CONFORMAL TRANSFORMATIONS THAT CLICKEPISODE 10 (FINALE) / Ten episodes folded into a single table

"Which one is real?" — ten episodes spent building a shape in which that question can be answered

Which frame is the real one? Episode 3 lined up three pictures, Episode 4 showed they are gauge,
and Episodes 8 and 9 showed that the quantum theory has breakings. So which is real?

What you need: the previous nine episodes, and nothing else Only dimensionless invariants can answer

"Is it expanding, is light slowing down, or are atoms shrinking?" — when Episode 3 set the three pictures side by side, everyone must have thought the same thing. So which is it, really? This series has kept answering "any of them, they are gauge". But is that a dodge? In the finale we set out the two positions that take this question seriously, and then give the series' answer. To say it first — only dimensionless invariants are entitled to answer. And that is not a dodge.

01Position A: "the question itself is misguided"

The position represented by Flanagan is unambiguous: the effort to decide which frame is correct is beside the point, at least within classical physics. Why? Because every observable quantity is a conformal-frame invariant.

This is exactly what the series has been demonstrating. In Episode 3 we computed the redshift in three pictures and got \(1+z=t_0/t_e\) every time. The dimensionless ratio at recombination was 52.6 in either picture. In Episode 4 we saw "distance between galaxies ÷ atomic radius" is gauge independent. In Episode 6, \(N=mc^2t/\hbar\) came out as literally the same expression in two pictures. If everything measurable agrees, no experiment exists that decides which is real.

Position A

Observables = conformal-frame invariants.
Therefore the question "which frame is physical?" has no content.

02Position B: "put matter in and you must choose"

Faraoni and others reply: once matter is included it is not so simple.

The reason is concrete. Under a conformal transformation, every quantity with mass dimension is transformed (Episode 3's table of weights). So a particle that was in free fall in the original frame — moving along a geodesic — does not follow a geodesic in the transformed frame. It feels a force from the dilaton's gradient.

What is happening

The original frame

Mass is constant. The particle "goes straight through curved spacetime" — a geodesic.

The transformed frame

Spacetime is flat. But mass varies from place to place as \(\tilde m=\phi m\), so the particle feels a force from the gradient of mass and bends.

The same motion, seen either as "straight, on a curved floor" or as "on a flat floor, under a force". The trajectory is the same either way, but the definition of "free fall" differs between frames.

Hence Faraoni and others say: if you want to speak plainly about "free fall" or the equivalence principle, you are forced to declare one of the frames to be the physical one.

03The two are not in conflict

This is the most important point of the episode. A and B are not actually quarrelling. The whole dispute is about what you call an observable.

A
"Observable" = the dimensionless ratios you actually measureClock ticks ÷ atomic period, angles to galaxies, shifts in colour. These agree in every frame → the question has no content.
B
"Observable" includes the structure of the theoryWhat counts as a geodesic, what counts as free fall, what counts as "no force". These differ between frames → you must choose.
Both are rightThe numbers that come out of the laboratory agree (A). The words used to explain those numbers do not (B). In the previous series' idiom — numbers are physics, wording is bookkeeping.
◇ ◇ ◇

04Quantum mechanically, nobody has settled it

So much for the classical story. Quantise and it genuinely becomes unclear. Episodes 8 and 9 have supplied the reasons.

Episode 8's anomaly. Quantisation required a standard of resolution \(\mu\). But swapping frames transforms that standard too. The calculations in two frames may differ by anomaly terms — two descriptions that were exactly identical classically may cease to be identical quantum mechanically.

Episode 9's ghost. In a path integral, "which field you choose as the integration variable" is itself the choice of frame. And because of the conformal factor's reversed sign, that choice affects whether the integral converges.

The state of the literature is, honestly, a mess — there are several papers concluding "equivalent even at the quantum level" and several reporting "it breaks". This is an open problem.

05This series' answer

Here is the answer we have spent ten episodes preparing.

The conclusion of "Conformal Transformations That Click"

Only dimensionless invariants are entitled to answer "which frame is real".
If one moves, it is physics; if it does not, it is bookkeeping.
And that verdict can always be delivered.

This is not a dodge, because it is a decision procedure. Whatever claim arrives, the same steps handle it. "The universe is not expanding, atoms are shrinking" (Episode 2)? Build the dimensionless ratio. "The singularity is a coordinate artefact" (Episode 6)? Build the dimensionless ratio. "\(\alpha\) is varying in time" (bonus episode 3 of the previous series)? Build the dimensionless ratio. The answer comes out mechanically, every time.

Figure: the last knob. Move the gauge and the four on the left change violently while the four on the right do not move at all (values at \(t=0.5t_0\), \(a=0.5\))
s = 0.00 (the standard picture) the four on the left move, the four on the right do not — only the right is physics
Bookkeeping (moves with the gauge) Physics (does not move with the gauge)

06Ten episodes, in one table

Ep.What we moved (bookkeeping)What remained (physics)
3The choice of time coordinate (\(t\leftrightarrow\eta\)) — this is what makes the speed of light change\(1+z=t_0/t_e\), the recombination ratio 52.6, \(\alpha\)
4The gauge \(s\) (splitting the expansion between metric and field)\(\tilde a\cdot\phi\) = distance between galaxies ÷ atomic radius
5The apparent shape of the potential (an effective potential)The equation of state \(w\); the triple boundary at \(w=-1/3\)
6Curvature \(R\) (it carries units; you can even set \(\tilde R=0\))\(N=mc^2t/\hbar\), the Planck time
7The marks on the ruler (1 of the metric's 10 components)Light cones = causal structure (the other 9)
8──The \(\beta\) function, \(1/137\to1/128\), 99% of the proton's mass
9──The conformal factor problem (no gauge can erase it)

Only the right-hand column says anything about the universe. The left-hand column is about how we choose to describe it. What we spent ten episodes doing was separating these two columns accurately.

Joining the previous series The backbone of "Cosmology That Clicks" was "quantities with units are bookkeeping; only dimensionless ones are physics". You could say this series has done nothing but promote that one sentence to a theorem — a conformal transformation is "the operation that moves quantities with units and leaves dimensionless ones alone", and bonus episode 9 ("a map of the constants you are allowed to fix") and bonus episode 5 ("you can fix at most three") were both restatements of that single operation. What the previous series grasped by intuition has been given a name and an equation. That is what these ten episodes are.

07Doors still open

Let us be honest and list what has not closed.

Are frames equivalent quantum mechanically?Episode 8's anomaly and Episode 9's ghost are entangled here. The literature is split.
The conformal factor problemUnresolved since 1978. There is no non-linear generalisation of the contour-rotation prescription.
Is conformal symmetry "real"?'t Hooft argues it is an exact, spontaneously broken symmetry. As Episode 4 showed, it can also be bolted on afterwards, so evidence is needed to tell the two apart.
What is \(c\cdot t=\text{const}\) actually claiming?As a way of speaking it is perfectly equivalent — so the phrase itself is not what goes on trial. What does is the expansion law \(a\propto t\) folded into the name. It goes to trial in the bonus episode.
Being straight with you

The arrangement of positions A and B is this series' reading. The two camps have not shaken hands over "the dispute is about the definition of observables" — the actual debate is more involved, and opinions still differ in particular about how to handle matter with non-minimal coupling, and about how quantum corrections act.

And the conclusion in section 05 is not universal. "Build a dimensionless invariant" is a decision procedure; it does not tell you which dimensionless quantity to build. We chose \(N=mc^2t/\hbar\) in Episode 6 because we already knew it had physical meaning. The procedure delivers answers mechanically, but framing a good question is, as always, human work.

Exercises for the finale
  1. Someone claims "the universe is not expanding; atoms are shrinking". How do you adjudicate?
    Show answer
    Build the dimensionless ratio. "Distance between galaxies ÷ atomic radius" is proportional to the same \(a\) in both pictures (Episode 4). Therefore they are indistinguishable = the same physics. The claim can be neither proved nor refuted; it is a choice of gauge.
  2. "After the conformal transformation the curvature was zero, so there was no singularity." How do you adjudicate?
    Show answer
    Curvature carries units and cannot be used to judge. Build the dimensionless ratio \(N=mc^2t/\hbar\) and it breaks as \(t\to0\) in both pictures (Episode 6). So the claim does not stand. Geometry vanished; the ratio remained.
  3. "\(\alpha\) is varying in time." Bookkeeping or physics?
    Show answer
    \(\alpha\) is dimensionless, so if it moves it is genuine physics. That is why an atomic clock can adjudicate (this is the critique of VSL in bonus episode 3 of the previous series). By contrast "\(c\) is changing" carries units and, on its own, is bookkeeping.
  4. (Harder) Why is Episode 9's conformal factor problem not solved by "just re-choose the gauge"?
    Show answer
    Because a path integral is the operation of summing over every value of the field, which is incompatible with choosing a particular gauge. Classically "\(\phi\) is gauge, so fix it" sufficed; quantum mechanically all values of \(\phi\) enter the sum. In other words this is a difficulty that no choice of frame removes — which is why Episode 9 is the one row in the table with an empty "bookkeeping" column.

SUMMARYSeparating the two columns accurately

"Which frame is real" has two positions. A (Flanagan and others): all observables are conformal-frame invariants, so the question has no content. B (Faraoni and others): with matter included, geodesics do not carry over, so one frame must be declared physical. They do not conflict — the dispute is only about what counts as an observable. The numbers coming out of the laboratory agree (A); the words used to explain them do not (B). At the quantum level, with the anomaly and the ghost entangled, nothing is settled.

This series' answer is a decision procedure: only dimensionless invariants are entitled to answer. If it moves, physics; if not, bookkeeping. What ten episodes did was separate those two columns accurately. The choice of time coordinate, the gauge \(s\), the curvature \(R\), the marks on the ruler — all left column. \(1+z\), \(N=mc^2t/\hbar\), light cones, the \(\beta\) function — all right column. And Episode 9's conformal factor problem alone cannot be moved to the left by any choice of gauge — there really are things that rewriting the books cannot erase.

This document is Episode 10 (the finale) of the series "Conformal Transformations That Click", written for physics-minded high-school and university students. The dispute over the physical equivalence of conformal frames is a real, unsettled issue; the position that observables are conformal-frame invariants at the classical level is represented by Flanagan (2004), and the position that with matter the geodesic equation does not carry between frames so that one must be declared physical is represented by Faraoni & Gunzig among others. The reading in section 03 that "the dispute is about the definition of observables" is this series' own and is not an agreed position of either camp. At the quantum level, results claiming equivalence and results claiming its breakdown coexist in the literature, and the matter is unsettled. The values in the tables (\(1+z=t_0/t_e\), the recombination ratio 52.6, \(\tilde a\cdot\phi\), \(N=mc^2t/\hbar\), \(\alpha^{-1}(M_Z)=127.95\), the roughly 1% of the proton mass carried by quark masses) were derived and checked in Episodes 3 to 8. The figure displays values at \(a=t/t_0=0.5\) under the gauge \(\tilde a=a^{1-s},\ \phi=a^{s}\). The academic standard is the ΛCDM model including inflation. — To print, use your browser's Print → Save as PDF (in the print version the slider is frozen and answers are hidden).

Print / PDF: ⌘+P (Ctrl+P on Windows). On screen the slider changes the gauge and shows that the four on the right do not move. "Show answer" opens the solutions.