(Blaine from the Predator by Dunechaser on Flickr. "I don't got time to bleed.")
I'm re-reading Emergence
A passage just forced me to put down my (paperback) book, log in to Gogo inflight wireless from 27A and reflect. In one section, Johnson describes emergent software systems and the work of Danny Hillis.
One of the many things Hillis has done is write software that "evolved" sorting algorithms. At the time (maybe still?), world-record sorting algorithms took 60 steps to organize 100 numbers. Using an initial pool of code snippets and a process that coalesced full algorithms, Hillis created an evolutionary environment in which algorithms created themselves, each successive generation an improvement on the prior generation. The evolutionary process worked brilliantly until it eventually stalled at 75 steps. The evolutionary process stalled at "local maximums" because the system only rewarded generations that improved upon the prior generation. To get better would require taking a step backwards and certain death.
The solution? Introduce predators. Here's Johnson:
"Before the introduction of predators, a miniprogram that reached a seventy-five-step ridge knew that its offspring had a chance of surviving if it stayed put at that local maximum, but faced almost certain death if it descended to search out higher ground. But the predators changed all that. They hunted down ridge dwellers and forced them to improvise: if a miniprogram settled into a seventy-five-step range, it could be destroyed by predator programs. Once the predators appeared on the scene, it became more productive to descend to lower altitudes to search out a new peek than to stay put at a local maximum."
After 30 minutes or so, the new environment produced algorithms operating in the 62-step range.
Huh.
You have to be willing to eat your own young. Or at least nudge comfortable people down into the valley to learn a little, recombine and search out higher peaks.
Leaders of organizations really need to think about this one.
