A proposal that many phenomena come from one source becomes a causal theory when it restricts how preparations, interventions and observable responses are connected. The transition can begin with a modest finite model whose assumptions and failures are visible.
Specify a calculation someone else can repeat
Begin with a state or history space, an allowed preparation, an update rule and a readout. Then identify an intervention: which part of the model changes when a setting is adjusted, and which parts are supposed to remain stable? Finally, state the units and the noise or probability model needed to compare the calculated readout with an observation.
These ingredients should share parameters wherever the proposed source is meant to explain a connection. If every output gets an unrelated adjustable rule, the phrase “one source” may add an interpretation without reducing the independent choices. A useful model makes that bookkeeping explicit enough for a reader to assess the claimed explanatory gain.
A small response calculation
For an original discrete example, let a retained variable x and a hidden variable h evolve as x at the next tick = ax + bh + u, and h at the next tick = rh + cx. The input u is a controlled drive. The measured output is x, and the starting state is x = h = 0. All quantities here are dimensionless.
Apply one unit of input at tick zero and no input afterwards. The first four measured values are x₁ = 1, x₂ = a, x₃ = a² + bc and x₄ = a³ + 2abc + bcr. For a = 0.4, b = 0.3, c = 0.2 and r = 0.5, they are 1, 0.4, 0.22 and 0.142.
A one-state alternative x at the next tick = ax + u matches the first two values but predicts 0.16 and 0.064 at the next two ticks. Those later observations distinguish these particular models if measurement uncertainty is small enough. A model comparison is stronger than the statement that either curve can be described using a hidden source.
Successful prediction need not identify the hidden ontology
Eliminating h gives x at the tick after next = (a+r)x at the next tick + (bc−ar)x + u at the next tick − ru. The scalar response depends on b and c through their product. Replacing b by sb and c by c/s, while rescaling h by 1/s for nonzero s, leaves the measured x dynamics unchanged when the preparation is transformed consistently.
Consequently, those output data alone cannot identify b and c separately. For the zero hidden preparation, b = 0.6 and c = 0.1 give the same response as 0.3 and 0.2. An independently calibrated hidden measurement or intervention could break that ambiguity. Merely collecting more of the same exact output sequence cannot.
The example demonstrates both predictive content and an identification limit. A hidden-variable model may explain a response economically while sharing that response with other realizations. The expanded book's tests with explicit alternatives makes a related distinction for continuous memory models.
Separate calibration from a demanding test
A practical study would specify its candidate models before inspecting the validation results. It would estimate parameters using one input sequence, propagate uncertainty, and predict the output for another drive or preparation. It would record where detector saturation, drift or unmodelled disturbances could imitate the predicted difference.
Testing a second drive matters because the model claims more than one fitted trace. Testing a different preparation probes whether the hidden-state assumptions transfer. Neither test automatically establishes a unique universal origin; each constrains a stated class of alternatives.
A useful publication should provide enough information to reproduce both successes and failures. Report the preparation, parameter values, update convention, readout, comparison model and error measure. Keep adjustable quantities distinct from values fixed independently.
Apply this standard to SMM/DTT
The expanded book's auditable model links this discipline to spatial readout, propagation, phase, memory and constants. Its reconstruction requirements separates the postulate from additional mathematics and physical identifications. These distinctions help readers engage with a concrete research programme instead of being asked to accept all consequences at once.
The published completion and phase-selection papers provide specific conditional structures. A next causal model can use those structures while identifying its extra dynamical assumptions and independent tests. The discrete two-variable system above is a teaching example, not a claimed derivation from SMM/DTT and not evidence for its physical ontology.
Sources and further reading
- Mohamed Haj Yousef, completion principles and real-commutant phase selection, published 2026 research records.
- Claims and evidence and what a common origin explains.