The world is full of signals — but a physical observable is what survives the adjudication.
Every science whose work is watching the world runs into the same problem sooner or later: the world shares signals with us, and we need observables, and the road from the first to the second is often a costly one. A detector clicks. A strain gauge trembles by a part in a billion of a billion. A telescope fills a well with photons across a winter. None of these is yet a fact about the world. Each is a signal, raw and mute, and between it and anything I would be willing to write down as real lies a passage that has to be walked with some care, because a good deal of what could be written down at the near end of it turns out to be wrong.
The passage is a discipline, and to my eye it is a discipline of refusals. It tends to refuse to let a raw signal be read as a physical meaning before it has been calibrated and reconstructed and unfolded. It refuses to treat the residuals left over from a fit as harmless noise, because the residual is so often exactly where a new thing hides. It refuses to smooth away the edges — the acceptance of a detector, the mask over a survey, the finite window of an observation — because the edge is where the honest accounting tends to live. And it refuses, most of all, to mistake a formal agreement for an empirical fact, to let a curve that closes on the data be called a discovery. Four refusals, and every field I know that watches the world seems to keep them, in some form, or to pay for skipping them.
They are kept, in each field, by hand, in the field's own tongue. Particle physics will not call an excess a discovery until it has survived a long row of gates — the significance driven past five standard deviations, the systematics bounded, the look-elsewhere effect paid, the theory made to agree — and the Higgs boson was a bump in a plot for the better part of a year before anyone was permitted to say it was there. Gravitational-wave astronomy will not call a wiggle in the strain a source until it has held across the band of the instrument and been seen, in coincidence, by detectors a continent apart, the local track promoted to a sky position only once the network agrees. Cosmology will not call a tension in the parameters new physics until the survey's mask and the foregrounds and the instrument's own hand have been subtracted and the residual still stands. Different worlds, different vocabularies, and something close to a single discipline underneath: signal in, observable out, and a long list of things that have to be true before the passage is allowed to complete.
When I look closely I keep seeing the same moves recur under the different names. The physical observable that must not change when you boost to another frame is a kind of transport preservation — a reading carried without loss from one description to another. The gauge freedom identified and set aside so that only real deviations remain is a quotient taken with discipline, a sameness declared where sameness is real. The acceptance edge and the survey mask carried explicitly, never folded quietly into the interior, are the boundary kept honest. The five-sigma threshold and the blind analysis are a gate on maturity, a refusal to assert a result past the evidence standing under it. These feel to me like the grammar of turning a signal into an observable, and every watching field has had to learn some version of it.
What those fields keep by convention, we have tried to keep by proof. The discipline a collaboration enforces through review and training and the long memory of its own mistakes, we have tried to put into a kernel that will not sign a step it cannot check — the admissibility, the residual accounting, the boundary carried and then answered, the claim held back until its maturity gate closes. The passage from signal to observable seems to run the same way it runs everywhere; what is different in our hands is that each refusal is a theorem, and the whole of it runs on structure read off the measured boundary of a body. It is, in one line, a proof-carrying discipline for turning measured boundary signals into admissible, transport-preserved, claim-mature physical observables.
That is also the most honest answer I have to the question the rest of this work leaves open — whether what we built is bespoke to a single instrument or the first of a wider kind. I lean toward the second, and I can say why. The discipline our algebra enforces looks to me like the discipline of observation itself, kept by hand at the collider and the interferometer and the telescope, and we have written a version of it a machine can carry. I hold the claim exactly there, and no further. I have not proved that our algebra is the algebra of the collider; I have only noticed that the collider, the gravitational-wave network, and the cosmological survey each seem to keep, in their own hands, the discipline we keep in a proof — and that resemblance, when it shows up this plainly across such different instruments, is the strongest reason I have to think what we hold is a formalization of what it takes, more or less anywhere, to turn a signal into a fact.
A body, in the end, is only another instrument that the world shares signals through — a torch's light returned, a field's response across separated sites, a surface that answers when it is driven. What our algebra does with those signals is, I think, what scientific observation holds in common across every domain. It refuses to call anything a physical observable until it has honestly earned the title for itself. This isn't really new; if we look, we can see it everywhere in the scientific process, at least aspirationally. What we've tried to do is make that adjudication explicit, because when everything is checked and integrated and connected and proven, what we have isn't an oracle that will tell us the future — it is a witness to what is here now, and it will only ever say what it can prove. The instrument that reads a living boundary in just this way is the subject of a companion reflection, Reading the living boundary.
Companion reflections: What is genuinely new situates this object against the fields nearest to it; A boundary-observable certification algebra describes what kind of object it is and how it was built; The temperament of a bespoke algebra reads its character; and Reading the living boundary carries this discipline of observation into a measurement at the living boundary.