Singularity Watch
HomeNewsletterReading GroupsConferencesPublications"Singularity Studies" LinksDegree ProgramsCritiques


Convergent Evolution (Convergent Development)

Uncovering the "standard attractors" for continually accelerating physical-computational development in our universe

© 2005-2008, John Smart. Reproduction, review and quotation encouraged with attribution.




Evolution and Development

Troodon and the Dinosauroid Hypothesis


A Growing Zoology of Convergent Developments

References (very partial list)


Convergent evolution is the evolution of species from different taxonomic groups toward a similar phenotypic or functional form. More generally, it is the hypothesis that in our universe of particular physical constants, laws and properties, there must be optimally convergent structures waiting to be "computed" by physical systems at every level of complexity as they evolve and develop in the environment. This hypothesis makes the assumption that all physical systems in the universe are engaged in some form of competitive and cooperative computation concerning their physical environment, and that optimal adaptive forms must emerge in similar but separate environments, as all universal systems seek to compute their local reality to the greatest extent they can, given their limitations.

Computation-centric or "infodynamic" perspectives on universal change have long been championed by physicists, computer scientists, and mathematicians such as Alan Turing, Ed Fredkin, John A. Wheeler, Stephen Wolfram, and others. They represent a general paradigm that has been quietly gaining ground every year.

Probably the most important group helping us understand the emergence of universal computational complexity as an evolutionary developmental ("evo-devo") process today are a pioneering group of biologists, such as Stan Salthe, Development and Evolution, 1993, Ian Stewart and Jack Cohen, Figments of Reality, 1999, Gerd Müller, Origination of Organismal Form, 2003, Simon Conway Morris, Crucible of Creation, 1998; Life's Solution, 2004, Mark McMenamin, The Garden of Ediacara, 1998, John Maynard Smith and Eörs Szathmary, The Major Transitions in Evolution, 1995; The Origins of Life, 2000, and John Odling-Smee et. al's, Niche Construction: The Neglected Process in Evolution, 2003.

These new biologists are perhaps best called "meta-Darwinists," as they encompass Darwinian and neo-Darwinian evolutionary insights yet also refuse to ignore the broad evidence for what is often called "convergent evolution" seen in life's evolutionary developmental record. Examples of convergence can be found at all scales, from biochemistry to plant and animal forms to life cycles to human culture, as noted by Connie Barlow in "Let There Be Sight!: A Celebration of Convergent Evolution", 2004..

Simon Conway Morris is by many accounts the lead investigator in this distinguished group of meta-Darwinists. The best popular work discussing the universe as an evolutionary developmental system to date is probably Conway Morris's, Life's Solution, 2004, though it still gives only a framework for these ideas. In Natural History magazine, December-January, 1998, Conway Morris famously debated neo-Darwinist Stephen Jay Gould about the inevitability and persistence of Cambrian phyla, the ubiquity of evolutionary convergence, and the independent evolution of all the higher human attributes, such as intelligence and self-awareness. Conway Morris cites octopi (mollusks) as a classic example of independently developed higher intelligence (did you know octopi have two neurologically prehensile tentacles, for example?), and notes the inadequacy of neo-Darwinian "contingency" models of the macroevolutionary process to explain both the stable persistence of the 35-odd Cambrian body plans and the redundant discovery of classic physical computational forms (homoplasy) by a variety of distinct genetic lineages.

These exciting new theories are actually updates of older models about the directionality of evolution, advanced by key paleontologists and biologists of the 1940's and 1950's, chief among them the Jesuit priest-paleotologist Tielhard de Chardin (The Human Phenomenon (mistranslated as The Phenomenon of Man), 1955/99), and less centrally the zoologist Ernst Mayr (Systematics and the Origin of Species, 1942), the humanist biologist Julian Huxley (The New Systematics, 1940; Evolution: The Modern Synthesis, 1942; Evolution and Ethics, 1943; Towards a New Humanism, 1957, etc.) and others.

Mayr talked carefully about the evidence for speciation as a "mark of evolutionary progress", even as he fashioned the modern non-directional evolutionary synthesis. Huxley also talked carefully about progressive evolution, including the way local evolution was now being directed primarily by human culture (through culturally transmissable "mentifacts" "socifacts, and "artifacts", early versions of Richard Dawkin's "memes"), making earlier biologically heritable forms of evolution so slow as to be future-irrelevant, and he noted the way evolution generates ever greater variety, complexity and specificity of organization over time, at the same time as this process usually leads into dead-ends, for any average path taken.

But it was De Chardin who talked the most openly and boldly about the "evolutionary" trajectory toward local superintelligence amongst a special subset of emergent forms, and of the universe engaging in what he called "cosmic embryogenesis" (the best phrase I have yet heard to describe universal change), through a hierarchy of statistically determined phases of development (geosphere, biosphere, noosphere, in his language). De Chardin's is the first coherent discussion of universal systems developing, vs. simply evolving, to my knowledge, even though he did not always use the term development to describe this progressive and directional process of intelligence emergence. While often on the mark, early paleontologists have occasionally confused the terms "evolution" and "development" when describing macrohistorical biological change, which has slowed down the emergence of the new evolutionary developmental paradigm.

This language problem persists today. Many of our leading modern theoretical biologists and directional paleontologists, talk about "convergent evolution", but only a small number of them describe the emergence of structures like eyes (which modern molecular genetics tells us were invented independently at least 20 times) as a process of "convergent development" or "convergent evolutionary development." The phrase "convergent evolution" is both incorrect and oxymoronic, as evolutionary processes, while they are the large majority of the average processes we observe in any sample of events, have no intrinsic directionality.

In other words, the emergence of "homoplasy" (convergent function via independent evolutionary pathways) is best seen as a process of not simply evolution but evolutionary development, occurring multilocally throughout the universe. Stan Salthe, who wrote the fascinating theoretical work Development and Evolution, 1994, is probably the closest individual to my way of thinking that I have yet discovered in this regard, though I am sure there must be scores of others.

Putting together the insights of modern evo-devo and meta-Darwinian biologists like those described above, and positing that the same process of evo-devo is going on at universal scales (the most parsimonious assumption, if the universe is just another complex adaptive system) is my own contribution to the grand detective story, and the subject of my forthcoming book, Journey and Destiny. All I'm really attempting is to slightly update the work of De Chardin and other progressive evolutionists (progressive evolutionary developmentalists, in my language) of the early and mid 20th century, and to provide the mechanism that they lacked, as well as documenting some of the new evidence for convergent evolution (convergent development) that has emerged in the decades since their work.

As one such example, John Odling-Smee (updating Julian Huxley, and others) is a meta-Darwinist who proposes that evolutionary developmental directionality is aided by a process of local niche construction. In its most obvious implementation, this is a process where increasing planetary biological intelligence constructs increasingly immune and interconnected niches on Earth which then select for the further evolutionary development of intelligence within a dominant subset of hierarchically emergent forms, in an ongoing developmental co-evolution between dominant organisms and their increasingly constructed environment, at the leading edge of computation on the planet.

In this perspective the physical computational networks and feedback loops sustaining "Gaia" are simply lower level versions of the feedback loops sustaining a modern city. Both of these "built environments" exert broad constraints on the randomness of evolutionary selection once they have stably emerged. A developmental trajectory (or "womb" for the next evolutionary developmental emergence), ultimately based on the physical laws of the universe we inhabit, is thus increasingly locally constructed over time.

While directional meta-Darwinian views of the universe are presently still speculative, their great advantage is that they account for the six-billion-year history of continually accelerating universal change (e.g., Carl Sagan's Cosmic Calendar), dramatically better than any other paradigm yet set forth (most neo-Darwinian models either deny or ignore accelerating change, or at best treat it dismissively as a 'random accident.') We reject this view, and instead recognize that the special physics of this particular universe we inhabit not only appears rigged to favor computation, but also continually accelerating computation within a special developmental subset of emergent forms.

Such views are deeply congruent with modern insights in developmental biology that virtually all embryonic structure is chaotically, randomly, evolutionarily constructed, in an iteratively tuned process that leads to developmental convergence on a special subset of macroscopic emergent structures and properties. Finally, if one can show a developmental value to the accelerating emergence of local intelligence, then the fundamental attractiveness of developmental programming for accelerating computional emergences becomes clear. It is my contention that when that emergent intelligence can nonrandomly influence the developmental parameters expressed in the subsequent iteration of the system, such value is always present.

If you dismiss the evolutionary developmental paradigm, we challenge you to come up with an alternative model that implicitly predicts our universal record of continuous and ever more local accelerating change.


Evolution and Development: Unpredictability and Predictability

As we have discussed elsewhere, the emerging paradigm of universal evolutionary development, or "evo-devo", tells us there is a major difference between evolutionary and developmental mechanisms when viewed at the universal scale. Evolution and development appear to be two fundamental classes of change occurring in all physical systems.

With regard to evolution, science has shown us that the overwhelming majority of universal events are predictably unpredictable (e.g., highly chaotic and contingent). Hence, those futurists attempting to forecast various (typically evolutionary) events, such as whether a particular company, nation, policy, or cultural feature will emerge at a particular time in a particular environment, have historically had a dismal performance record within any average sample. Furthermore, the faster planetary sociotechnological change occurs, the shorter our prediction horizon becomes for all evolutionary events. Business plans for many technology companies, for example, are now commonly based on forecasts of quarters, rarely years, and almost never decades.

Yet at the same time, there is a special subset of computationally-related developmental trends that have become increasingly easy to predict, the more decades of performance data we have in hand. These trends include accelerating technological capacities of all types, such as a generalized Moore's law over the last 110 years, independent of specific computer manufacturing paradigm, Dickerson's law for proteomics over the last 40 years, Poor's law of network node density for at least the last 30 years, Cooper's law of wireless bandwidth over the last 30 years, etc. Each of these capacity growth models, and many others, have remained highly predictable over long spans of time, and will remain so for the forseeable future in our current understanding of physical law. Such evidence strongly argues that these are developmental, not evolutionary events.

Thus a new group of developmental futurists, armed with insights into the predictable infodynamic nature of the universe, is beginning to make testable, falsifiable short-term forecasts with regard to a range of accelerating technological capacities. As these continue to be verified in coming years, our developmental models of accelerating change can only become increasingly accurate with time. Universal meta-trends such as increasing space, time, energy, and matter efficiency and density (aka, 'STEM compression') are becoming increasingly apparent. This view of the purpose of planetary developmental history leads us to understand that certain biological forms represent computational optima, given the sharp constraints of building computing systems on top of DNA-guided protein synthesis and cellular architectures that must emerge, bottom-up, out of delicate peptide bonds and wet-solution chemistry.

Specifically, studies in convergent evolution (e.g., convergent universal evolutionary development), are proposing an ever-growing number of physical structures and processes that appear to be inevitable emergent forms, global developmental attractors for increasingly STEM efficient general purpose computation. The fields of cosmology, astrobiology, evolutionary developmental biology, and science and technology studies, for example, are helping predictive futurists to understand that there has been convergent evolutionary development in a long succession of physical, chemical, biological and technological structures over time.

Troodon and the Dinosauroid Hypothesis

As one prominent example, consider the work of paleontologist Dale Russell, who along with Ron Séguin was an early champion of the dinosauroid hypothesis (Russell and Séguin 1982). Briefly, this is the idea that our own anthropoid/humanoid forms were a preexisting standard attractor waiting patiently in the universe's computational phase space to be eventually discovered by dinosaur evolutionary development, had they gone on living past their sudden extinction event.

In making his case, Russell noted that a number of small dinosaurs (raptors and oviraptors such as Troodon, Deinonychus, and Stenonychosaurus) had all developed bipedalism, binocular vision, complex hands with semi- and fully-opposable thumbs, and brain-to-body ratios equivalent to a number of today's birds. They were intelligent pack-hunters of both large and small animals (including the mammalian precursors to us) both diurnally and nocturnally. They would likely have become the dominant planetary species due to their superior intelligence, hunting, and manipulation skills, had not environmental conditions suddenly and drastically changed 65 million years ago, at the Cretaceous-Tertiary (KT) boundary.

The current best model for the sudden extinction at the KT boundary is the metorite impact hypothesis, supported by planetary iridium deposition data. In another example of rare intuition, Russell co-published a supernova KT extinction hypothesis in Nature in 1971. Walter Alvarez later credited him as the first paleontologist to seek an extraterrestrial origin for the KT extinction.

Russell received a lot of academic and popular criticism, even scorn, for the publication of his dinosauroid conjectures in the early 1980's. In hindsight, the ferocity of these responses seem quite out of proportion to the reasonableness of the idea. The most obvious motivation was the academic turf-wars famously engaged in by neo-Darwinists against any competing paradigm to "frozen accident" dogma. But the criticism may have also been assisted by more subtle psychological factors. In the same way that Copernicus and Darwin both advanced theories that were initially unsettling and deflating to the human psyche, acceptance of Russell's hypothesis requires us to to realize that our own lineage, our entire pongid/ anthropoid heritage, may have been in many ways unnecessary to the accelerating development of all the higher human qualities and ethics we prize. While we have certainly taken our own unique and intrinsically valuable path, many aspects of the humanoid destination may have been statistically, developmentally determined all along. As another explanation for the irrationality of criticism he faced, some of us might feel an instinctual, species-level fear when we contemplate what could likely have been our sauroid successor (illustrated to the right), had environmental events taken a less catastrophic pathway 65 mya. Perhaps it is psychologically easier to simply deny these insights and attack the messenger. Yet Russell's model was a conservative extrapolation of well-known trends in higher brain masses and brain-to-body mass ratios within the more successful terrestrial species over evolutionary time (sometimes called Marsh's law), as well as a recognition of the superior energetic efficiency of upright posture in slower-moving bipedal organisms.

Here's an important thought question to consider: Why would the most successful dinosauroids necessarily become slower moving? Developmental STEM compression insights become especially useful here. It is a predictable hypothesis that the more that mimicry and (later) memetic intelligence are encoded in any species, the slower they will tend to move in physical space. Once a species is culturally computing using behavioral mimicry (and later, sounds), within high-density living environments, and coordinating such collective mimicry defenses as throwing rocks at 80 mph (which requires opposable thumbs and strong arms) at any advancing predators, such favored bipedal species no longer need to be particularly fast, thick-skinned, or sharp-taloned.

From this point forward, by moving more slowly on average, they can more rapidly increase the complexity of their newest substrate for evolutionary developmental computation: mimicry and memetic culture. Of course, the most refined example of this trend today is the mostly-hairless hominidae one finds inhabiting modern civilization. Consider how our species' previously more horizontal, Troodon-style backs have recently been coaxed by civilization into an almost exclusively upright position (for maximum hand manipulation ability), yet our skeletal systems apparently haven't had sufficient evolutionary time to fully reconfigure themselves for this new orientation, hence the sharp "scoliosis curve" of our lower back with its attendent back pains. Today we spend most of our days sitting mostly inactively inside large boxes (now mainly in front of electronic boxes), or moving between boxes inside smaller wheeled boxes, while our collective computations flow across the planet at the speed of light. The brains of our electronic successors (not their sensors and effectors) will most certainly be even more immobile still, if the developmental singularity hypothesis is correct.

Stephen Jay Gould once stated that if large brains were inevitable, we should have seen them appear much earlier in dinosaur development. But this proposition ignores important developmental context. Large brains might not have provided superior adaptation until they were able to achieve some minimum new performance threshold, some preexisting attractor that could only be achieved within the right development environment. Collective rock throwing is one obvious early applied technology for large social brains. It may have been among the first highly effective tool uses for groups, rather than for individual hominids, and once it emerged, it permanently moved the leading edge of planetary evolutionary development out of body plans and into culture. Yet rock throwing alone simply could not have been effective against the large and speedy sauropods (e.g., Allosaurus, Tyrannosaurus) that dominated the Mesozoic era. In general, environmental conditions in that era were very likely not yet conducive to a rapid selection for significant cultural intelligence. Nevertheless, such intelligence did slowly emerge anyway, in rudimentary form, in the pack hunting behavior of the small raptors.

As the density of raptors increased in environmentally favored areas, and as their average range of travel was able to decrease given their greater hunting and metabolic efficiencies, it seems very likely that various subspecies with greater and greater intelligence (and strong forearms, unlike the dinosauroid above) would have naturally emerged. Eventually, unintelligent sauropods the world over would have been decimated by growing enclaves of rapidly-replicating, technology-using, security-oriented, pack-hunting dinosauroids, in the same way that a highly imitative and innovative Cro-Magnon apparently extingished the technologically and culturally-bereft Neanderthal.

We might not want to believe these informed speculations, but what other credible alternatives are there that are consistent with our planetary history of continuous accelerating innovation in a special subset of complex systems over time? The developmental acceleration appears to be statistically inevitable, deeply engineered into universal physics and achievable via a wide range of exploratory evolutionary paths.

A Growing Zoology of Convergent Developments

Today, astrobiologically ubiquitous universal emergences such as carbon-based organic chemistry, proteins, nucleic acids, lipids, cells, cell nuclei, nervous systems, eyes, binocular vision, trichromatic vision, jointed limbs, wings, opposable thumbs, bilateral symmetry, humanoid forms, mathematics, wheels, internal combustion engines, automobiles, silicon-based computing systems, and the paradigm of scientific investigation are among a large and growing class of proposed convergent evolutionary developments. Such events are now suspected to be not simply evolutionary adaptations for particular niches, but rather universal developmental attractors for accelerating local computation, given the unvarying physical laws of the common universal environment that these systems inhabit.

Learning to discriminate between evolutionary and developmental events when considering the future is both an art and a nascent systems science.

Critiques? Other examples to add? Send us your feedback at johnsmart{at}


References (very partial list)

Hecht, Jeff. 2007. Smartasaurus. Cosmos 15:40-41
Russell, Dale and Séguin, Ron. 1982. Reconstruction of the small Cretaceous theropod Stenonychosaurus
    inequalis and a hypothetical dinosauroid. Syllogeus 37:1-43.