Google Tech Talk on the Dichotomy of Design
Many of the interesting challenges in computer science, nanotechnology, and synthetic biology entail the construction of complex systems. As these systems transcend human comprehension, will we continue to design them or will we increasingly evolve them? As we design for evolvability, the locus of learning shifts from the artifacts themselves to the process that created them. There is no mathematical shortcut for the decomposition of a neural network or genetic program, no way to "reverse evolve" with the ease that we can reverse engineer the artifacts of purposeful design. The beauty of compounding iterative algorithms (evolution, fractals, organic growth, art) derives from their irreducibility.
Google itself is a complex system that seeks to perpetually innovate. Leadership in complex organizations shifts from direction setting to a wisdom of crowds. The role of upper management is to tune the parameters of communication. Leaders can embrace a process that promotes innovation with emergent predictability more than they can hope to dictate the product of innovation itself.
Innovation is critical to economic growth, progress, and the fate of the planet, yet it seems so random. While innovation may appear inscrutable at the atomic level, patterns emerge in the aggregate nonetheless. A critical pattern, spanning centuries, is that the pace of innovation is perpetually accelerating, and it is exogenous to the economy. Rather, it is the combinatorial explosion of possible innovation-pairings that creates economic growth.
I arranged the talk to overlap with a SFI brain spa @ Google. Some quotes from that event (without attribution per Chatham House rule):
“Unlimited power limits intellectual parsimony.”
“With machine learning, we are creating electronic savants. They are happy in a high-dimensional space. They have no desire to reduce. What we want is electronic Keplers that can recognize the ellipse, not savants that can force fit a heliocentric model.”
“The target of evolution can’t be more complex than the selection pressure itself. If you can come up with the selection pressure, you might as well design it.”
“I don’t think there is any natural process that is incompressible. It’s not random.”
[I disagree with the premise of those last two quotes]