A calculation can reproduce a measured number without independently explaining it. The key question is where the information entered. Following that dependency is essential when evaluating claims about fundamental constants, fitted parameters or the predictive power of a new model.
A result hidden in the input
Consider the invented model y = Ng, where N is an integer and g is an adjustable scale. We set N = 10, take an observed value yobs, and choose g = yobs/10. Substituting gives:
The calculation is correct. Its agreement with the observation follows from how g was chosen. It would be circular to present that agreement as an independent determination of y. It would be entirely reasonable to present the procedure as a calibration of g.
This example is deliberately constructed to illustrate a dependency. It is not a characterization of a particular published theory. The distinction applies equally to established models and proposed alternatives: a fitted input should be named as a fitted input.
Calibration can support a new prediction
Suppose a model describes two quantities, y₁ = Ag and y₂ = Bg. The coefficients A and B are fixed independently of the two measured values, and the same parameter g genuinely applies to both processes. A measurement of y₁ can calibrate g and thereby predict y₂.
For a numerical example, take A = 2, B = 3 and a measured y₁ = 10 in specified units. The calibration gives g = 5 in the corresponding units, and the second equation predicts y₂ = 15. Measuring y₂ provides a new comparison, provided that B was not selected afterwards to make it agree.
The ratio makes the independent content especially clear. The scale cancels, leaving a relation that the two observations can test. This does not make a theory parameter-free; it identifies a prediction that survives the calibration. If the coefficients themselves depend on additional fitted quantities, those dependencies must also be disclosed.
Follow every dependency
A useful derivation can be accompanied by a short ledger. It records what enters the calculation and how each item was fixed. The ledger is more informative than a general claim that a result was “derived from first principles.”
| Item | How it is fixed in the example | Role |
|---|---|---|
| A = 2, B = 3 | Specified before using the observations | Model structure to justify |
| y₁ = 10 | Measured input | Calibration datum |
| g = 5 | Inferred from y₁/A | Fitted scale |
| y₂ = 15 | Calculated from Bg | Prediction for a separate comparison |
“Specified before” is necessary for this example, but it is not a complete defence against hidden fitting. A coefficient might have been chosen after inspecting similar data, or a whole model family might have been searched until one member matched the target. That search history affects how surprising the agreement is.
A clean evaluation freezes the model and its choices, states the allowed uncertainty, and tests consequences that were not used to choose those ingredients. Independent measurements and independently implemented calculations can strengthen the check. They address different possible sources of error and should not be treated as interchangeable.
A number still needs a physical identification
Suppose a mathematical construction yields a particular integer. That establishes a fact about the construction. Connecting the integer to an electromagnetic response, a mass ratio or a cosmological quantity requires a physical dictionary and an argument that the relevant observable is governed by that expression.
Dimensionful numbers require additional care. A length written as 1 metre becomes 100 centimetres without any change in the object. A bare numeral cannot carry the whole physical claim. Units, normalization conventions and the measured operational quantity must be specified. Dimensionless ratios remove dependence on the choice of units, but they still require a justified identification and an independent test.
The symmetry essay makes a parallel point: preserving a mathematical structure is one step; identifying its physical meaning is another. Both steps can be valuable without being conflated.
Reading the research programme constructively
The existing DTT constants explorer is a useful place to apply this method. Read each proposed route with a pencil: mark its structural assumptions, normalization choices, physical identifications and remaining matching conditions. Then ask which consequence could be checked without reusing the value being explained.
The same habit helps when reading the published completion-principles and phase-selection papers. Identify the starting definitions and the quantities being calculated before evaluating the proposed physical interpretation. A carefully delimited result can be a substantial contribution even when a larger programme remains unfinished.
Sources and further reading
- DTT constants explorer: the existing programme guide, including assumptions and open matching questions.
- Mohamed Haj Yousef, “Completion principles, real commutants, and positive-frequency geometry for orthogonal Hilbert representations,” Open Journal of Mathematical Sciences 10 (2026), 996–1012. Published paper.
- Mohamed Haj Yousef, “Real-commutant phase selection from real Hilbert representations,” Physics Letters A 594 (2026), 132085. Published paper.
- Claims and evidence: distinguish a definition, a derivation, a fit, a prediction and an interpretation.