Chapter 3 left us with a question that has been waiting since the start. If nothing is ever sealed — if every thing is a flow open to every other, leaking and pulling and exchanging without stop — then is anything ever truly cut loose? Is anything ever random?

This is not a small question. It sits under everything. Whether the world is, at bottom, lawful or chancy decides what kind of place we are living in. And the answer most people carry is that the world is chancy at the bottom — that physics found, in the last century, real randomness woven into the floor of things. Quantum events, we are told, are not just unpredictable to us. They are undetermined in themselves. The universe rolls dice.

I want to look at that claim slowly, because I think the axiom of this book puts a crack in it. Not a victory over it. A crack — and an honest account of how wide the crack is, and what it would take to close it either way.

First, let us be clear about what randomness would have to mean.

There is a weak kind of randomness, and nobody argues about it. A coin flip is “random” in the weak sense — not because the coin escapes the laws of motion, but because we cannot track the thumb, the air, the spin, the table, finely enough to call it. The coin is fully determined. The flip obeys mechanics exactly. It is only random to us, because our knowledge runs out before the causes do. This is randomness as ignorance, and it threatens nothing. The causes are all there; we just cannot see them.

The strong kind is different, and it is the one that matters here. Ontic indeterminacy says that the prior physical state does not fix one unique outcome, even when the probability law and all admissible conditions are specified. It need not say the event is severed from every influence, produced from nothing, or governed by no law. That distinction matters: indeterminate is not the same as isolated, and “not uniquely fixed” is not the same as “without conditions.”

It is this strong randomness the axiom has trouble with. And the trouble is simple to state.

An event wholly severed from every condition would indeed be a closed island in time. The axiom presses against that image. But ontic indeterminacy does not require such an island. An event may be embedded in an open system, constrained by its state and environment, governed by a probability law, and still not have one outcome fixed in advance. Openness and indeterminacy are logically compatible.

That admission removes the shortcut from openness to determinism. What remains is a philosophical preference: I find a fully relational world easier to read as determined through relations than as containing an irreducible choice among allowed outcomes. A critic may find the opposite. Neither preference is a deduction from “no severance.” The event that looks random may have causes running into a whole flow we cannot trace; it may also be genuinely indeterminate while remaining fully coupled. Determinism is therefore a lean, not a result of the axiom, and the chapter will keep it in that lane.

And I want to be careful that this “determined” is not heard as the old clockwork — the world as a film already shot, every frame waiting in a sealed canister to be played. That picture is itself a closure, a box drawn around the future, and the axiom forbids it as surely as it forbids the uncaused spark. The determination I mean is stranger and lighter than that. The next event is fixed by the whole connected flow, yes — but it is not written anywhere in advance, not sitting finished in some register the present is merely reading off. It has no existence until the flow arrives at it; the only way to find what it is, is for the current to run forward and resolve it. Fixed, and yet unwritten — determined the way the next figure of an unending calculation is determined, certain to come out one way, but not stored ahead of the working, reachable only by doing the work. So the world is not a sealed future being unrolled. It is a determination with no closed box anywhere in it — caused by everything, written nowhere, and made only at the moving edge where the flow comes to its point.

Here the argument runs out, and I will not push past it.

I have not proven the world is determined. I have said why a fully relational picture inclines me that way. The axiom is compatible with both hidden determination and lawful indeterminacy, so the lean belongs to the philosophy, not to logic. Here is exactly where the argument runs out.

The person who believes in strong randomness can say: you cannot show me the hidden causes. You are telling a story in which every quantum event is secretly piped by the whole universe — but you cannot trace one thread of that piping, not one, and a cause you can never in principle detect is not so different from no cause at all. Your determinism is as invisible as my randomness. Neither of us can show the other the goods.

And he is right that we may be stuck more deeply than present precision. Some deterministic and indeterministic interpretations reproduce the same quantum predictions by construction; no merely finer instrument can distinguish empirically equivalent accounts. A test becomes possible only when a specific deterministic completion predicts an observable departure from standard quantum theory. Until such a model names that departure, the disagreement is metaphysical rather than an unfinished measurement.

That is the honest position. The axiom does not settle determinism, and indeterminacy does not require severance. Anyone who says a generic increase in resolution must decide between them is reaching past both logic and evidence.

And yet the tie is not perfectly even, and there is one thread of evidence worth weighing, lightly, before we move on — because it tilts the table, even if it does not flip it. It comes from the people who need real randomness and cannot make it. A computer, however vast, cannot generate a truly random number. It can only compute — follow rules — and rules produce sequences that look random but are not, that a sharp enough analysis unwinds back to their seed. This is not a small engineering nuisance; it is a known limit. So when randomness is genuinely needed — for the keys that guard secrets, where a predictable "random" number is a broken lock — it cannot be calculated. It has to be harvested. The machines reach outside their own computation and sample the physical world: thermal noise in a resistor, the timing of radioactive decay, the jitter of electrons. They go to the thermodynamic floor and scoop up what looks, to them, like chaos.

Now weigh what that means carefully. Deterministic algorithms can generate sequences unpredictable to bounded observers, but their algorithmic output is fixed by program and seed. When applications require fresh physical entropy, machines sample thermal noise, decay events, timing jitter, or other processes. This shows a practical distinction between computed pseudo-randomness and sampled physical entropy. It does not show whether the sampled process is ontically indeterminate. The lesson belongs to openness only as illustration: what the formal system did not specify enters through an interface to a physical environment.

So the disagreement remains, but it is not weightless for the philosophy. The axiom inclines me toward determinism; the behavior of formal machines illustrates how rule-bound systems produce pseudo-randomness and must sample physical processes for fresh entropy. That is not evidence that the sampled processes are deterministic. It is an image of openness, and no more.

But a metaphysical preference can generate a scientific wager if it is given a model. A deterministic completion must specify hidden state, dynamics, accessible observables, and a prediction that differs from standard quantum theory. Then the test is finite: preregister the observable, compare the models, and let the result constrain or kill that completion. Finding unexplained correlation would not by itself vindicate determinism; ordinary dynamics, detector memory, selection effects, and standard non-Markovian physics must be excluded. Failing to find correlation would kill the registered model within its tested range, not prove that no deterministic account can ever exist.

So I do not possess one experiment that decides determinism in general. I possess a demand placed on every physical version of the lean: become specific enough to lose. If a completion permits signaling, violates operational equivalence, or misses its held-out prediction, it falls. Ontic indeterminacy would cost the philosophy its deterministic extension while leaving openness intact. Actual severance would strike deeper. The sizes remain distinct, but the weapon is now honest: a finite test of a named model, not an instrument instructed to “look as deep as looking goes.”

I should say where this argument has company, because I am not the first to make it.

A serious physicist, David Bohm, spent much of his life building exactly this kind of picture — a reading of quantum mechanics in which the apparent randomness is incompleteness, in which hidden variables, woven into an undivided whole, determine what looks like chance. His view is a minority one and contested, and I do not lean on his authority. I name him because a first-rate physicist looked at the same dice and concluded, on grounds of his own, that the world beneath them is whole and connected and very likely determined. The idea that quantum randomness might be the surface of a deeper, undivided order is not a thing I invented to save my axiom. It was already there, in the physics, argued by people who knew the mathematics far better than I do.

Bell's theorem and the experiments behind it matter here, but only at their actual scope. They rule out broad classes of local hidden-variable explanations under explicit assumptions. They do not prove determinism, permit faster-than-light signaling, or show that every thing in the universe is dynamically coupled to every other in the book's sense. They establish quantum correlations incompatible with a local-causal factorization of the tested kind.

That result is congenial to a philosophy suspicious of simple separability, but congeniality is not confirmation of the whole axiom. It gives one precisely defined arena in which classical local explanation fails. Bohmian mechanics remains logically possible because it is nonlocal, though it brings its own commitments and does not follow from Bell. The disciplined conclusion is enough: Bell constrains how a deterministic completion could work, and it warns us not to treat local separability as an unquestioned default. “The axiom, measured” would claim far more than the experiment supplies.

The formal discipline for all this is in the back of the book. It does not supply one universal determinism prediction. It requires any proposed deterministic completion to state a distinct, finite, losable test. Here, in the philosophy, the claim is only a preference for determination within relation — a preference the axiom permits but does not force.

So where does that leave chance?

Right where the axiom puts it — real the way the horizon is real. The horizon is a true feature of how the world looks from where you stand. Walk toward it and it moves; it is not a wall out there at the edge of things. It is the place your seeing runs out. Randomness, I think, is like that. It is a true feature of how the world looks from inside our limited reach — the place our tracing of causes runs out. Real, in that sense. Reliable. You can build on it, calculate with it, trust it to behave. But not a wall at the edge of the world. Not a true cut in the flow. The place our knowledge ends, mistaken, understandably, for the place the causes end.

And I hold even that with an open hand. Maybe the horizon is not merely epistemic; maybe outcomes are genuinely indeterminate while remaining fully coupled. That would cost the deterministic lean and not the open-system axiom. The lean earns scientific standing only through particular completions that name their own finite tests. Until then it remains metaphysics, held because I find it coherent, not because an instrument is guaranteed to settle it someday.

One question is left, and it is the one a reader feels in the chest before the head. If the flow determines everything — if every event is piped by the whole connected current, fixed though no one can trace it — then what becomes of choosing? Am I a puppet of the flow, my sense of deciding only a story I tell after the current has already moved my hand?

I do not think so, and the reason is in the shape of the determination, not against it. The flow does not push you around from outside, the way a hand works a puppet. You are not outside it to be pushed. You are one of the places it resolves — a knot where the whole current gathers and becomes one definite thing rather than another. The deciding is not something done to you by the flow. The deciding is the flow, narrowing to a point, at you.

Watch where it happens. A car drifts into your lane, and in the half-second you have, either you see it and pull the wheel, or your attention is elsewhere — on the phone, on a worry, on nothing — and the moment closes over you. The difference between the dodge and the crash is not muscle. It is where your attention was resting when the fork arrived. Attention sits at the edge this book keeps circling — the live edge between holding a shape and losing it — and where it falls is what tips the event one way rather than the other. That fall of attention is what choosing is. Not a faculty bolted onto the machine. The live settling of a fork the whole flow has loaded but not yet landed.

Here is the place to be careful, because it is easy to hear more in it than is there. The fork is loaded — heavily. Everything pours in: what you are, what was done to you, the state of your body in that second, the long training that shaped where your attention even tends to go. The choosing is not free of cause; a choice cut loose from all of that would be the uncaused event this chapter already set aside, the severed island that cannot be. What is live is not the absence of cause. It is that the resolving happens here, now, at the fork, and not somewhere before it — the current arriving at a point and becoming definite there, in the present, which is the only place anything ever happens. Conditioned to the root, and still settled live. Both at once.

So is the choice determined, or is it free? Here the chapter’s own answer comes back, and it is the tie. From inside the moment, “the whole flow fixed this” and “I settled it, live, just now” are not two findings you could ever tell apart. They are one event, read from two distances. The one who could stand outside and see whether the fork was sealed before you reached it — that one does not exist, for the same reason no one can trace the hidden causes. So the choice is as determined as anything in this chapter and as free as anything ever is, and the two are not rivals. They are the inside and the outside of a single resolving, and you only ever get the inside.

And there is a will at every scale, and they are not all yours to place. The body has its pull, older and deeper than your attention. The long currents that carry a life — the language you think in, the conditioning you did not choose and cannot fully see — move you in ways you do not author, more a will you undergo than one you exercise. And the largest current of all, the whole turning itself — whether that, in any sense, wills, the way you do — is the question this book has refused from the start, and refuses here too. This much is enough, and honest: you choose, at your own scale, live, at the fork; and you are moved, at the scales above you, by currents you did not set going and cannot fully see. From inside, you cannot always sort which is which. That not-sorting is not a flaw in the account. It is the same open hand the whole chapter has been holding out.

We said the box was never there in space. We have now followed it into time, and found that an uncaused event would be a box in time — and that the same axiom forbids it, though it cannot yet prove the forbidding. Next we turn from cause to time itself: not whether events are determined, but how the past, the present, and the future are made — and why only one of the three is ever the place where anything actually happens.