I read Digital Biology a few years ago and there's one topic I keep revisiting as I can't seem to reconcile all of it's elements--swarm intelligence.

Your criteria for effective swarm intelligence were something along the lines of

1. Randomness of events

2. Positive feedback

3. Negative feedback

4. Disproportionate fluctuation

The first three items make perfect sense to me, but the fourth doesn't seem to be absolutely necessary--can't a bee hive or ant colony survive without it? I understood your lottery example but wasn't able to translate it into something absolutely necessary for a colony of ants. If you could explain the need for disproportionate fluctuation in the context of an ant hill perhaps it would drive the point home.

I really enjoyed your book. I picked it up because I hoped to learn about naturally occurring types of organization in the hopes that I could apply them to business. It was one of those rare instances when the book covered exactly what I'd hoped it would cover. An unintended benefit was that your book has helped me to think through business problems on a more elemental level in order to better isolate the problem from the symptoms.

Thanks and regards,

I believe the fourth one was "amplification of fluctuations" - and it was thought of by an Entomologist. You need to amplify the fluctuation in order for the "choice" to be made by the system. I agree that these can be stated more concisely, however - which is what I tried to do elsewhere in the book. Glad you enjoyed it :)