Efficient according to which criterion?
Calling a collective efficient appears informative until one asks what it is efficient at. Does success mean a rapid response, a low cost per response, broad coverage, or resilience after an interruption? These goals can conflict. A procedure that acts quickly by using every available participant may leave little capacity for the next event. A procedure that saves resources may respond slowly. An evaluation must identify the quantity being improved before declaring an improvement.
Optimisation supplies a precise language for this question. In the standard form “minimise f(x) subject to x belonging to a feasible set,” x describes the decision, f the objective, and the feasible set the permitted choices. Boyd and Vandenberghe’s textbook is a foundational resource for this mathematical framework. Assigning biological meaning to its terms requires additional evidence; the notation cannot choose that meaning on the researcher’s behalf.
An invented allocation example
Consider a deliberately synthetic system with six identical units of effort and two locations. Let x units protect the first location and 6 − x protect the second, with x an integer from 0 to 6. Suppose, purely for illustration, that the value of protection is the smaller of the two allocations. Maximising that value selects x = 3. The balanced result follows directly from the chosen objective and available budget.
Now change the objective to 2x + (6 − x), a score that values effort at the first location twice as much as effort at the second. The score simplifies to 6 + x, so maximising it selects x = 6. Nothing about the six units has changed. The selected allocation changes because the definition of success has changed. Neither answer is inherently the correct description of a colony, and neither score is an empirical measure of shrimp fitness.
Constraints carry part of the explanation
Suppose the system must now assign at least one unit to each location. The second objective selects x = 5. The constraint changes the answer again. In an engineered system, such a restriction might represent a service requirement. In a biological model, it might stand for a physical or behavioural limitation—but that interpretation needs justification. A constraint inserted to make an output look plausible is still an assumption, not a recovered fact.
This example also shows why comparisons need a common setting. Two methods should receive the same budget, access to information, and permitted actions unless the point of the experiment is to examine those differences. A rule using a complete view of the system has an informational advantage over a local rule. If that advantage is unreported, a performance comparison can confuse algorithm quality with access to information.
From a benchmark to a research claim
A proposed computational study should begin with a simple baseline and a declared evaluation set. For the invented allocation problem, exhaustive enumeration of the seven possible allocations gives a checkable optimum. More complicated problems may not offer that convenience. In such cases, an author should distinguish “better than the tested alternatives” from “optimal.” The latter is a mathematical claim requiring stronger support than a favourable set of demonstrations.
A study should also examine the distribution of outcomes and the cost of obtaining them. A method may achieve a high average score while occasionally failing badly. Another may perform consistently at a slightly lower level. Which is preferable depends on the application. Reporting both behaviour and costs allows readers to judge the trade-off. Reducing everything to an unexamined league table hides precisely the decision that matters.
Biological inspiration with a stated scope
Theoretical work on collective movement, including Couzin and colleagues’ 2005 model, shows the value of explicit local rules for investigating group outcomes. Such work can inspire new questions without establishing that another species uses the same mechanism. For eusocial shrimp, a claim that a colony optimises a particular quantity would need observable variables, a defensible comparison, and evidence that alternative explanations do not fit equally well.
AESRD’s proposed contribution is to make these requirements part of ordinary communication about collective systems. A useful paper would say what was measured, what was assumed, what alternatives were tested, and what “better” means. This note contributes an original teaching example, not a biological result. Its conclusion is deliberately limited: before evaluating the apparent wisdom of a collective, inspect the objective function assigned to it by the observer.