Path I · Physics Companion

What Do We Actually Observe?

What arrives at the detector is measured. What happened at the distant source is reconstructed.

Published essay
IntuitionThe ArgumentDeep NotesRead all

Each depth is written as a self-contained route. Choose one without needing to read the other two, or use Read all for a continuous article.

Intuition

We do not receive the event itself

A distant galaxy, supernova or other source does not arrive at the telescope. Electromagnetic radiation arrives.

source event ↓ emitted radiation ↓ propagation ↓ detector response ↓ processed data ↓ source reconstruction

The detector can measure received flux, observed frequency, direction, timing, polarisation and related local quantities.

It does not directly display the source’s age, distance, emitted luminosity or complete original waveform.

What arrived is measured. What happened at the source is reconstructed.

Reconstruction can still be excellent

A reconstruction can be precise and strongly constrained. The important point is not that inference is unreliable. It is that inference and direct reception are different stages of the experiment.

The Argument

1. The detector provides a local record

A spectrograph may provide a quantity such as

\[F_{obs}(\lambda_{obs},t_{obs})\]

together with calibration, noise and angular information. These are properties of the signal received here.

2. A source quantity enters through comparison

An observed feature is matched to a laboratory reference:

\[1+z=\frac{\lambda_{obs}}{\lambda_0}=\frac{f_0}{f_{obs}}.\]

The measured spectrum plus the identified reference give redshift.

3. Source history comes later

Distance, lookback time, intrinsic luminosity and physical scale are obtained only after the measured quantities are inserted into a physical model.

received signal → identification → measured parameters → physical model → source history

4. Propagation belongs inside the chain

If the signal can change during travel, that transformation belongs between source and detector. It cannot be removed from the inverse problem merely because one particular propagation model has already been adopted.

5. The conclusion

\[\boxed{\text{received data}\neq\text{reconstructed source history}.}\]

Path I begins here because every later claim about the distant universe inherits this distinction.

Deep Notes

Start with the physical experiment. A remote event changes an electromagnetic field. That field propagates to an instrument. The instrument responds locally and produces data. The source event itself never reaches the detector.

Astronomy then works backwards from that local record. The reconstruction may be exceptionally precise, but it still requires a chain of identifications and physical assumptions connecting what arrived here to what happened there. Deep Notes begins by making that chain explicit before introducing the mathematics used to represent it.

remote source state ↓ emitted electromagnetic signal ↓ propagation history ↓ local detector response ↓ processed observation

1. What is local to the detector

A telescope can record quantities such as received flux, observed wavelength or frequency, arrival time, direction and polarisation. A spectrograph may represent part of that record as

\[F_{obs}(\lambda_{obs},t_{obs}).\]

These quantities belong to the signal at reception. They are not direct readouts of the source’s distance, age, emitted luminosity or earlier waveform.

2. Even redshift already contains a comparison step

Suppose a received spectral feature is identified with a laboratory feature of wavelength \(\lambda_0\). The measured displacement is expressed as

\[1+z=\frac{\lambda_{obs}}{\lambda_0}=\frac{f_0}{f_{obs}}.\]

The detector supplies \(\lambda_{obs}\). The reference value \(\lambda_0\) and the line identification come from laboratory physics and a source model. Redshift is therefore tightly measured, but it is already a relation between received data and an identified reference.

3. The forward problem

Let the source state be \(S\), the propagation operator be \(\mathcal P\), the detector response be \(\mathcal D\), and the measured observation be \(O\). A compact forward description is

\[O=\mathcal D[\mathcal P(S)]+\varepsilon,\]

where \(\varepsilon\) represents noise and unmodelled error. If \(S\), \(\mathcal P\) and \(\mathcal D\) are specified, this equation predicts what should be received.

4. Astronomy usually solves the inverse problem

The actual observational task runs in the opposite direction: given \(O\), infer the source state \(S\). That inversion requires a model of the source, a model of propagation and a calibrated detector response.

Quantities such as distance, lookback time, intrinsic luminosity and physical size enter at this stage. They can be very precisely inferred without becoming direct detector readings.

5. Why propagation cannot be skipped

If the signal changes between emission and reception, those changes are part of \(\mathcal P\). A reconstruction that assumes a particular propagation law will attribute the remaining structure to the source according to that law.

If a different propagation law produced the same received data from a different source history, the detector alone would not distinguish them:

\[\mathcal D[\mathcal P_1(S_1)]\approx\mathcal D[\mathcal P_2(S_2)].\]

Independent observations can break such degeneracies. The logical distinction remains even when one model is overwhelmingly better supported.

6. The conclusion

\[\boxed{\text{what is received}\neq\text{the reconstructed history of what emitted it}.}\]

The purpose of this distinction is not to weaken measurement. It is to keep every physical step between source and conclusion visible.

Further reading

Open discussion

Questions, objections and alternative readings

This discussion is public and connected to GitHub Discussions. Specific objections, competing interpretations, relevant evidence and corrections are especially welcome.