🌲 Opathorlokan University · ← back to all labs opathorlokanuniversity.net
CASE STUDY Agent-based ant colony — simple rules, complex behavior. Notebook → Nest → Dig → Emergence. For Kelly Thompson, age 16, Notebook 23, Volume 6.
Kelly's Colony Lab
v0.2 · Emergence · Agent-Based Sim
OPA · 4.3.4 · Building 3 · College III
Tab 01 of 04 · Notebook

Simple rules.
Complex behavior. Emergence.

A single ant is almost stupid. A colony of ten thousand ants — same brain, no boss, no plan, no blueprint — builds a structured city underground with brood chambers, food stores, waste dumps, and ventilation, all in the right places. This lab takes the lid off and lets you watch the city grow.

🐜
Notebook 23 · Volume 6 · Entry 412
"Dad says emergence is when simple rules make complex behavior. Mom sees it in markets. Jenny sees it in crowds. Dad sees it in lint. I see it in my colonies every single day. Same principle, different substrates. Colony Four's workers are doing the spiral thing again. I'm noting it."
— Kelly Thompson, 16
Dundee, Omaha  ·  Sunday

§1.1What's in here

Tab 02 · The Nest opens up a labeled cross-section of an ant nest — brood chamber, food storage, midden, foraging tunnels, surface mound — and compares Kelly's two species side by side. Tab 03 · The Dig is the simulation. Set your variables (species, moisture, soil, temperature, food, colony size), hit play, and watch them tunnel in real time. The chambers emerge from the workers' individual decisions — nobody plans them. Tab 04 · Emergence runs the same parameters twelve times. Twelve different nests. That's the Thompson family's whole thesis in one screen.

§1.2Easter eggs from the spiral notebooks

Omaha · Dundee neighborhood
"Eight is the perfect age to understand that observation changes reality."
Professor Matrix Thompson — MIT mathematician, chaos theorist, proprietor of Dr. Suds' Quantum Laundromat — handed his 8-year-old daughter a lint trap and asked her if she could see the fiber distribution pattern. Three weeks later she was watching ants carry a cricket across the patio for forty-five minutes. He bought her an ant farm that weekend. Twenty-three notebooks later, here we are.
// Thompson family methodology · est. 2017
UNO Allwine Hall · 3rd floor
Dr. Sarah Cane and the vibration sensor.
When Kelly was twelve, her dad introduced her to a complex systems researcher who flipped through ten minutes of her notebooks and told her she was conducting a longitudinal behavioral study. They put Colony Seven (Formica subsericea) on a vibration sensor — and discovered the digging patterns correlate with USGS seismic data at 60–70% confidence. Six hours of pre-detection ahead of the larger species.
// Cane Lab · UNO Complex Systems
Colony Four · December 2023
"They moved the brood chamber three centimeters deeper."
After a cold snap dropped surface temperatures, Kelly checked Colony Four's thermal gradient. The brood chamber had relocated three centimeters lower — toward the thermal optimum. No central planning. No queen issuing orders. The workers individually felt the temperature gradient and, following the simple rule "tend the brood where it's warmest," moved them. Same rule, observable result.
// Notebook 19 · Vol 4 · Dec 14, 2023
The spiral
Logarithmic food storage. Nobody told them.
In Formica colonies, food storage chambers often arrange in a slow spiral winding around the brood chamber. The spiral is not designed — it emerges from two simple worker rules: "drop food near other food" plus "never inside the brood chamber." The geometric solution to those constraints, played out by thousands of workers over weeks, is a logarithmic spiral. The same shape as galaxies, nautilus shells, and sunflower seeds.
// Field observation, Colony Four · 2024
The Sunday dinner
Four frameworks. One dataset.
Every Sunday, the Thompsons eat together. Kelly's raw observations (ants spiraling food storage) feed into Jenny's behavioral analysis (collective decision under constraint), then Mom's strategic framework (distributed optimization with no central authority), then Dad's mathematics (chaos theory and emergence). Four interpretive lenses on the same underlying physics. Kelly figured out at thirteen that dinner is her dad's research methodology.
// Professor M. Thompson, MIT '94, Dr. Suds' Laundromat '09
Why ants matter
100 million years of distributed intelligence.
Ants evolved their colony structure during the Cretaceous, when dinosaurs were still warm. They've been solving the same coordination problem ever since: how do you organize ten thousand individuals with no leader, no language, and no shared memory, into something that builds cities, farms crops, raises livestock, makes war, and survives every climate on Earth? The answer is the same answer to every other emergence problem. Simple rules, applied billions of times.
// Bourke & Franks, "Social Evolution in Ants" · 1995
⚠ Disclaimer & scope

This is a teaching lab built around an agent-based simulation. Each ant on screen is a tiny program following a few rules — move toward a tunnel boundary, prefer digging downward, deposit food near existing food, tend the brood at the thermal optimum. The chambers, spirals, and overall nest shape are not drawn by the code. They emerge from the rules. That is the entire point.

Numbers and species-specific behaviors are simplified from real myrmecology literature. Real ant nests are wildly more sophisticated — three-dimensional, with ventilation engineering, fungus gardens (in leafcutters), seed milling stations (in harvesters), and species-specific architectures we are only beginning to map with X-ray tomography. If this lab makes you want to learn more, Mark Moffett's "Adventures Among Ants" is where to start.

§1.3Notation

Through the lab: brood is the chamber containing eggs, larvae, and pupae — the most carefully tended and thermally optimal location. food storage is where foragers deposit returning food — often spiraling around the brood. midden is the trash dump, always kept far from the brood. queen sits at the center of the brood chamber. pheromone trails are invisible to humans but everything to the colony.

Tab 02 of 04 · The Nest

The Nest · anatomy of underground architecture

No two ant nests look identical, but they all have the same parts. Once you know what to look for, you can read a cross-section the way an archaeologist reads a dig site.

Select species

Worker length
Colony size
Max depth
Chambers

Cross-section · labeled

profile · cross-section

Parts of the nest

The mound & entrance. A pile of excavated soil ringing one or more openings. The entrance is the bottleneck — guards stand here, foragers traffic in and out, debris gets ejected. Lasius makes the classic crater-shaped anthill you see on sidewalks. Atta has a single massive entrance crater up to a meter wide.

Foraging tunnels. Roughly vertical near the surface, branching outward as they go deeper. Width is matched to body size — about 1.5 to 2 worker-widths. Wide enough for two-way traffic, narrow enough to defend.

Brood chamber. Where the eggs, larvae, and pupae are kept. The most carefully tended location in the colony. Workers actively move the brood vertically to chase the thermal optimum — colder surface temps push the brood deeper. The queen is here.

Food storage. Returning foragers deposit food where workers can access it later. In Formica species the food often arranges in a logarithmic spiral around the brood — an emergent geometric solution to "drop near food, but never inside the brood."

Midden / trash. Dead workers, undigestible remains, fungus mat in leafcutters. Always far from the brood, often near the entrance for ejection.

Queen chamber. Center of the brood chamber. In most species there's exactly one. The queen does nothing but lay eggs; she doesn't direct anything. The colony's "intelligence" lives nowhere — it lives in the interactions.

↗ Roadmap · v0.1 and beyond

v0.1 3D rotating cross-section · drag to orbit, slice plane to cut through.

v0.2 Real species-architecture data from Tschinkel's casting work (he pours plaster into nests and excavates the casts).

v0.3 Termite mound comparison — different evolution, similar emergent solutions to ventilation.

v1.0 X-ray tomography overlay for real-nest comparison.

Tab 03 of 04 · The Dig

The Dig · live agent-based simulation

Set the variables. Hit play. Watch the chambers emerge. Nobody draws them — they arise from the workers' individual decisions, applied a few thousand times.

The colony

Species Lasius neoniger
Colony size 120 workers
small=20 · typical=120 · large=300+
Food availability Moderate
scarce · low · moderate · abundant — drives forager fraction

The substrate

Soil moisture 45%
<20 dry · 30–50 ideal · >70 waterlogged
Soil hardness Loam
sand · loam · silt · clay — clay is hardest to dig
Surface temperature 72°F
drives thermal-optimum depth · 75°F is ideal · cold = deeper brood
Sim time
0days
Tunnel length
0cells
Max depth
0cells
Brood chamber
Food deposits
0
Active workers
0

§3.1The dig in progress

cross-section · live · west looking east
paused — hit play

What you're watching

Each yellow dot is one worker following a few rules: move toward an unexcavated boundary, prefer digging downward, carry dirt back to the surface mound, follow pheromone trails laid by other workers, never tunnel into a chamber already designated for brood. The pheromone field (faint green wash) reinforces successful dig routes — that's how main highways emerge.

The brood chamber emerges where workers cluster to tend the queen and her eggs — at the thermal optimum depth, which moves with the surface temperature slider. The food storage emerges from returning foragers depositing food near existing deposits while avoiding the brood. Watch the food chambers in Formica colonies — they tend to wind around the brood in a slow spiral.

↗ Roadmap · v0.1 and beyond

v0.1 Click to drop food crumbs · watch foragers find and recruit.

v0.2 Click to drop a predator stimulus · watch defensive behavior emerge.

v0.3 Seasonal cycle · winter cluster deep, spring expansion.

v0.4 Vibration sensor (Kelly's discovery) · seismic input shifts digging direction toward source.

v1.0 3D nest with rotating cross-section · same simulation, third dimension visible.

Tab 04 of 04 · Emergence

Emergence · twelve colonies, same rules

Set the parameters, hit run, and the simulation builds twelve colonies in parallel with identical inputs. The rules are the same. The starting conditions are the same. The outcomes are not. That is what emergence looks like.

Shared inputs

Species Lasius neoniger
Colony size 100
Soil moisture 45%
Surface temperature 72°F
Run length 14 days

Run the experiment

Each thumbnail below shows an independent colony with identical inputs but different random seeds. Look for: where the brood chamber settled, where food storage emerged, whether tunnels branched east or west, how deep they pushed. Same rules, no two outcomes alike.

Mean depth
cells
Depth σ
Mean tunnel
Brood depth range

§4.1The twelve colonies

What you're seeing

This is the Thompson family's whole methodology in one screen. Same species. Same colony size. Same soil. Same temperature. The rules each ant follows are identical in every panel. And yet — twelve different nests. Different depths. Different branching patterns. Different chamber placements.

This is also why ant colonies are so robust. There is no single right answer that the colony has to find. There are many valid solutions to "build a viable nest in these conditions," and the simple rules find one of them every time, robustly, without ever needing to know which one. The colony doesn't optimize toward a blueprint — it satisfices toward viability.

The same logic explains why Mom's markets, Jenny's crowds, Dad's lint, and Kelly's ants all look like the same kind of system. None of them have a designer. All of them are stable. All of them are slightly different every time. Emergence.

↗ Roadmap · v0.1 and beyond

v0.1 Click any thumbnail to view its full-resolution time-lapse.

v0.2 Side-by-side comparison panel · pick any two outcomes to study differences.

v0.3 Statistical overlay · all 12 nests rendered transparently on top of each other to show the consensus shape.

v0.4 Parameter sweep · vary one input across the 12 runs to see sensitivity.

v1.0 Cross-system emergence — same engine, different agents — for Jenny's crowds, Dad's lint fibers, Mom's markets. Watch the same statistics rules generate stadium ovations, lint patterns, and market crashes.