Why do I keep going on about ants and nature
The author explains their ant analogy for the MaidSafe network, based on discussions with Deborah Gordon at Stanford’s Gordon Institute who studies Harvester ants.
Key Observations from Deborah Gordon’s Research
- Ants act independently
- The queen only lays eggs and has no control
- Ant colonies are 150 million years old and highly sophisticated
- Four types of ants: harvester, forager, cleaner, soldier
Ants detect patterns by rubbing against each other and change roles based on colony needs.
Author’s Conjecture on AI
Rather than building large supercomputers (like IBM Watson), AI should be like ants—many small things with few, simple rules. Complex systems should have limited functions; it’s the combination of systems that creates larger capabilities.
MaidSafe Network Design
Network nodes behave like ants, reacting to messages and changing their “persona” (similar to ant role-switching). The network has ~7 main personas.
Three Guiding Principles for Nodes
- Protect self
- Protect data (food)
- Protect network (colony) Nodes will delete bad data, self-terminate when infected, sacrifice themselves to save data, or lose data to save the network.