Policy Trial

A comparison of two queue-ordering policies against the same declared dataset, ending in a defended choice between them.

Due: 9 April 2027 · Weight: 30%

Purpose#

Weeks 4 through 6 introduced first come first served, shortest job first and a priority rule, then compared them on one declared log in week 6. The Policy Trial asks for the same comparison done properly: a declared dataset, both policies run against it in full, and a written defence that owns the trade-off rather than hiding it.

The task#

Choose two ordering policies to compare — you may use two of first come first served, shortest job first, or a priority rule, or propose a variant of your own, as long as it is precisely specified. Declare a dataset (you may build on the Office of Everything logs from weeks 4 through 6, extend them, or construct a new one of comparable size) and compute the outcome — individual wait times for every entry — under each policy.

Then choose one policy and defend it in writing. The defence must name a specific group of requests that would wait longer under your chosen policy than under the alternative, not only describe who benefits.

What you submit#

The declared dataset, the computed results for both policies (shown as working), and a written defence of your choice, as one document.

You may use Normal Day from the Queue Policy Lab with two policies. Include the declared inputs, service order and individual waits, and explain at least one selection decision; a screenshot alone is not working. Other declared datasets and policy variants remain permitted, but require your own calculations because the Lab only runs its fixed inputs and rules. The written defence is your judgment, not a simulator recommendation.

Scope boundaries#

This assessment is a comparison of exactly two policies on one declared dataset — it does not ask you to build a working scheduler, simulate randomised arrivals, or survey real-world scheduling literature. Any dataset you construct yourself must be stated as constructed, not presented as observed.

Relationship to other work#

The Policy Trial reuses the Queue Autopsy’s discipline of separating description from assumption, applied here to a policy choice rather than an observation. Its declared dataset and defended trade-off are also the method The Honest Queue’s final redesign has to use when it accounts for who its own ordering rule favours.