Here is a cool article on Swarm Intelligence in National Geographic Magazine. In a nutshell, swarm intelligence is a process for making decisions that doesn't have a centralized control mechanism. The effect from these mass actions are called emergent; they arise out of the sum total of the interactions of the agents as a side effect of the individual agents independent actions. The emergent consequences are not the goal of those individual actions but rather the individuals collectively cause an unintended side effect.
It is extremely important for the system that these actions are taken without centralized control or even centralized information. Of course there must be communication, but it can't be centrally controlled. Ants, bees and other social insects use it to a large extent in their decision making process.
I covered swarm intelligence in markets here. A good example is the stock market. At any time, the collective intelligence of the world has processed all the given information and determined a price for the stock. The stock could be going up today so today's seller is wrong from your perspective, but the seller could be hedging or rebalancing their portfolio; each individual must make their own determination of the stock's value through the price. The problem that can occur is that the market may lose the swarm characteristics that ensure correct pricing and cause a bubble or other imbalance.
There is one very crucial part of the definition that I don't think gets enough notice. For the swarm magic to happen
the system has to reach a combined threshold of diversity, organization, and connectivity before emergent behavior appears.It the tension between connectivity, organization and independence of individual actors that causes the great benefits, but if that balance is thrown off, the market no longer prices correctly. Lots of things can disrupt the works. It's obvious that if the world is in a state of chaos there can be no market and that the ordering of society in general has to allow people to come together and buy and sell safely. Bad actors must be punished to protect the market from fraud. Connectivity can be face to face or over the internet. People have known this for ages but have long missed the hidden role diversity plays in supporting the whole structure. By diversity I mean independent individuals making decisions for their own reasons. During bubbles in particular this mechanism can break down.
Benefits of Swarm Intelligence
- great algorithm for decision making using many simple nodes. Simple nodes are easier to make and organizing those nodes is easier than creating more complicated nodes.
- on average extremely effective - makes reasonable decisions most of the time.
- If you are the individual whose risk pays off then you benefit greatly from the risk.
- Local Minima. If the coverage of the swarm is too small, it can be deceived by what looks like the correct answer locally, but is not the correct answer overall.
- Sucks for wrong individuals. If you are the individual who has to pay the price of failure then the system did not work for you. The system falls apart when nodes go along with the swarm instead of doing the hard work of determining their individual path.
There are many reasons why people might not pull their weight.
- Fear of taking responsibility - scared to take a risk and pay the price.
- Lazy. 'Nuff said.
- Deference to perceived authority
- Going along with the crowd
- The Bigger Idiot theory. The idea that someone else will come along and make the same mistake you did, except worse, and so save you from your own stupidity. So if, for example, you pay too much for a big house you and figure that can sell it to to the next sucker that walks down the street. Of course, only the biggest idiot actually believes in this theory.
Updated to add the Bigger Idiot theory.
No comments:
Post a Comment