I Believe But Cannot Prove: Speculations on Evolutionary Development and
the Cosmic Purpose of Accelerating Change, © 2005 by John
this 1,000 word piece as a personal response to the 2005 World
Question at Edge.org:
"What Do You Believe is True Even Though You Cannot Prove It?"
Belief plays an important role in all our lives. This question brought
out a number of the beliefs underlying my own intution with regard to
Like the living systems within it, I believe our universe utilizes two unique processes of change: evolution and development. I also believe our universe follows an evolutionary developmental plan to do some generalized form of computing, and that its emergent intelligences are each evolutionary experiments to this end. Though imperfect, I think universe-as-living-system and universe-as-computer are among the best analogies available today.
In life, evolutionary and developmental processes share infrastructure and information, yet have independent, complementary mechanisms and unique dynamics. Evolutionary processes are nonlinear, chaotic and poorly predictable, while developmental ones are convergent, cyclic, and can be quite predictable, if you know where to look. Both exhibit heritability and variation. Scholars use the phrase "evolutionary development" (or "evo-devo") for this yin-yang model of change in biological systems.
Astrophysics predicts much about the development of simple, non-generative physical structure in the universe (e.g., entropy, black holes, and the accelerating universe). But today's physics is largely silent on the future of complex evolutionary developmental structure, such as life and intelligence. There are a number of "cosmic questions" one might ask with regard to the latter subject. Let me share four of these questions here, and my tentative responses to them.
First, why does evolutionary development of novel complex physical structure, from the slow emergence of galaxies to the rapid advent of today's computers, continuously accelerate over the last half of our universe's lifespan? Carl Sagan was perhaps the first to popularize this idea in his "Cosmic Calendar" metaphor in The Dragons of Eden, 1977. More recent treatments are Richard Coren's The Evolutionary Trajectory, 1998, Laurent Nottale, Jean Chaline, and Pierre Grou's Les Arbres De L'Evolution (The Trees of Evolution), 2000, Eric Chaisson's Cosmic Evolution, 2001, and Ray Kurzweil's The Age of Spiritual Machines, 1999 and The Singularity is Near, 2005. Is cosmic acceleration an observer-selection bias? Or is it a developmental trajectory for intelligence? And what about our belief that technological acceleration will continue unabated in special domains? So far this has proved to be a highly practical intuition.
Second, why do our most complex newly emergent systems use dramatically less physical resources (space, time, energy, and matter, or STEM) to do their "computing"? In other words, why has dominant universal computation moved from galaxies to replicating solar systems to the living surface of special planets to the animals on these planets, to the talking animals, to their even faster and more efficient technology? Expecting the ability to create further resource efficiency in our technology is today only a belief, yet our industries invest tens of billions in this expectation annually.
Third, if hospitable conditions for organic life are ubiquitous, as many astrobiologists suspect, and if life must on average develop complexity at accelerating rates, why is our visible universe and galaxy not saturated with observable intelligence? This question, Fermi's Paradox, led me to surmise that outer space is a "rear view mirror" on the growth trajectory of universal intelligence. Our visible universe is a history of older, slower, and simpler computational systems that came before us. It appears to be the domain of the past, not the future.
Fourth, why do rates of emergence during the first half of universal history stand in inverse to the second half? For billennia after the Big Bang, emergence takes increasingly longer, as physical processes occur with decelerating energy and interaction rates in an expanding cosmos. Juxtapose this to accelerating and increasingly local self-organization in the second half, as in Earth's six billion year history, and we discern a U-shaped curve for rates of change. In biology, global deceleration occurs as seed creates soma (body), and later, multi-local acceleration as the mature soma generates new seeds. Is our local intelligence a universe seed? Are we engaged in a universal developmental cycle? This analogy may offer deep insight into the future role of local intelligence.
If we are becoming a seed for the next phase of our developmental journey, this would explain both our accelerating local knowledge about universal systems, and the rapidly increasing "STEM density" of the leading edge of computing systems in universal history. I suspect universal intelligence gravitates to the microscopic universe, into the denser and computationally richer environments of "inner space," not outer space. Consider the trillions of connections in a human brain, or the billions of even faster and more compact (in free energy rate density, as Chaisson observes) logic gates in our still primitive electronic brains, and the astounding simulations of external and internal reality they create. I expect an "Einstein of Information Theory" will eventually prove, using evolutionary developmental physics, that progeny computing systems in maturing universes must in fact become relentlessly more local, going from chemo to bio to techno to cyber, and from outer to inner space. Eventually, that trajectory might require them to transcend the fabric of what we call physical reality, into some multiverse or hyperspace whose presence our current theories can only hint at.
In the nearer term, we may soon transition from our information age to a symbiotic age, where computers engage with us in a "human-level" of dialog, through a conversational interface (CI), simplistic at first. Statistical natural language processing (NLP), through search platforms like Google, utilizing exponentiating stores of archived human data, may incrementally deliver a crudely human-level CI over the next generation, or perhaps longer depending on our foresight and resolve. Soon after, to better serve us, I believe our computers will be modeling our internal states ("personality capture") and mapping and sharing our preferences (the "valuecosm"). I see many potentials for damage and dehumanization in early versions of these developments, but believe their ultimate effect will be profoundly empowering.
Meanwhile, in biological space, hierarchy and development theory suggest sharp limits on our species' capacity for further genetic complexity generation. Our species' newly-discovered genetic homogeneity, by comparison to other mammals, may exist not because of mass extinction (e.g., Lev Zhivotovsky et. al., 2003) but due to the decreasing value of human genetic variation after language emerged. I suspect language was developmentally inevitable, but our genes blundered randomly into it, using evolution. Once complex mimicry language is present, the most important local computation apparently occurs in our collective semiotic mind, and increasingly, in our artifacts.
If further genetic evolutionary variation must create many semiautistic mutants for every occasional improvement, I suspect (and we should eventually prove) that selective pressures would act increasingly against such variation as linguistic societies emerge. Such unproven developmental "path dependency" of the human genome would make sense to me in a world where biological change, in the human sphere, has become so slow as to be future-irrelevant by comparison to our technological extensions. It will be our cyberselves, not our bioselves, that inherit our extraordinary universal future.