Paper 2: the two-level architecture

Paper 2 is the architecture paper. It answers a question that sounds administrative and is actually load-bearing: when the framework says an observer is inside one branch of a multiverse while a substrate holds many branches, which one is the quantum mechanics happening in? The answer is both, and keeping the two levels apart without contradiction is the paper’s real work.

The two levels

At the observer level, physics looks ordinary. An observer inside a branch sees one world, measurements with outcomes, statistics obeying the Born rule, and information conserved in the everyday sense that nothing is lost in a measurement, only redistributed into correlators with the environment. The observer’s quantum mechanics is closed and needs no reference to anything outside the branch. This closure is a requirement, not a convenience: an observer who needed substrate facts to do physics would be doing physics that no one could check.

At the substrate level, the picture is different. The substrate holds the global state, of which our branch is one sector among many. It does its own bookkeeping, and its conservation law is the Information Conservation Principle from the foundations series: information is neither created nor destroyed, and apparent loss is redistribution across sectors. The substrate manages branch identity, isolation, and the leakage channel that the boundary-permeability work describes.

The paper’s theorem is that these two descriptions are consistent: there exist states of a substrate such that every sector’s internal physics is unitary and Born-weighted while the global object also conserves information across sectors. The construction is the superselection structure from the walkthrough series, with the squeezed-state algebra doing the work of keeping sectors mutually invisible.

Why keeping the levels apart is hard

The temptation, every time, is to let one level reach into the other. Popular Everett talk does it constantly: the observer is told they are in a branch, which is information from the wrong level, and no experiment could deliver it. Copenhagen does it in the opposite direction, promoting the observer’s collapse into a physical process at the substrate level, which is a category error that has fuelled ninety years of confusion. The paper’s discipline is that neither level may borrow machinery from the other: the observer’s statistics must close without substrate facts, and the substrate’s conservation must hold without observer notions like measurement.

The hard technical content is in the interface. The paper proves a no-go worth stating in one line: no operation available inside a branch can transfer substrate-level information across the branch boundary, other than through the leakage channel, which is exponentially suppressed and patterned rather than thermal. That is what connects this paper to the testable programme: the interface theorem is what makes Paper 13’s transmission calculation and Paper 14’s lab protocols meaningful rather than decorative.

The Born rule, again

The observer level’s Born rule is not postulated twice. It is derived once, at the substrate level, from the conservation structure, and inherited: each sector’s effective dynamics is exactly the kind whose unique consistent weighting is squared amplitude. The observer inherits the rule the way a shadow inherits the geometry of the object casting it. If the ICP derivation fell, both levels would lose the rule and the architecture would need a replacement, which is why the derivation’s axioms are stated with more care than anything else in the paper.

What Paper 2 buys the rest of the series is a permission slip: it is now legitimate to talk about a substrate computing branches, and about observers inside them doing honest physics, without the two statements colliding. The next paper builds the first physical piece of that substrate, and it is the piece the whole boundary story grows from.