Cosmological natural selection (fecund universes)

From Evo Devo Universe
Jump to: navigation, search

Cosmological natural selection (CNS), also known as fecund universes, is a prominent theory of universe evolution, development and reproduction originally proposed by eminent theoretical physicist and quantum gravity scholar Lee Smolin in 1992. See also our page on Evolutionary Cosmology.

Universe reproduction via black holes

According to CNS, black holes may be mechanisms of universe reproduction within the multiverse, an extended cosmological environment in which universes grow, die, and reproduce. Rather than a ‘dead’ singularity at the center of black holes, a point where relativity theory breaks down and spacetime and matter-energy become unmodeled, what occurs in Smolin’s model is a 'bounce' that produces a new universe with parameters stochastically different from the parent universe. Smolin theorizes that these descendant universes will be likely to have similar fundamental physical parameters to the parent universe (such as the fine structure constant, the proton to electron mass ratio and others) but that some of these parameters, and the laws that use them, may be altered in some stochastic fashion during the replication process. Each universe therefore potentially gives rise to as many new universes as it has black holes.

In a process analogous to Darwinian natural selection, those universes best able to reproduce and/or adapt would be expected to predominate in the multiverse. As with biological natural selection, universal mechanisms for reproduction, variation, and the phenotypic effects of alternate parameter heritability must be found for the model to be valid, and may be explored by simulation. To assess adaptation, proposed universal fitness functions (black hole fecundity, universal complexity, etc.) may be simulated to the extent present physical theory and computation allow, by exploring phenotypic features in the ensemble of possible universes adjacent to our present universe in parameter space. But strategies for validating the appropriateness of fitness functions remain unclear at present, as do any hypotheses of adaptation with respect to the multiverse, other universes, or other black holes.

Smolin states that CNS originated as an attempt to explore the fine-tuning problem in cosmology via an alternative landscape theory to string theory, one that might provide more readily falsifiable predictions. Most CNS models propose an organic type of reproduction with inheritance for universes, with an extended, branching chain of universes exploring a ‘phenospace’ of potential forms and functions within the multiverse. The question of what constrains this exploration, either within the multiversal environment (adaptiveness) or via self-selection, opposing the fecundity that establishes it, remains undetermined at present. McCabe (2006) states that research in loop quantum gravity “appears to support Smolin’s hypothesis” of a bounce at the center of black holes forming new universes (see also Ashtekar 2006). There are loop quantum gravity models in which parameter memory is rigorously retained at the bounce (Corichi & Singh, PRL 100, 161302, 2008) and others in which parameters can randomly reassort (Gambini & Pullin, Discrete quantum gravity, gr-qc/0306095, 2003). In Smolin's model, there is no gravitational singularity at the center of black holes, but rather a structure that allows a new universe to emerge. Roy Kerr ("Do Black Holes Have Singularities?" (arXiv:2312.00841, 2023) proposes that all real black holes rotate (most at incredible speeds), and challenges the standard interpretation of the Penrose singularity theorems that physical singularities exist inside rotating black holes. Ashtekar & Singh ("Loop Quantum Cosmology: A Status Report", Classical and Quantum Gravity, 2011) propose that the Big Bang singularity is replaced by a quantum bounce when matter density reaches the Planck scale. There are also regular black hole models that replace the singular interior with regular geometry (see Ansoldi, arXiv:0802.0330, 2008 for a review). With respect to white hole theories (new universe creation via black holes), Carl Rovelli and collaborators have a series of papers (2014-2020) arguing that black holes quantum-tunnel into white holes over very long timescales in our universe due to black hole time dilation, with the interior singularity repllaced by a bounce. At present, we simply don't know which theories are the most explanatory. Brian Cox and Jeff Forshaw's trade book, Black Holes: The Key to Understanding the Universe, 2022, is a good popular overview of some (but not all) of these competing theories.

According to The Life of the Cosmos (1997), Smolin's trade book on CNS and other subjects, by the mid-1990’s his team had been able to explore, via mathematical simulations (parameterization), eight of twenty then-known fundamental parameters in the standard model of particle physics. In such simulations, Smolin claims our present universe appears to be fine-tuned both for long-lived universes capable of generating complex life and for the production of hundreds of trillions of black holes, or ‘fecundity’ of black hole production. In 2004, he updated these simulations to show that almost every small change in the Standard Model parameters either leaves black hole production unchanged or decreases it. In other words, a significant fraction of universal parameters are selectively neutral for universe reproduction. They allow (are robust to) "neutral drift". This is the same dynamic found by evo-devo scholar Andreas Wagner (Arrival of the Fittest, 2014) that developmental genetic networks are not only robust to hierarchical convergence on complexity transitions, they are robust to drift. How and when learning networks bias internal complexity to increase or decrease with such drift, allowing the self-organization of complexity, is one key area of investigation.

Smolin's theory has been critiqued on occasion (Vaas 1998; Vilenkin 2006), and continues to be elaborated and defended (Smolin 2001,2006). Vilenkin calculated that the cosmological constant (lambda) could be varied to produce greater numbers of black holes than exist in our current universe. Vilenkin proposed that this "falsifies" the CNS model, but that is overclaiming. It is true that in a 1994 preprint (gr-qc/9404011), Smolin proposed that universal parameters have been selected for values that "extremize" the production of black holes, but Smolin did not argue for a black hole maximum. Even the term Smolin chose for the average black hole production function, fecundity, does not connote a global maximum. In a 2004 preprint (hep-th/0407213), Smolin proposed the fitness function for black hole production would presumably require small changes across generations, a path dependence that would allow at best a local maximum. This fits with life's dynamics, in which seeds are never maximally produced in any species, only in numbers fecund enough for survival.

If our universe's "parameter genetics" are anything like biological genetics, we may expect to find an evo-devo model, where previously successful parameters and laws gain "inertial memory" (developmental canalization, see Waddington 1942) over replication history, while others with less history are randomly evolved at the new replication event. We find this dual-category view in the theory of facilitated variation (Gerhart & Kirschner, The Plausibility of Life, 2006), which divides life's genetic parameters into a small, robust, and vitally important core, which facilitates variations in the periphery which are biased to be viable. Systems biologists have explored a distinction betweeen stiff and sloppy parameters in genes and other factors. Gutenkunst et al. ("Universally sloppy parameter sensitivities in systems biology" PLoS Comp. Biol. 3, 2007) and Transtrum, Machta & Sethna (Science 342, 2013; PRE 83, 2011) model parameter spaces generically. They find a few stiff directions where behavior is exquisitely sensitive (analogous to universal fine-tuning), and many sloppy directions where it barely responds. If internal complexity has any nonrandom effect on universe reproduction, as it does in life, one might expect co-selection for fecundity and intelligence in cosmic phylogeny, and persistent tradeoffs between computation and experimentation with seed (big bang) parameters, organism (universe) emergent structure and function, and the niche-constructed (multiverse) environment (if it exists) (SOE partitioning) (Smart 2008).

A promising analytic approach is Lee Altenberg's "Implications of the Reduction Principle for Cosmological Natural Selection" (arXiv:1302.1293, 2013). Altenberg enlarges the universe's state to a meta-state that encodes both the effective law and the state it acts on (Smolin's own phrase, which he quotes). Then he lets transmission (inheritance) laws vary themselves across universes. Via Karlin's theorem of growth-versus-mixing (explored in Altenberg, PNAS 2012) he proves that the ensemble comes to be dominated by the transmission law with the highest inheritance fidelity. The model proposes that tradeoffs between fitness and faithfulness goals provide a better model for evolved fundamental parameters as compromises, and not simply local optima seeking to maximize universe production.

This dual-category parameter model is consilient with the work of Fred Adams, "Stars in other universes: stellar structure with different fundamental constants" (JCAP 2008) and "The degree of fine-tuning in our universe and others" (Physics Reports 807, 2019), who argues that stars remain viable over a surprisingly wide swath of parameter space. Harnik, Kribs & Perez, "A universe without weak interactions" (PRD 74, 2006), also construct a weak force-free universe with stars, chemistry, and structure. Thus it appears that many of the universes ~31 presently known fundamental physical parameters look sloppy with respect to complexity production.

Gardner and Conlon (2010) apply the Price equation to CNS, proposing that if black holes create universes in which constants can vary randomly, then CNS offers a fully naturalistic explanation for both the apparent fine-tuning of universal constants and our universe's fecundity of black holes. Taking an adaptationist view of CNS, in an approach called CNSI (CNS with Intelligence) Price (2017) argues that intelligence is more likely than black holes to be an adaptation self-designed by natural selection. This contrasts with Smolin's original view that black holes are the central adaptation, and intelligent life a byproduct (spandrel) of selection.

In more recent years, Smolin and collaborators have begun to explore Cosmological Learning (CL) models in which the universe is considered a learning system, with internally accumulated information and intelligence, a universe capable of creating new physical laws within its own life cycle. The arXiv paper, The Autodidactic Universe (Alexander et al. 2021) is an attempt to bring self-directed learning theory, including network-centric machine learning models, into correspondence with gauge and gravitational field theory, matrix models in quantum cosmology (a background-independent formulation of string theory), and theories of universal evolution and development.

Antecedents to CNS

The earliest antecedent to CNS may have been the oscillating universe model of Alexander Friedman (1922). Another oscillating model was the phoenix universe of Georges Lemaître (1933). In 1973 and 1977, John A. Wheeler proposed that the basic laws and constants of the universe might fluctuate randomly to new values at each successive bounce (new universe birth) in an oscillating universe, and thus provide a natural mechanism for anthropic selection, how we come to find ourselves in a universe that is fit for life.

Beginning in the 1980’s theorists in quantum gravity began postulating that our universe might ‘give birth’ to new universes, or causally disconnected regions within our own universe, via fluctuations in spacetime over very short distances (Steinhardt 1983; Vilenkin 1983; Strominger 1984; Linde 1986; Hawking 1987,1988,1993; Coleman 1988). Some theorists (Hawking 1987; Frolov 1989) proposed that new universe creation might be particularly likely in the singularity region inside black holes.

Perhaps inspired by the work of Hawking and Frolov, philosopher Quentin Smith published a paper (Smith 1990; commentary by Stenger 1999) proposing that random symmetry-breaking events in the initial Big Bang singularity might lead to the production of new universes via black hole singularities, and that extrapolating this process to past universes could provide a naturalistic explanation for the basic laws and constants of our universe.

In 1992 Smolin (independently of Smith) described essentially the same process in more rigorous terms, in the first peer-reviewed hard sciences paper on CNS (Did the Universe Evolve? Classical and Quantum Gravity 9:173-191).

The Smolin-Susskind CNS Debate

In a series of public exchanges, including a 2004 debate on the anthropic principle on Edge.org, string theorist Leonard Susskind argued that Lee Smolin's cosmological natural selection (CNS) hypothesis is flawed because it predicts we live in the fittest of all possible universes, with maximum black hole production, and then refutes it with eternal inflation, which if it counts as reproduction, would predict a maximum cosmological constant, which we clearly do not have. Unfortunately from the start, this was a mischaracterization of CNS. The fittest of all possible universes is a global claim, and hill climbing with successive small mutations in the parameters never predicts a global optima, both in biology and in CNS cosmology.

Susskind's main points in that debate were:

  • The Prediction of Maximal Black Holes: Susskind argued (similar to Vilenkin on this point) that if Smolin's theory were correct, then our universe should look very different. The fundamental physical constants should be tuned to produce the absolute maximum number of black holes possible.
  • Contradiction with Observation: Physicists can model universes with slightly different constants. It is possible to create a universe that would produce far more black holes than our own. For example, a universe with lumpier initial conditions or different stellar physics could theoretically be more "fertile."
  • The "Wrong" Parameters: Susskind pointed out that the observed parameters of our universe seem tuned for creating long-lived stars, complex chemistry, and life, not simply for churning out black holes. The fact that our universe is not a maximal "black hole factory" suggests that maximizing black hole production is not the driving evolutionary principle.

This critique was a key part of the intellectual sparring between Smolin's CNS and Susskind's continued support of the weak anthropic principle as an explanation for the fine tuning problem in physics.

Unfortunately, this debate between two distinguished physicists did not explore how replication and selection work in biological environments. Living replicators are never maximal producers of gametes (seeds) in order to be adaptive. As biologist Tom Kirkwood has proposed in his disposable soma hypothesis (1977;2000;2008), living systems are always forced to divide their energies between germline (seed) and soma (organism, body) production and maintenance. There are physical, computational, and selective costs to maximizing or protecting against entropy with either form of existence, both of which are persistent. Think of how Cicadas "choose" to hide out for 13 or 17 years as nymphs (nymphs may be thought of as an juvenile form of organism, closer to eggs than adults, in many ways) as a viable adaptive strategy. Consider also that universes may be constrained to interact with other universes in the multiverse environment, and compete with them for space and resources, as CNS theorist Ward Blonde considers (Blonde 2025). If there is any competitive or other limit on the use of multiversal resources, or any interaction between replicating universes, selection will be at work.

Systems theorist John M. Smart, in the SOE partitioning of intelligence hypothesis, proposes that every evo-devo replicator opportunistically partitions its intelligence between three actors: the seed, the organism, and the environment (Smart 2008). This partitioning is obvious in complex replicators, like humans, who extensively niche-construct their environment to facilitate adaptation. Yet we also see niche construction even in simple replicators (paramecium, fungi, insects), as they cognitively choose where, when, and why to make habitats and act, irreversibly modifying their microenvironment, often to their collective benefit. We see extremes of this in the spider's web and the termite's mound, but such modification is universal to life. Some modifications may also be macroenvironmental in effect, and largely unconscious network dynamics, as in the Gaia hypothesis (planetary chemogeobio homeostasis). If black hole intelligence can nonrandomly influence the parametric structure of a new universe's replication, the multiverse itself may have been incrementally modified over past replications in support of greater autopoetic complexity and intelligence. Might such niche construction be an explanation for the nature and distribution of dark matter?

As summarized above, a 2013 preprint, theoretical biologist Lee Altenberg explored implications for CNS of the Reduction Principle (Feldman 1972) an observation that when mechanisms of variation themselves vary, selection favors greater faithfulness of reproduction (perhaps faithfulness of reproduction only in the "core" that facilitiates variation) and of Karlin's Theorem 5.2 (Karlin 1982) in a multiverse. Under these assumptions, Altenberg proposed that natural selection entrains two completing goals: maximization of fitness and faithfulness of inheritance. Tradeoffs between fitness and faithfulness open the possibility that evolved fundamental parameters are compromises, and not local optima seeking to maximize universe production. The paper's main implication is that parameters that are not local optima for black hole production may be involved in the inheritance mechanism, and subject to contravening selection for preservation of universe properties. In his view, inding a parameter that is not fecundity-optimal doesn't refute CNS; it flags that parameter as part of the high-fidelity transmission machinery--analogous to the evolutionarily "conserved core" in metazoan replicators. A better understanding of the Reduction Principle in evo-devo biology, and of which genetic and other life parameters are "stiff" versus "sloppy" (Gutenkunst et al.), or "core" versus "periphery" (Gerhart and Kirschner) would appear to offer promising new ways to investigate the roles of fundamental parameters in the evolutionary development of universes.

Other CNS criticism and current research

While early oscillation models did not survive criticism, a number of theorists continue to work on more complex and in some cases, more biologically-inspired versions. Examples include cyclic models in brane cosmology (Steinhardt and Turok, 2001; Lehners, Steinhardt, and Turok 2009 and Baum and Frampton 2007), and the cosmological natural selection hypothesis (Smolin 1992,2006). These models remain controversial. Brane cyclic models presently offer no satisfactory description of the bounce via string theory. Likewise, the CNS description of the bounce via loop quantum gravity also remains tentative, and CNS as a framework remains difficult to evaluate without fitness function models. Is there self-selection for autopoetic stability or complexity? Or some form of competitive selection among potential or actual universes? What are the resource constraints and selective variables in the environment?

Since the late 1990's we have had empirical evidence (observation of distant supernovae as standard candles, and the mapping of the cosmic microwave background) that universal expansion is not slowing but is in fact accelerating via dark energy dynamics. In most dark energy models, a future Big Crunch seems very unlikely. Nevertheless, many such models do allow for a multitude of gravitationally-driven "Small Crunches" in an accelerating universe. For example, Nagamine and Loeb (2003) project that our universe must self-fractionate into a number of informational 'islands,' or supergalaxies, each of which undergoes gravitational collapse, while the space between supergalaxies rapidly grows. Over time, the stellar black holes in each galaxy merge with the supermassive black holes at the center of each galaxy, and all matter in each supergalaxy eventually becomes entrapped inside black holes. Our local supergalaxy will comprise the collided Milky Way and Andromeda galaxies and several dwarf galaxies, and our Milky Way and Andromeda begin to collide in just 20 billion years.

When we combine CNS with dark energy models, rather than an oscillating universe which returns all its evolutionary species to a single replication point, our universe appears to be generating a diverse set of continually branching replications, just as we observe in any evolutionary developmental history of living systems exploring a phenospace. Such a continually branching replicative pattern (phylogenetic tree) is also seen in many nonliving replicating systems, such as stars replicating across a developing galaxy. Branching universe ensembles may be a way to maximize diversity, resilience, and adaptiveness, over an alternative "all eggs in one basket" Big Crunch. Questions such as these remain to be made more rigorous in some future information theory of replicative evolutionary developmental systems. A fruitful path toward such a theory may come from careful comparison of evolutionary and developmental similarities between all known and theorized replicative complex systems, whether they be quantum, chemical, biological, cultural, technological, or universal.

Another topic that remains to be explored would be to extend Altenberg's modifier framework (see previous sections) with a third selective objective: fecundity, transmission fidelity, and complexity-permittingness, in both living systems and in CNS. These three objectives map well to evolutionary exploration, developmental conservation, and evo-devo adaptiveness as entrained goals in living systems. We might then ask a tractable question: does the fidelity-optimal parameter subset overlap with the complexity-permitting subset? Is it orthogonal to it? Or does it trade off against it? If there is overlap, we may posit that the conserved cosmological "developmental toolkit" is the same subset that makes complexity possible, offering a model for parameter function. If they trade off, we also get something interesting: if a universe cannot maximize both fidelity and complexity, and the observed parameters are a compromise, perhaps similar to specific cases in living systems under selection. Or perhaps similar to self-organization in nonliving systems. Such a model might even tell us something falsifiable about the replication environment (multiverse). To our knowledge, such an inquiry into the fine-tuning problem has yet to be done.

Cosmological Learning (CL) models

CL models (also called CIS (Cosmological Intelligent Selection), CNSI (CNS with Intelligence) and CAS (Cosmological Artifical Selection)) are a family of CNS extensions that seek to bring information theory, self-organization, learning, intelligence, and network science into the replicative universe framework. They propose that accumulated end-of-universe, or more precisely, end-of-black-hole evolutionary intelligence may nonrandomly improve the values of such variables as the stability or frequency of universe/black hole replication, the capacity for intelligence to influence and predict the improvement of future universal complexity or adaptiveness, or the robustness of universe selection within the hyperverse or multiverse. CL models assume that any universe where emergent intelligence was able to play a less-than-random role in replication or selection might become favored, more resilient, or dominant in some extrauniversal environment, over lineages where emergent intrauniversal intelligence does not increasingly factor into replication, as in Smolin's original CNS model.

Some CL models propose that increasingly learning-oriented, internally intelligent universes might naturally grow out of simple CNS universes at the leading edge of universal complexity, just as we have seen learning and intelligence emerge within environmentally dominant lineages in life's history on Earth. In the most functionally and morphologically complex species on Earth, we may observe that life's learning, memory, and intelligence mechanisms have progressed from "random" recombination of prebiotic or prokaryotic genetic elements, to a much more culturally-guided replication in higher eukaryotes. In this process, we can argue that individual and collective intelligence (memes, knowledge, self-awareness) increasingly influences and constrains the original more fundamentally "random" replicators (genes, DNA).

CL models are thus consistent not only with weak observer-selection anthropic models, where our universal parameters are presumed to be anthropic (intelligence-favoring) primarily because we are here to observe them, but also with strong anthropic arguments such as the fine-tuned universe problem, which postulate that several of our universal parameters appear improbably fine-tuned for the purpose of the emergence of life, complexity, learning, and intelligence. In other words, in CL/CIS, not only universe replication and selection but also universal evolutionary developmental intelligence emergence, conservation, and immunity, and robustness may be considered intrinsic to the developmental telos (purpose, drive, life cycle goal) of the universe, again as long as that intelligence has a less-than-random role to play in universe selection and replication. While today's science lacks a sufficiently advanced information theory to describe the functional role of learning and intelligence in biological evolutionary development, CL models are at least suggestive of the outlines of a such an information theory, and thus worthy of research and critique.

To get complexity influencing the transmitted parameters, you need intelligence to influence or engineer the bounce. One model for this is Louis Crane's Meduso-Anthropic principle, in which very small engineered black holes might be produced by advanced civilizations in astronomical numbers, influencing the population of parameters more than a comparably smaller number of (still astronomical) stellar mass black holes. An alternative is Clement Vidal's Cosmological Artificial Selection, in which advanced civilizations use their scientific knowledge to engineer a subset of black holes toward their own ends, with adaptiveness still determined by selection in the multiverse. In those (miniscule) universe lineages it is no longer natural selection in operation but artificial selection, and it's Lamarckian. That's a big assumption, and Vidal is candid that it is. Nevertheless, accelerating miniaturization trends in the manipulation of matter, energy, space, and time by complex adaptive systems, both life and human history, as described in John Barrow's Barrow Scale of engineering (Impossibility, 1999) do arguably support both proposals. If the "bounce" architecture at the center of black holes is somehow entangled with other black holes, network models, analogous to gene networks in prokaryotic replicators, with horizontal gene transmission, might be the best early examples of parametric intelligence emerging to influence the bounce (Smart 2026).

Cosmologist Stephon Alexander at Brown U, Lee Smolin at the Perimeter Institute, and collaborators at Microsoft Research, have proposed a Cosmological Learning model in which the universe progressively learns laws governing its structure and function. Their arXiv paper, The Autodidactic Universe (Alexander et al. 2021) is an attempt to bring self-directed learning theory, including network-centric machine learning models, into correspondence with gauge and gravitational field theory, matrix models in quantum cosmology (a background-independent formulation of string theory), and theories of universal evolution and development. Contact Stephon Alexander if you would like to contribute to this pioneering research.

Brief history and critiques of CL models

As Cyril Stanley Smith notes, philosophy, art and play often arrive at useful new ideas first. The earliest exploration of CL/CIS appeared first in both philosophy and science fiction. Pragmatist Philosopher Charles Sanders Peirce first proposed the application of natural selection to cosmology (Peirce 1893). A century later, philosopher Quentin Smith explored the idea of new universes emerging from black holes as seeds (Smith 1990). Sci-fi author and physicist David Brin considered that universal intelligence may have evolved via both intrauniversal (supernovae) and extrauniversal processes of black hole replication (Brin 1991). Sci-fi author and physicist Gregory Benford also briefly explores CL in one character's musings in Cosm (1998).

In the scientific literature, Louis Crane's Meduso-anthropic principle (1994) proposed, in an arXiv preprint, that we need to consider a role for intelligence in the cosmological replication process. He also provided perhaps the first clearly biological framework (the medusoid to polyp life cycle) from which to consider this prospect. Crane published an update to this bio-inspired CL model in Possible Implications of a Quantum Theory of Gravity: An Introduction to the Meduso-Anthropic Principle (2010).

Cosmologist Edward Harrison (1995), offered the first CL hypothesis to be published in a peer-reviewed journal, independently proposed that the purpose of intelligent life is to produce successor universes, in a process driven by natural selection at the universal scale.

Complexity scholar James N. Gardner (2000,2003,2007) subsequently explored CL/CIS ideas at length in his selfish biocosm hypothesis. Building on Richard Dawkins (1976) approach to evolutionary biology, Gardner envisioned self-preserving and self-selecting universal replication mechanics, and explored the hypothesis of closed timelike curves with respect to universal replication dynamics. Gardner further proposed that this recursive selection process leads to the likelihood of advanced ancestor intelligences as architects of our own fine-tuned universe.

Systems theorist John M. Smart (2000,2008,2011) authored a developmental singularity hypothesis, proposing that spatial, temporal, energetic, and matter efficiencies in the production of information and computation must lead intelligent civilizations to a black hole-like densities, as a developmental attractor. He coined the CNSI term in his 2008 publication, Evo Devo Universe?. In 2025 he changed his preferred term to CL. Smart explores the speculative idea of a constrained developmental destiny for civilizational intelligence in black-hole-like environments. in his transcension hypothesis (Smart 2011), citing apparent advantages of black hole-like conditions for the perpetuation of life and intelligence in an entropic universe. In contrast to Gardner's end-of-universe 'architects', Smart proposes only a very limited capacity for end-of-universe intelligence to fine-tune universal physical parameters in each replication cycle, as in the way current genetic "engineers" do not engineer, but rather engage in very limited fine-tuning, aided by finite simulations.

Philosopher Clément Vidal (2008, 2009a, 2014), also takes an evolutionary and developmental approach to CL/CIS. Vidal proposes that an intervention of intelligence in the universal reproduction process should be named "Cosmological Artificial Selection" (CAS). In Vidal's CAS hypothesis, our current cosmic blueprint would be expected to have been at least partly fine-tuned, by previous universe-creating intelligence and our own future intelligence, using extensive computer simulations, for the production of universes that are robust to the production of intelligence. The selection process would then not be random or natural, but mediated by intelligence, or artificial.

Physicist Rüdiger Vaas (2009, 2010) has critically examined the extension of intelligence to CNS, noting its potential utility as a way to integrate three streams of philosophical thought: (1) the origin and apparent fine-tuning of our universe, (2) the possible value and meaning of life in the universe, and (3) the survival of universal information and complexity. At the same time, Vaas notes a range of major unresolved research and validation problems. A response to some of these criticisms can be found in Vidal (2009b).

Evolutionary theorist John Stewart (2010) takes a broad evolutionary perspective extrapolating the trajectory of evolution. He also builds on the work of Crane and other authors to explore the idea that "our universe and the evolution of life within it is a developmental process that has itself been shaped by evolutionary processes of even wider scale". He further argues that at a particular point, evolution will continue to advance only if we decide to advance the evolutionary process intentionally.

Evolutionary biologist Michael Price (2017) argues that just as selection is the root source of adapted order in the known universe, if universes replicate in the multiverse, intelligence may be even more likely than black holes to be an adaptation self-designed by natural selection. This contrasts with Smolin's original view that black holes are the central adaptation, and intelligent life a byproduct of selection.

See also

References

Retrieved from "https://evodevouniverse.com/wiki/index.php?title=Cosmological_natural_selection_(fecund_universes)&oldid=8911"