Lesson 2 · 14 October 2026
Systems, feedback and emergence
Why do fish stocks collapse, even when no fisher wants them to?
| Session | Date | What |
|---|---|---|
| 6 | Thu 05 Nov | Quiz 1 |
| 7 | Thu 12 Nov | Project topics published, allocation opens |
| 9 | Thu 26 Nov | Project kick-off |
| 11, 12 | Thu 10 and 17 Dec | Presentations of preliminary work (P1) |
| 14 | Thu 14 Jan | Quiz 2 |
| 16, 17 | Thu 28 Jan, 08:30 and 10:00 | Final presentations (P2) |
Winter break between sessions 12 and 13.
| Part | Points |
|---|---|
| Quiz 1: basics | 15 |
| Quiz 2: analysis and LLMs | 14 |
| Group project | 40 |
| Individual final report | 15 |
| Participation | 16 |
| Grade | Points |
|---|---|
| 1 (Excellent) | 89 to 100 |
| 2 (Good) | 77 to 88 |
| 3 (Satisfactory) | 65 to 76 |
| 4 (Sufficient) | 51 to 64 |
| 5 (Fail) | 0 to 50 |
15 + 14 points
16 points
Took part? Come to me at the end of the session and I award you the points.
40 points, groups of 4 to 5
20 points
A GitHub repository with working, documented code. Research question and approach count.
10 points
10 to 15 minutes plus 5 minutes of discussion. Sessions 16 and 17.
10 points
5 to 10 pages: question, methods, results, discussion.
Topics come from a given list and are allocated online before the kick-off in session 9. You present your preliminary work in session 11 or 12 (P1).
15 points
AI tools encouraged
You are responsible for correctness and for understanding your code. In the final presentation, I ask oral questions about it.
Why do fish stocks collapse, even when no fisher wants them to?
We fish a lake together.
A system.
Feedback and emergence.
0 to 5 fish
In every round, you choose how many fish you catch.
× 2 until the lake is full
The fish left in your lake double. A full lake holds 6 fish per fisher: 24 in a lake of four.
8 rounds
Catch as many fish as you can over all of them.
Your lake starts full. You see its fish, not what the others chose.
The game runs in the live session. Every student fishes on their own device, and the fish in each lake appear here round by round.
One sentence: what did you think about when you chose your catch?
For example: what you expected the others to do, what you saw in the lake, what you wanted to achieve.
In class, the answers appear here, grouped into themes.
“A system is an interconnected set of elements that is coherently organized in a way that achieves something.” Meadows (2008)
“Put simply, a system is a set of complementary, interrelated parts that does something as a whole that its parts can't do on their own.” Sillitto (2021)
The parts you can identify.
You, your lake mates and the fish.
How the parts affect each other.
Catches take fish out, the rest regrow, and your lake mates fish the same stock.
What the system does, which can differ from what anyone wants.
Turned fish into catches, round by round.
You and your lake mates, the fish, the catch limit and the regrowth.
A regulator and a quota, the price of fish, other lakes, your need for income, the weather.
The boundary depends on the question. To ask “would a quota help?”, the regulator and the quota move inside.
“Value-laden choices of what to include, and where the boundaries need to be articulated, are always involved and have consequences for how we understand the system.” Sitas et al. (2022)
Our boundary holds people and fish together: a social-ecological system.
Knows exactly how fast the fish regrow.
Can't tell how much you will catch.
Knows that each of you wants a big catch.
Can't tell what that does to the lake.
Our curve came from both, interacting. Taking the system apart loses exactly what we want to explain.
What do the arrows between fish and fishers carry?
Think of a round after which your lake had fewer fish than before.
Fewer fish → I take less → more fish
“If I don't catch them, someone else will” → fewer fish → more reason to catch now → more catches → even fewer fish
Few fish → few new fish → even fewer fish
+ moves the same way, − the opposite way
R is a reinforcing loop, B a balancing loop
The rules fixed the catch limit and the regrowth, not which lakes survive.
Three tests for an emergent result
When you fished mattered, and how the others reacted, not only how much.
You each chose a number. A collapsed lake is not a number anyone chose.
Same rules, same kind of people, different lakes. Could anyone have predicted it?
More emergence is not always better
Railsback & Grimm (2019)
A model is not the system. It is a simplified stand-in, built for a question.
Ours: why do shared stocks collapse? Unlike the fishery, a model can be run again, changing one thing at a time.
Starfield et al. (1990)
| Our game | The simulation | |
|---|---|---|
| Fishers | You, 4 per lake | Boats on one lake |
| Space | No places | A grid of places |
| Catch | You chose 0 to 5 | A boat takes up to 5 from the richest water nearby |
| Fish | Doubled, up to 24 | Fish breed and spread to nearby water, up to a limit |
| Talking | From the break on | Nobody talks or plans |
We made these rules up. Nobody measured them.
Fish, the darker the more Boats The second slider is your choice from our game.
The model: each bird reacts to its neighbours
Fly the way your neighbours fly.
Move towards your neighbours.
Keep a minimum distance.
The system: real starlings
No bird knows the shape of the flock, and no rule mentions it.
Reynolds (1987)
Every cell is alive or dead. In every step, for every cell:
2 or 3 live neighbours
fewer than 2
more than 3
a dead cell with exactly 3
What these rules do
No choices, no chance: everything is fixed by the rules.
Four rules that can compute anything. In general, nobody can tell what a pattern will do without running it.
Fishers and fish form one social-ecological system.
Each catch changes the next: the fewer fish, the stronger the race.
The collapse emerges from many reasonable choices. Nobody chose it.
Knowing the rules is not knowing the outcome. So how could we find out what saves the lake?
Name one rule that could have saved our lake. Why would it work?
One or two sentences. Which feedback loop does your rule weaken?
In class, the answers appear here, grouped into themes.
Chapter 28, Agent-based modelling, in The Routledge Handbook of Research Methods for Social-Ecological Systems (Biggs et al., 2022), pp. 383 to 389. On Moodle.
About 30 minutes. Read it with our lake in mind.
A short reflection (about 150 words), before lesson 4:
Biggs, R. et al. (eds., 2022). The Routledge Handbook of Research Methods for Social-Ecological Systems.
Railsback, S. F. & Grimm, V. (2019). Agent-Based and Individual-Based Modeling: A Practical Introduction. 2nd ed.
Meadows, D. H. (2008). Thinking in Systems: A Primer.
Metcalf, G. S., Kijima, K. & Deguchi, H. (eds., 2021). Handbook of Systems Sciences.
Also: Mobus & Kalton (2015) · Jackson (2024) · Reynolds (1987) · Gardner (1970)