-tuning is not nine separate target rings. See how structure, gravity, dark matter, and expansion trade leverage—and where the toy refuses to compensate.
Lukasz Szramuk · · updated · 5 min read
A one-dial experiment asks a clean question: what changes when this number moves and everything else stays fixed? Nature owes us no such convenience. A star responds to gravity, electromagnetism, particle masses, and reaction rates together. A galaxy forms through initial fluctuations, matter content, expansion, cooling, and time. The viable region is therefore not nine independent intervals but a shape in a space with many dimensions.
That shape can contain compensation. A change that closes one path may be offset partly by another change, or it may create a new failure elsewhere. ‘Partly’ matters: restoring galaxy formation does not automatically restore long-lived stars, calm planetary systems, chemistry, or observers. A compensation claim must name both the recovered condition and every condition it stopped checking.
A compensating pair reopens one model boundary. It does not prove the whole universe is rescued.
One dependency graph, not nine isolated windows
The Tuner contains one explicit multi-control structure rule. Q supplies the initial contrast; gravity and dark matter strengthen its collapse proxy; Λ competes with the result by accelerating separation. The dependency map below is generated from the same exported functions that decide every Tuner fate. Its three experiments start from our values, move one physical control, then add exactly one compensator.
Coupled constants · compensation lab
The dials do not move alone
Follow the five controls that meet at the Tuner's structure gate, then break one condition and add one declared compensator. The lab exposes the model's real dependency graph instead of treating nine windows as independent.
Model contract and scientific boundaries reviewed 28 August 2026
The Tuner's structure dependency map
Q, gravity, and dark matter combine into an effective structure amplitude. Dark energy competes with that result. Geometry, matter survival, nuclear binding, electromagnetism, and density retain separate first-failure gates in this toy.
Structure inputs
Qeff
Effective structure
Qeff = Q × √N × (0.4 + χ) · Λceiling = 15 × (Qeff / Q₀)³ᐟ²
Expansion pressure
ΛDark energyCompetes with growth by separating matter before it binds.Toy formula — not a cosmological fit
Bound structure before expansion wins
Five controls do not compensate here
The Tuner deliberately keeps these first-failure rules sharp. Real physics may connect some of them, but this model will not silently reopen their gates.
Choose a compensation test
Experiment stage
Observer gate closes
Expansion winsAt Λ = 80×, expansion separates matter before the original fluctuations build galaxies.
Changed controls
What this demonstrates
The model's Λ boundary depends on Q. A one-dimensional Λ window is therefore not a universal edge.
What it does not establish
This is an authored toy relation, not a measured probability distribution for Λ and Q or a claim that either parameter can adjust in response to the other.
Why the restored universes all fail the same way first
Lowering Q does more than approach the toy's minimum structure amplitude. It also lowers the amount of dark energy the same structure can tolerate. In the selected recipes, our unchanged Λ therefore outruns growth before the later ‘fog’ rule gets to become the first verdict. Adding gravity or dark-matter scaffolding raises the effective amplitude and its Λ ceiling together. The outcome change is a useful demonstration of causal ordering inside the code, not a calculation of real galaxy abundance.
Real parameter spaces can be wider—and stranger
Tegmark and Rees made the Q–Λ loophole explicit: an anthropic upper bound on Λ changes if Q varies too. Stellar calculations likewise find regions, rather than single intervals, in gravity, electromagnetism, and nuclear-reaction parameters. Some regions contain unfamiliar but sustained stars; others satisfy one nuclear criterion and fail primordial chemistry or planet formation. Moving from one dimension to several can widen a window without making the full space hospitable.
The opposite warning is equally important. A fundamental theory might correlate constants so that the compensating direction is unavailable. A measure over possible universes might make a viable corner extremely rare. And observer requirements are incomplete: stable carbon, a star, or a galaxy is not a probability for intelligence. Geometry of the allowed region answers ‘what combinations pass these stated tests?’ It does not answer ‘how likely are observers?’
What the Tuner deliberately keeps independent
D, η, ε, α, and Ω retain separate first-failure gates in this model. That is a disclosure, not a claim that deeper physics contains no coupling. Quark masses reshape nuclear stability; electromagnetism and gravity jointly constrain stars and planets; density and expansion share a cosmic history. Those relationships require calculations the Tuner does not perform. Its honest map includes only the compensation it actually computes.