Does better food win?

27 of 27 terms

Collective intelligence

Colony-scale behavior emerges from individuals with limited information and local interactions. There is no central controller directing bees in this model.

You can see sources A and B at once. A bee uses only its own encounters, memories and nearby dances.

Dance recruitment

After a successful return, a bee may dance. A nearby bee can observe it and probabilistically follow the reported source; information does not reach the entire colony at once.

When disabled, new dances and dance encounters stop. Previously learned sources may remain in memory; uninformed foragers wait.

Scout proportion

The initial share of bees able to search independently. Scouts also collect food and can follow dances. Roles do not change during a run in this version.

18% of 160 bees rounds to 29 scouts. Changing the proportion starts a fresh run with the same seed.

Communication noise

The size of bounded errors added to the direction and distance learned from a dance. It also affects acceptance through the dance’s confidence value.

0% removes added location error; it does not guarantee acceptance. 100% does not mean that 100% of messages are lost.

Seed · random seed

An integer that initializes the bees’ random choice sequences. The same seed, initial settings and interventions at the same ticks reproduce a run in the same model version and runtime.

You can repeat a run with 518394. Trying other seeds explores how much the result depends on a single random starting point.

Tick and model time

A tick is one simulation step. 1 tick = 0.1 model seconds; 600 ticks = 1 model minute. The clock displays progress within the model.

Step advances exactly 1 tick. 20× playback computes more steps; it does not change bee rules or the duration of a tick.

Food collected

The cumulative sum of delivered cargo × quality at collection, counted since the start of the run. It is a model reward unit, not grams or actual honey yield.

Delivering 1 unit of cargo from a source with quality 0.95 adds 0.95 to the total. Cargo still in flight has not been counted yet.

Food intake rate

The delivery rate over the trailing 60 model seconds, expressed per model minute. During the first 60 seconds, the denominator uses only the elapsed time.

The total can rise while this rate falls: fewer deliveries arrived in the latest minute. The chart shows about 600 model seconds of history; each point still uses a 60-second window.

First discovery

The model time when any bee first detects an active, stocked source within its local sensing range.

“—” means no discovery yet. The first source found need not be the richest source or the first food delivered.

Active scouts

The number of bees with the scout role that are not resting, dancing or observing a dance.

Scouts collecting at a source or returning to the hive also count. This is not just the number searching for new sources.

Recruited foragers

The current number of bees outside hive activities whose last recruitment flag remains set. Scouts that followed a dance can also be included.

This is not a cumulative counter of recruitment events. The same bee may accept several dances during a run.

Resource allocation

The share assigned to A or B among bees flying to a source, collecting or returning with a source target. Searching and hive activities are excluded from the denominator.

With 3 bees directed to A and 7 to B, B shows 70%. This does not mean 70% of the whole colony is at B. With no source-directed trips, both bars show 0%.

Source quality

A model coefficient multiplying the reward value of collected cargo. It differs from remaining food stock; bees learn quality through a local visit or a dance report.

Default source A has quality 0.46; B has 0.95. B is farther away; higher quality alone does not guarantee the outcome.

Private knowledge

A source location and quality estimate learned through a bee’s own visit or an accepted dance. Every record has an expiration tick.

Removing source B does not instantly erase memories. A failed visit can invalidate a bee’s record, or it can expire; the default lifetime is 140 model seconds.

Energy proxy

An abstract indicator from 0 to 100 that falls during flight and rises at rest. It does not constrain flight or decisions in this version.

20% is not a measured biological energy reserve. It cannot establish starvation, lifespan or metabolism.

Dance direction

The direction from the hive to the advertised source, shown in degrees clockwise from east. A sun compass is not modeled.

0° points right and 90° points down. Communication noise may be added to the direction learned by a follower.

Model distance

The straight-line distance between the hive and the remembered source. “u” is an abstract distance unit, not a conversion to meters.

At equal quality, a farther source receives a larger travel discount in the dance utility calculation.

Duration code

An educational time code calculated as advertised distance ÷ 250. It is not the length of time the dance remains available.

A distance of 250 u gives a 1-model-second code. This conversion is not calibrated from real waggle dances.

Utility

The bee’s own reward discounted by round-trip distance: reward ÷ (1 + 0.003 × distance). It affects the probability of dancing and the dance’s lifetime.

High quality raises utility while longer distance lowers it. It is neither net energy nor a source score shared by the colony.

Signal time remaining

The model time left before a dance report expires. An expired dance cannot recruit new followers.

An old dance can remain briefly after a source is removed. Duration code encodes distance; time remaining describes how much longer the report is available.

Control colony

A comparison colony with the same seed, initial environment, population and tick count, with dance recruitment disabled. Source and noise interventions apply to both colonies.

Results appear as experimental / control. Behavioral differences can change how random choices are consumed; one paired run does not estimate a general biological effect.

Intervention

A change during a run: toggling dance, changing noise or removing and restoring source B. It is recorded with the number of completed ticks.

An intervention at tick 100 applies before step 101 is computed. Changing seed, scout proportion or population instead starts a fresh run.

Run and replay

A run is an experiment with specific initial settings. A JSON export stores settings, versions, tick count and ordered interventions.

Import does not trust saved metric values: it recomputes from the start and pauses at the recorded tick. Unsupported or invalid files are rejected.

Landscape view

An observer map of the hive, sources and bee positions in model space. Bee decision rules cannot access this map.

Seeing A and B from the start does not mean the bees know them. Select a bee to inspect which records it actually holds.

Dance floor

A magnified view focused on dancers and observers in the hive. Each dance is an expiring signal that only nearby bees can encounter.

Inspect a dancing bee: direction, distance, duration code and utility are generated from its own experience.

Communication view

Links show recently accepted dance reports between bees. Bees on the outer ring are a schematic projection of those outside the hive.

Links extend back at most 300 ticks (30 model seconds); the journal limit can shorten this in busy runs. This is not the complete communication history.

Signals and performance

An observer layer showing signal details and application performance. FPS is frames rendered per second; ms is real computation time.

Paired tick timing includes both experimental and control calculations. FPS and computation time do not measure model time or colony success.