Probability and possibility rubric
This page fixes the vocabulary for reading the site. Its purpose is to stop probability language from being waved around carelessly.
Sometimes a chance-based story is still live. Sometimes it is not. The difference is not whether we like the explanation, but whether the proposed mechanism had enough real opportunities to do the work claimed for it.
Two different questions
When a source uses numbers, we ask two separate questions.
1) Single-shot odds. An argument may give a probability per try, written 1 in 10n (about one success expected in roughly 10n independent tries at that fixed chance).
2) Enough tries in the real world. An argument may also ask whether time and matter allow enough independent trials of a proposed mechanism. A first-pass picture:
- Trials available ≈ (relevant pieces of matter) × (relevant time in seconds) × (how often a new independent try can occur).
If a process would need vastly more successful random assemblies than could ever occur in our universe, it is not a viable explanation under ordinary assumptions. That is not a softer word for “unlikely.” It is starved for opportunities: the story never gets enough rolls to become credible.
Example: two fair six-sided dice
Roll two standard dice (each face 1–6, equally likely). There are 6 × 6 = 36 equally likely ordered outcomes—e.g. (1, 6) is different from (6, 1).
- Sum = 7 is the most common total (six ways: 1+6, 2+5, …), so P(sum 7) = 6/36 = 1/6—about 1 in 6, i.e. more likely than not among the eleven possible sums, and possible / plausible in this rubric for a single roll.
- Double six happens only as (6, 6), so P(double six) = 1/36—about 1 in 102 (between 1 in 102 and 1 in 103 on the single-shot table below): unlikely—sound arithmetic, no mystery.
The point of the example is not dice. The point is that probability by itself is not enough. We always have to ask how many real chances there were.
Sigma (particle physics)
Particle physics often states significance in standard deviations (σ) of a normal distribution, not as “1 in 10n.” A common gloss for a 5 σ discovery threshold is about 1 in 3.5 million (one-sided tail probability ~3 × 10−7, i.e. the same order as 1 in 106 or 1 in 107). That is a different convention from our 1 in 10n tables; it is still a useful anchor when you read physics news.
A rough cosmic trial budget
These figures are order-of-magnitude sketches, not exact physics. They exist to block a common mistake: treating the universe as if it had unlimited tries.
- Matter in the observable universe is often quoted at about 1080 baryons (a nucleons-scale count—definitions vary).
- Cosmic age is about 13.8 billion years, i.e. about 4.35 × 1017 seconds.
Rough bound (same order of magnitude only):
1080 × 1017 ≈ 1097.
That yields about 1097 atom-seconds: a back-of-the-envelope ceiling for naive “particle times second” sketches. No real process gets one independent lottery draw per baryon per second—coupling, chemistry, and causality fix what counts as a trial. 1097 is an intuition anchor, not a theorem. Treat it as a hard warning against multiplying your way to infinite opportunities.
Lloyd bound
Lloyd bound — Seth Lloyd, Computational capacity of the universe — on the order of ~10120 elementary physical operations on ~1090 bits of registered information (over cosmic history, in the paper’s model).
References:
- arXiv: quant-ph/0110141
- Journal: Physical Review Letters 88, 237901 (2002) — DOI 10.1103/PhysRevLett.88.237901
The ~10120 scale is a tighter, more meaningful cap than the naive 1097 atom-seconds sketch: it bounds how much computation the observable universe could have performed, not merely particles × seconds.
Under Lloyd’s model, ~10120 is the right order for an upper bound on blind search the cosmos could have afforded. It is not a proof about every metaphysical hypothesis. It is a physical ceiling on “try everything at random.”
Physical viability
If a proposed mechanism—under realistic assumptions—would require more independent trials than could have occurred in the observable universe (given matter, time, and known physics), then that mechanism is not a viable explanation here. Logical possibility is untouched. The failure mode is lack of opportunities—the mechanism never gets the runway it needs.
Not merely “1 in 10n” small—too many rolls for the universe to ever plausibly cash in.
Six levels
Per-try odds (1 in 10n) and trial-budget thinking (p × N: probability per try × number of tries) are blended in what follows. The symbols are scaffolding; the levels are what matter for reading the site.
| Level | Short name | Idea |
|---|---|---|
| 1 | Likely / expected | Many successes expected: p × N ≫ 1. Everyday “this should happen a lot,” like ordinary variation. |
| 2 | Possible / plausible | A few successes expected: p × N roughly in the 1 to 10 range. Reasonable chance within the opportunities at hand. |
| 3 | Rare but possible | p × N < 1 but not absurdly small. Could happen; you would not bank your life on it, but it is not crazy to treat it as still live. |
| 4 | Extremely improbable | p × N ≪ 1—often tiny even if you allow many tries (think “1 in 106 or harsher” per try when N is not astronomical). Would almost never occur in realistic opportunity counts. |
| 5 | Practical physical floor | Per-try odds so small that even “use the whole cosmic history as a search” does not help: order 1 in 10120 per independent random try lines up with Lloyd-style total-op budgets (see above). This is “not enough rolls in the universe” territory for blind search. |
| 6 | Physically unachievable for the proposed mechanism | The mechanism requires more independent trials than are physically possible in the observable universe or assumes fresh random assemblies beyond what cosmic limits allow. Conclusion: under those assumptions, that path is not a viable explanation—not merely beaten by a better story, but ruled out by the trial budget. |
Companion scale: 1 in 10n per try
When an argument quotes one number “per shot,” use this table:
| Approximate single-shot odds | What we mean |
|---|---|
| Better than 1 in 2 | More likely than not |
| 1 in 2 to 1 in 10 | Possible / plausible |
| 1 in 10 to 1 in 103 | Unlikely |
| 1 in 103 to 1 in 106 | Highly unlikely |
| 1 in 106 to 1 in 1012 | Extraordinarily unlikely |
| 1 in 1012 to 1 in 1020 | Remote—not a close race |
| 1 in 1020 or smaller (without a gigantic N) | Practically ruled out as a serious single-shot chance story |
| 1 in 10120 or smaller | Effectively impossible — cosmically negligible; undirected chance is not a live option per try (same order as total-universe operation budgets, e.g. Lloyd). |
| Nested scales (e.g. 1 in 1010123) | Beyond possibility for blind chance — astronomically past cosmic limits; not a competitor. |
"Effectively impossible" does not mean logically impossible. It means not a live empirical option under the stated assumptions. Reserve strict impossible for logical or mathematical impossibility.
What every cited number must answer
A number does no work unless the reader can see:
- Whose model it comes from.
- What p measures (one draw? one parameter? one universe in an ensemble?).
- What counts as a trial.
If a source does not quantify a claim, stay qualitative. Do not invent exponents to sound precise.
What follows from the rubric
Evidence in Level 5–6 does not mean another view is "a bit more plausible." It means—under the stated assumptions—the blind-process story is out of reach for the universe: too few trials, or per-trial odds so small that cosmic resources cannot close the gap.
That does not automatically prove design in every case. It does mean that unguided chance is no longer doing the explanatory work. At that point, law, constraint, bias, or design must be examined instead of repeating "maybe it happened randomly."