Kinds of Separation
Independence is never free — it has to be earned. Once a proposed separation has been run through the Separation Certificate, it settles into one of a few recognizable patterns. This page walks through what each looks like in practice, using concrete cases from the framework's own results rather than examples invented for illustration — and it keeps the honest middle ground, the separations that are still open, plainly in view.
Earned by observation
Some separations hold because observation itself pays for them. The split is real — but its reality comes from a measured fact about the world, not from the abstract constraints alone. These are honest and useful; they are just not claims the framework can say it derived.
Worked example: the observed particle spectrum
The specific set of particles, charges, and generations that make up the Standard Model is treated as an observed input, not something the constraints produce from scratch. Any result that leans on it is reported honestly as conditional on that observed content: given this particle content, the rest follows — but the particle content itself was measured, not derived. The product form of the gauge group is a second case of the same kind: it is earned in part by the observed fact that the Standard Model's gauge symmetry really does factor into commuting pieces.
Earned by the constraints alone
Some separations are stronger still: the foundational constraints force them, with no extra unproven assumption smuggled in. This is the strongest and rarest category, because a separation here was never a modeling choice — it is the only option left standing once everything else already established is imposed at once. The test is exacting: it is not enough for the split to be consistent with the constraints; the constraints have to leave no alternative.
Worked example: no unpaid "spectator" sectors
The question was whether extra, disconnected pieces of matter — particles with no forced coupling to anything else, bolted onto the observed spectrum for no structural reason — are allowed. The Nonseparability constraint answers directly: such an unpaid spectator would be an extra observable label, modulus, or excitation with no anchor in anything actually observed, and that is precisely what the constraint forbids. So the elimination is resolved — forced by the constraints alone, not merely consistent with them: unpaid spectator sectors and disconnected vectorlike additions are ruled out by the constraints themselves, not by a modeling preference.
It is worth being exact about the limit of that same result, because it is a clean example of honest scope-setting. Eliminating unpaid spectators does not by itself explain why there are exactly three generations of matter rather than some other number. That count rests on a separate, sharper topological calculation — a rigid whole-number index of the frozen shape, computed two independent ways — that takes the observed particle content as a given input and returns exactly three left-handed families from the geometry. Because it depends on that observed input, the generation count belongs with the observation-earned separations above, not in this stronger, forced-by-the-constraints category.
Fails the check
Some proposed separations look perfectly reasonable in ordinary language yet fail the certificate outright — most often because they would smuggle in an unpaid new scale, an unpaid new sector, or a split chosen after the fact to land on a desired answer.
Illustrative failures (not an exhaustive list)
- An independent hidden sector introduced with no observed anchor and no forcing argument — a split asserted rather than paid for.
- A geometric product factorization asserted with no gauge-group justification and no Nonseparability accounting — the very mistake the Standard-Model-as-interface work was built to catch and correct.
- A distinction drawn at a resolution finer than the framework's own finite-record floor can support — a split the world has no way to register.
- A branch or geometry chosen only after its numeric target was already in view — a violation of causal order, no matter how mathematically tidy the resulting split happens to look.
Calling one of these a failure is not a permanent verdict on the underlying physics. It is a verdict on a specific proposed argument for the separation. A different argument — one backed by a genuine observed anchor or a real forcing proof — could still carry the same split later. What fails here is the reasoning offered, not the possibility itself.
The honest middle ground: open
Most interesting questions do not resolve cleanly on the first pass, and this vocabulary would be dishonest if it pretended otherwise. Some separations are simply open — not yet settled either way. The live example is the fork over whether the Standard Model's product structure is the deep geometry itself or only the low-energy face of a single unified geometry (see The Standard Model as Interface). The product structure itself is solid and paid for; the deeper-unity reading is an explicitly open, unproven conjecture — nobody has yet produced a specific unified geometry and shown it reduces correctly. It is a research fork that changes nothing already established unless and until it changes a finite record: neither earned-and-final, nor forced-by-the-constraints, nor a failed argument — open, and named as such.
| Pattern | What it means | Worked example on this site |
|---|---|---|
| Earned by observation | Accepted because an observed anchor covers the cost; honest, but not derived. | The observed particle spectrum; the SM's product gauge group; the three-generation count. |
| Earned by the constraints alone | Forced by the constraints — no alternative left, not merely consistent. | Elimination of unpaid spectator / vectorlike sectors. |
| Fails the check | The proposed argument for it does not hold up (a different one still might). | An unpaid hidden sector; a product assumed with no gauge-group justification. |
| Open | Not yet resolved either way; named honestly as unsettled. | Whether the SM product is the deep geometry or a low-energy interface. |