Devo 3. Convergences, Optima (TINA Trends), and Predictions (AI)
Seeing convergence is of course seeing a developmental funnel. It involves recognizing often hidden processes and conditions that are interacting in a way that is reducing local variety and difference, moving complex systems toward a particular future state. Understanding when and why a system, or set of systems, is diverging (evolution) or converging (development) is a critical foresight skill. Diverging systems are in many ways increasingly unpredictable, and converging systems the opposite.
Convergences happen when previously separated products, services, or processes gain much closer interaction, interdependence, or integration than presently existed between them. Technological convergence, aka digital convergence, is a well-known example. Think of voice, data, and video all migrating to a common internet backbone, or many different kinds of operating systems running on one virtual machine in software. Think also of single devices gaining multiple functions, such as many systems on a chip, or many apps running on a smartphone, tablet, or smartTV. Think also of the convergence of the world’s socioeconomic systems on a common set of values, including evidence-based thinking and social democratic capitalism. For an optimistic view on how rapidly both East-West and North-South convergences are proceeding, read Mahbubani’s The Great Convergence (2014). It is a nice follow-on to Fukuyama’s The End of History (2006), which originally popularized this view. Global political convergence can be hard to see in the short term, where reversals are common. But when we look back (and forward) over decades, the signature is unmistakeable.
As we said in Chapter 2, the most important kind of convergence to look for and to understand is dematerialization, a process where growing intelligence is able to reduce or substitute for physical resources and devices. Convergence on adaptive intelligence appears to be the main goal of complex systems. When we understand this, we know why machine intelligence will be the most important story and process of the 21st century. Using accelerating IT to better solve human problems is the most important challenge for the modern foresight community, whether we see it or not.
Optima are a convergent process that that seems to be maximizing for a particular goal or value. This maximization is occurring within a set of stable laws and constraints, some for the system in question, and some related to its environment. We have previously argued that STEM compression is a process of optimizing complex systems for greater resource efficiency or density. We said balancing evo and devo approaches to building AI, via biologically-inspired approaches, may be necessary to optimizing the growth of adaptive machine intelligence in coming years. These are just two of many optimization proposals we can increasingly quantify and test by simulation.
Scholars sometimes talk about optimizing for evolutionary processes, and foresighters might think of optimizing for client preferences, but I prefer to reserve the term optimization for developmental processes. I’d say that we experiment with evolutionary processes, and our environment “satisfices” or “selects” for an adaptive degree of diversity, and for preferred outcomes. But developmental systems have enough stability in them that we can talk of “optimization.”
Developing systems have a framework of laws and constraints (associations, dependencies) that describe them, and any optimizations that happen to them must occur within that framework. That means seeing as many of the likely laws and constraints on a system that we can is as important as guessing what goal or value appears to be being optimized by the system.
Recall our discussion of TINA Trends in Chapter 2. We argued that the deepest understanding of such trends requires taking a developmental, optimization-centric perspective on their emergence. If we think of our planet as a complex system of finite size, becoming more technologically integrated and interdependent every year, and if we propose it is doing so in ways very much like a developing embryo, we can then ask which of our TINA trends (tech, economic, and cultural globalization, transparency, human rights, etc.) look developmental, in a rigorous biological sense.
For example, if we don’t understand that something (in our current model, Hox genes, and their regulatory networks) is acting to constrain developing embryos into expressing a particular framework of segmented body plan and tissue architectures at certain future places and times, we can’t then anticipate in a rigorous fashion how those networks, along with cellular signaling, cellular migration, and chemical diffusion in a developing brain will optimize for the emergence of specific patterns of neural connection. We find many predictable patterns in all higher brains, so we know some kind of optimization is going on, even though the vast majority of the microarchitectural patterns in each brain will different, even when we compare genetically identical twins. But at each stage of development, the more of the laws and constraints on the developmental system and its environment we understand, the better our ability to describe optimization.
Developing a quantitative model of optimization is today a very tall order, for any system, and such models escape us many domains. Even biological development still has few such models, though the amazing Eric Davidson (1937-2015) was able to develop a fully predictive, optimization-rich model of the first few weeks of sea urchin development over the course of his long career. Davidson won the International Prize for Biology for this work in 2011. I wish he’d gotten a Nobel prize, as quantitative models of development are so very important to the advance of evo devo biology and evo devo thinking. In the meantime, given the scarcity of such fully predictive models, we make the best guesses about optimization processes as we can, waiting for our science and senses to grow sharper.
Predictions, forecasts of emergences of specific events, structures, or outcomes in future space and time, are the last developmental factor we shall consider. Unlike optimizations, predictions don’t have to be developmental. We can predict an evolutionary possibility (an experiment that will be tried), an evo devo preference (and its associated strategies and plans), or a high-probability or “inevitable” development. We can attach probabilities to all of these predictions, but the those probabilities will have a very wide range and/or be very low for evolutionary events, and they’ll be much higher probability, with much narrower confidence intervals for developmental events.
Prediction is an art much more than it is a science, but that doesn’t mean it isn’t a valuable art. It is our hope that this Guide will give you confidence to do a lot more prediction, across all the probabilities, as imprecise as those are to you today. Doing predictions will give you weeble stories, that you can test against your colleagues, and then offer to your clients. It will improve your scanning and sensemaking ability, and help you see opportunity and risk ahead of everyone else.
Would you predict that a basic income guarantee must emerge everywhere, as some function of growing technological productivity and social wealth? Will China and the Soviet Union eventually get more representatively democratic? Will English becoming a globally taught and even more dominant language?
There are also negative predictions. Could Chinese ever overtake English as the dominant global language? I consider such a future statistically impossible, for a variety of reasons. English is just too redundantly established, with too large a vocabulary, and Chinese is just too hard to learn by comparison. After the AIs arrive, we’ll surely all be taught a common new global language from birth, which might be English, but might also be something even more useful, designed by them, not us. In the meantime, how many new English speakers do you expect between now and 2050? What handful of other languages will grow speakers during this time? All the rest will continue to lose speakers, of course. What can we say now about the world wide web, smartphones, and the internet of things a generation (25 years) hence?
Many future events or structures can be perceived and predicted in advance, if one has good enough systems knowledge and sufficient clarity of thought and vision. Let us know what you see, and why, and thanks for predicting. The more we do it, the better we get, individually and as a community.
Let’s take brief survey now of a few categories of evo devo foresight. As we do this survey, please keep in mind both the unpredictable and the predictable features of physics, chemistry, and biology that very likely caused our own emergence here on Earth. We will look first at biological and psychological evo devo, then scientific and technological evo devo, then organizational and industry evo devo, then global societal evo devo. This will set us up for thinking about evo devo activism, how we can get better at seeing and advancing “what the universe may want,” even as we are still quite early in discovering those intentions, via the Eight Goals and other approaches.