Nobody tunes the dials

Every other page here argues about fine-tuning. This one measures it — inside a toy, which is the only place a claim of that shape can be measured at all. Nine constants, nine tracks, a uniform random position on each, ten thousand times over. Every draw goes through the same fate engine the Universe Tuner runs when you move a slider, and the same one the Graveyard uses to decide what killed each of its exhibits. There is no second model hiding behind this page.

A draw is uniform in slider space rather than in value space. Five of the nine tracks are logarithmic, so a universe is as likely to be drawn with a tenth of our gravity as with ten times ours. The sixth constant is the dimension dial, which offers one through six, and only three produces stable atoms and stable orbits — so one draw in six dies of geometry before anything else gets a chance to go wrong.

The run is a pure function of its seed, and the seed is in the address bar. A link to this page shows you the same ten thousand universes it showed whoever sent it, in the same order, down to which square in the grid is which. Press Draw again for a different ten thousand.

The fine-tuning literature contains numbers like one part in 10^120. This page reports one in a few thousand. Both can be true, because they answer different questions.

The famous figure for the cosmological constant compares the value we observe against the value quantum field theory predicts for empty space — a range about 120 orders of magnitude wide. This toy's dark-energy track runs from a hundredth of our value to a thousand times it, which is roughly the range the anthropic-bound literature works in after Weinberg, and a vanishingly thin slice of the other one. Widen the track and the survival rate collapses with it. Narrow it around our own value and the universe stops looking tuned at all.

Which range is the right one to draw from is not a question this page can settle, and nobody else has settled it either: that is the measure problem, and it is why a probability over universes is so much harder to state than it first appears. What this page can do is be explicit about the ranges it used. They are one line per dial in the source, and the windows they are judged against come from the same file.

Three things, none of which depend on the exact prior. The first is when universes die. Most of the deaths counted above happen in the first three minutes: the nuclear glue is either too weak to light fusion at all or so strong that every proton pairs off immediately, and between them those two outcomes take more than half of every run. Cosmology gets its turn much later, and has far less to do.

The second is that the tightest constraint in this model is not the one popular articles lead with. Dark energy has the famously narrow window; here it is among the most forgiving dials on the board, because its ceiling moves with how lumpy the universe is. The nuclear constant and the number of spatial dimensions filter harder than anything else, and they do it in the first few minutes.

The third is the gap between the two rates in the box above. Treating the nine windows as nine independent locks understates how often a universe survives, by a factor this run computes while you watch. That is not a rounding error. It is Victor Stenger's objection to the fine-tuning argument — that varying one constant at a time exaggerates how special our settings are — turning up here as a measurement rather than as an argument.