Last updated: 2026-11-06
Lab 2: Feedback Loop Mapping
A deployed system doesn't sit still once it's live — it sits inside loops, feeding its own output back in as input, some of them stabilising and some of them amplifying. This lab asks you to find those loops in a platform you already use, rather than take the mechanism on trust.
Learning Objectives FoundationalKnowledge that endures for decades — core principles
- Identify the sensor, comparator, and actuator in a real feedback loop.
- Distinguish negative (stabilising) feedback from positive (amplifying) feedback in a software system you actually use.
- Predict, rather than just describe, what happens if a loop's gain changes.
Pre-Lab Reading FoundationalKnowledge that endures for decades — core principles
Read Feedback Control and Requisite Variety and the "Software Meets Reality" section of Software Meets Reality. Come with a platform in mind — a social feed, a recommendation system, a game's matchmaking, a CI/CD pipeline, a thermostat app — that you've used enough to describe its behaviour from the outside.
Materials
- Large paper or a shared whiteboard per group, for a loop diagram (sensor → comparator → actuator → system → sensor).
- Each group picks one platform; the tutor should hold back a reserve list (a video platform's "watch next" queue, a rate-limiter, a spam filter, a thermostat, an exam timetabling system) for groups that get stuck.
Session Plan (120 minutes) Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice
| Time | Block | Activity |
|---|---|---|
| 0:00–0:15 | Briefing | Recap sensor/comparator/actuator and Ashby's requisite variety; groups pick a platform. |
| 0:15–0:45 | Loop diagram | Draw the platform as a loop: what is measured (sensor), what it's compared against, and what action the comparison triggers. Label the loop positive or negative and justify the label from the diagram, not from a guess. |
| 0:45–0:55 | Break | — |
| 0:55–1:25 | Requisite variety check | List the range of situations the system actually has to handle (its variety of disturbances) versus the range of distinct responses its controller actually has. Where the two don't match, predict what happens to the disturbances the controller can't distinguish. |
| 1:25–1:45 | Gain experiment | Pick one parameter in your loop (how fast it reacts, how strongly, how often) and trace through, step by step, what changes if that gain is turned up. Does the loop stay stable, oscillate, or run away? Write the sequence of states, not just the conclusion. |
| 1:45–2:00 | Score the runaway case | If your loop is positive (or could become positive under the gain change you just traced), score the "runaway" outcome using the likelihood/impact matrix from Risk Management & Mitigation, then name a concrete mitigation for each of its four strategies — Avoid, Reduce, Transfer, Accept — even where one of the four turns out to be unworkable for this loop. Say which strategy the platform you chose actually seems to use today. |
Deliverable Applied / MethodologicalKnowledge with a 5–10 year half-life — stable practice
A labelled loop diagram, a one-paragraph requisite-variety verdict (where does this system's variety fall short of what it needs to handle?), the step-by-step gain trace, and the runaway-risk score with its four mitigation options from the final block.
Related Topics
- Software Meets Reality: Systems Thinking, Frame Analysis, and the Future Software Engineer — the argument this lab puts into practice.
- Feedback Control and Requisite Variety — the full control-theory background, including Beer's Viable System Model.
- Learning as a Feedback Loop (and the AI Partner) — the same loop applied to your own study process.
- Risk Management & Mitigation — the scoring matrix and Avoid/Reduce/Transfer/Accept framework this lab's final block applies to a runaway loop.
- Lab 1: Perspective Switching — previous in the series.
- Lab 3: Wicked Problem Framing — next in the series.
References
-
Wiener, N. (1948). Cybernetics: Or Control and Communication in the Animal and the Machine. MIT Press / John Wiley & Sons. Held by the University of Reading Library.
-
Ashby, W. R. (1956). An Introduction to Cybernetics. Chapman & Hall. Held by the University of Reading Library.
Graduate Attributes
Directly developed
- Subject Specialist: Readers apply requisite variety theory to assess where a chosen system's capacity falls short of its requirements.
- Critical Thinker: Readers trace the gain through a system's components to identify stabilising or amplifying feedback loops.
Modelled on this page
- Self-Directed Learner: Participants select their own familiar platform for analysis rather than working with a prescribed case study.
- Confident Communicator: Complex system dynamics are distilled into a single paragraph verdict and a labelled diagram.
University of Reading Graduate Attributes framework.