Quick post this time. I answered this in an email today and figured, posting it may mean I repeat it less 🙂 So here it is:
PS Can you remind me of your ant analogy wrt your network?
Yes no problem ( I visited Deborah Gordon in the Gordon Institute at Stanford to discuss)
Deborah studies Harvester ants and did a TED
video about it (very interesting). She noted several good points
- The ants act independently
- The queen lays eggs and can have no impact
- The ant colonies are 150Million years old and are very sophisticated and complex
- There are 4 types of ant in the study (harvester, forager, cleaner and soldier), she painted them all to test this.
As ants leave the nest they rub against each other and detect patterns. So an ant may have been a forger yesterday and therefore goes to do that job today. Brushing past other ants they bump into each other and from their scent they can see the pattern. If there are many cleaner ants then the ant becomes a cleaner, too many harvesters, she (all ants are female except for the breeding season where some males are born and fly to new nest sites) becomes a harvester (no need to forage today) and so on …
People talk about AI as building larger smarter computers (IBM Watson etc.). My conjecture is that it’s not a large thing you build, it should be like ants. Lots of very small things with very few rules that are easily followed. Deborah did like this approach as I attempted to dispel the AI myth of a supercomputer. Each complex system should not do too much, like ants they have a limited function to do, too many functions is inordinately difficult to cope with. It’s the combination of complex systems that make up larger systems, not building Goliath systems, humans are not that smart (yet).
So in maidsafe
our network nodes are like ants, they will react to different messages and routes of messages and change their persona (harvester -> forager etc.). This builds a very large scale complex and sophisticated ability, in the same way nature does (cells in the body are the same, neurons in the brain etc.). We have about 7 main personas.
The variables involved to build a single complex system like Watson are in my opinion unnatural and not found in nature (it’s like creating God or something silly like that).
So complex colonies or cell groupings is in our ability and in maidsafe we have done this to create a very complex system that manages data. It does so using three guiding principles for the node.
1: Protect self
2: Protect data (food)
3: Protect network (colony)
Each of these rules govern what a node will do, so it will delete bad data, it will self terminate on getting a virus (others will terminate it as well). It will terminate itself to save some data. It will lose data to save the network.
It’s a bit of a mind shift for everyone, but in my wee world inside my head it makes perfect sense and it’s a replica of natural systems in action,