Week 3 Lab: Query completion audit

A lab auditing a real search box's autocomplete for suggestions its owner probably wouldn't choose to put there on purpose

11 March 2027

Before the lab#

No setup beyond a browser and a willingness to type slowly. You’ll be working with whatever search box you use most, so there’s nothing to install.

In the lab#

Everyone runs the same short list of neutral prefixes — starts of names, of places, of common questions — through their chosen search box, and records every suggestion offered. Comparing lists across the room does most of the work: the same prefix on the same search engine, typed by two people in the same lab, sometimes returns different suggestions because of personalisation you can’t see directly but can infer from the difference.

Then a narrower pass: prefixes chosen specifically to probe for the kind of suggestion a platform would rather not show — and a look at what’s absent rather than what’s present, since the interesting evidence here is usually a gap.

Afterwards#

This lab’s method — a fixed prefix list, run and compared across several systems — is close to the shape Assignment 2’s audit wants from you later in the semester, just aimed at a different kind of suggestion. Keep your prefix list; you’ll reuse the method before you reuse the topic.