A useful research question identifies an unresolved distinction, the evidence that could resolve it and the assumptions needed to interpret that evidence. Naming a large theory becomes more productive when these smaller questions are already clear.
Turn a theme into a question
“What is time?” can open a valuable discussion, but it leaves many different investigations possible. We might be asking how a clock is calibrated, why a process has a direction, how duration is experienced or whether temporal relations can arise from a deeper structure.
Choose one distinction before choosing an explanation. For example: does a delayed signal reveal a slower propagation speed, a processing delay or both? This question names alternatives that can be expressed in one measurement model. It also suggests what additional observation would help.
One observation can leave two unknowns
Suppose a signal crosses a known distance d at constant speed v and then experiences a fixed processing delay b. Its measured latency is L = d/v + b. Assume the clocks and distance have been calibrated independently and that the processing delay does not depend on distance.
A distance of 10 metres and latency of 3 seconds fit at least two possibilities: v = 5 metres per second with b = 1 second, or v = 10 metres per second with b = 2 seconds. Greater precision in that same single latency does not distinguish these exact alternatives.
The missing information is structural. One equation relates two unknown parameters. Before debating the nature of the signal, we can ask whether the proposed observation determines the quantity we wish to explain.
Design the separating comparison
Repeat the measurement over 20 metres while preserving the processing conditions. The first model predicts 5 seconds; the second predicts 4 seconds. Two distances now separate the alternatives. More generally, two measurements give:
v = (d₂ − d₁)/(L₂ − L₁), b = L₁ − d₁/v.
These expressions require distinct distances, a nonzero latency difference and the stated constant-speed, fixed-delay model. If the processing delay changes between trials, the inference is no longer justified by these data alone. The experiment must therefore control the condition that makes its comparison informative.
This is an original teaching calculation, not a proposed measurement of a fundamental constant. Its lesson is methodological: an effective question links an ambiguity to a feasible way of removing it.
State the kind of answer being sought
A mathematical question may ask whether a specified update preserves a quantity. A physical question may ask whether that update predicts a detector response. A historical question may ask what a passage meant in its linguistic setting. A philosophical question may ask what counts as an adequate explanation of dependence.
The answers use different kinds of evidence. A proof does not establish that its premises describe an instrument. A successful instrument model does not settle a textual attribution. A compelling interpretation does not replace an experimental comparison. Keeping the intended answer visible helps several disciplines contribute without talking past one another.
Questions for the monadic programme
The broad SMM/DTT aim is to connect inner generation with the structures available to outer observation. It can be divided into questions that admit explicit work: which histories are equivalent under future tests, which distinctions survive completion, which operations select a spatial comparison and which clocks share a common calibration?
The expanded book's reconstruction requirements and auditable-model programme provide a practical framework. Each proposed result should identify inputs, a calculation and an observable consequence. A small example with a definite failure condition may advance the programme more than a much broader claim whose assumptions cannot yet be inspected.
Write a short research note
A useful note can begin with five sentences: the question; the competing possibilities; the assumptions held fixed; the calculation or observation that separates them; and the result that would require revision. This structure works for a numerical experiment, a conceptual objection or a close reading.
Theory then has a clear job. It organizes the proposed relationships and tells us what should follow under changed conditions. Starting with questions does not diminish the value of a broad vision; it gives that vision a sequence of concrete tests and constructive tasks.
Sources and further reading
- Mohamed Haj Yousef, What a reconstruction must contain, Tests with explicit alternatives, and A model that can be audited.
- Continue with Matching claims with evidence and Time as a research question.