Software patterns from biology

Evolution has had a long time to solve coordination problems at the scale of molecules, cells, organisms and populations. I wanted to know which of those solutions are useful when designing software, so I built a catalogue.

The patterns

There are eight so far. A few examples:

• Bounded compartments, from cellular compartmentalisation: contain overload so billing keeps working while analytics rejects work.
• Bow-tie core, from metabolism: many inputs and outputs meet at a small shared core, which cuts down pairwise integrations.
• Integral feedback, from cellular signalling: keep a service at a utilisation target as demand changes.
• Persistence filter, from a common gene-regulation motif: ignore short spikes and act only on sustained signals.
• Quorum signalling, from bacteria: coordinate optional activity based on local participation.

Each entry has the biological basis with citations, a mapping table, an architecture diagram, implementation notes, failure modes, where the analogy breaks down, and an experiment to try.

Here is the bow-tie pattern, where many input formats meet a small shared core before fanning out to consumers:

And quorum signalling, where a decaying shared signal decides whether workers join in:

Keeping it honest

The catalogue separates three things: what has been observed in biology, explanations of how it evolved, and the engineering hypothesis. A good analogy is a reason to run a test, not proof that a design is right. Each pattern also says what would need to change for real deployment. For example, the quorum example does not implement consensus.

The examples use only the Python standard library and need no setup:

git clone https://github.com/timini/nature-inspired-architectures.git
cd nature-inspired-architectures
python3 -m natural_architectures all