(I write this primarily in order to rid myself of an intermittent fog of nihilistic thought that arises every time I contemplate the future in the context of my identification with present-day society and the human species. Unfailingly, I begin brooding and digging deeper and deeper into futorological writings, reading frenetically to seek out bright prospects in humanity's future to dispel the existential darkness of the moment. This tends to be a drain on my time as well as being a useless (in the traditional sense of the word) and addictive cycle of repetitive activity. I have resolved, therefore, to put my views on the matter down in print, in the hope that the process will either end my fretting, or at the least channel it into a tangible medium.)
As an avid reader of science fiction during my younger days, I have often remarked at the pedestrian nature of the concepts and gadgetry (though not the literary value) deployed in contemporary science fiction writing. As our understanding of the near impossibility of human extra-solar space travel has gradually consolidated, and as advances in scientific discoveries have declined in cultural significance, writers increasingly have to look to realms of fantasy or sociology to construct their story-lines. An unfortunate consequence of this shift is that the delighful art of peering into the future has been decoupled from artistic imagination and is now the domain of sober scientific speculation either through fitting curves to empirical evidence from the modern industrial period or through seeking predictive power in the present by devising sociological theories of history.
To me, either methodology appears to result in an unfortunate narrowing of focus. In the first case, the predictors tend to be stalwarts of the avant garde of technological innovation (Kurzweil, Joy, Moravec etc), who look upon evolution as a best-case scenario, i.e., they operate on the premise that once technology has been developed, nothing can stop it from being adopted. Their statistical analysis of 50 years of technological discovery leads them to believe that advances will continue to accelerate, leading, in the next few decades, to a situation where, depending on the degree of sanguinity of the predictor (a) scarcity and the need to work will have been eliminated or (b) humans will have been marginalized or exterminated. I suggest that these observers suffer from a sampling bias, i.e., they consider human history in the period before 1800 to have made no contribution to ensuing developments. This arises from the assumption that progress must be physically present and demonstrable to count. A second fallacy lies in the assumption that technological feasibility will inevitably cause adoption. A stark counterexample is the fate of the space program, which has seen no progress in the last three decades through the lack of social incentive. I feel that their predictions, while quite specific, are unlikely to be accurate, since they operate entirely using inductive reasoning and cannot account for feedback effects and other social variables. In the second case, that of sociological analyses, I feel that these observers suffer from the opposite problem, i.e., their models are not flexible enough to be able to make good predictions about the volatile period of human existence that we are living through. While observations regarding social disruption, anomie and commentaries on crime rates, urbanization etc are certainly apposite, these analyses do not have predictive power, since they are based entirely on analogous as opposed to inductive reasoning.
I title this essay `Cycles of life' because the basic thesis herein is a characterization of four stages of existential behavior in all systems that learn to make fewer errors in prediction based on their interactions with their environment. This framework, I then claim, presents a good description not just of an individual human's actions while in existence, but also those of a society's and lastly humanity's. I assign to these four phases of existence names assigned in the Indian Classical system of ashrams since the original system emerges as a special case of the more general model I propose. I first define the ashram framework and show characteristics of systems to which it will apply. Then, I examine the degrees to which it may be applied to entities that we recognize well - the liberal democratic Western culture, the scientific enterprise, less developed societies in other parts of the world, and finally humanity as a whole. By hypothesizing and providing evidence for a parallel between the existential trajectories of humans and societies it becomes possible to draw specific conclusions about the future of humanity in a heuristic and socially interpretable manner, as opposed to empirical methods which, as described above, make severe and dehumanizing assumptions about cultural homogeneity into the future. Furthermore, our mathematical formulation of the generative process affords our model the rigor and predictive ability that sociological approaches tend to lack.
A meta-comment before we begin: I am continually perplexed by what I formulate to myself as the 'dilemma of citation'. It is hard for me to keep track of the observations of technologists, social scientists and philosophers since my reading generally follows Brownian motion rather than a linear scholarly process. At the very outset, therefore, I disclaim all pretensions to originality and virtue in this work and apologize to those whose ideas I may appear to plagiarize throughout this exposition. Since I stand to gain nothing save some emotional satisfaction from the writing, I hereby consider myself to be ethically absolved.
The ashrams of Manu
The Manu Smriti is a Sanskrit text dated to the 2nd century B.C. that contains social commentaries and injunctions as preached (if not necessarily practiced) in India towards the end of the Maurya empire. It has attracted considerable controversy, since it defined and was used as spiritual sanction for the imposition of the infamous caste system. While post-colonially, it is possible to see that the practice was much caricatured by Western interpreters to increment the white man's putative burden, the fact remains that the Manu Smriti remains a controversial reference. Therefore, I wish to clarify that I borrow no ideology from the text, only some convenient nomenclature.
The Manu Smriti divides a man's life into four stages. The first, brahmacharya, is associated with the life of a student: minimal transactional involvement with society, exploratory expeditions and vocational and intellectual learning. The second, grihastha, is identified with the life of a man of the world: extreme transactional involvement, well-established social networks and relationships, sense of duty, responsibility and exaltation of the spirit of hard work. The third, vanaprastha, is identified with the lifestyle of a wanderer: great freedom of movement, lowered sense of transactional responsibility, greater social egalitarianism, increased affinity for simplicity. The fourth stage is called sannyasa, the life of the renunciate, which reflects a complete end to transactional as well as social involvement and a gradual dispersal of the sense of identity, preparatory to its physical cessation. The hypothesis of this essay is that the existential evolution of humans - both as individuals and as communities - proceeds along these broad stages of development. I further claim that this process is characteristic of systems that 'live', for reasons that I discuss next.
To see how the four-fold trajectory is an extremely plausible mechanism for the evolution of living systems, it becomes necessary for us to conduct a gedankenexperiment motivated by computational learning theory. Consider a space of possible outcomes, each of which can affect an agent that chooses them positively or negatively. The manner (positive or negative) in which the agent is affected is governed by the agent's preference relation, which is essentially a collection of statements that the agent stores in its memory to allow it to make choices intelligently. For example, the preference relation 'if hair on fire, then effect of going to movies strongly negative' will allow an agent to choose wisely between going to the movies and finding a bucket of water. It is not known whether all agents are initialized with the same preference relation, but the question is moot, since at every instant in an agent's existence span, it makes choices which affect its preference relation. Thus, it is almost impossible to find two agents with the same preference relation.
Now, if an agent is to survive, it is necessary for it to learn to avoid negative outcomes, a process that is accomplished using both prior knowledge as well as experience of outcomes. The greatest ability for an agent is to be able to generalize, that is, find similarities and analogies between past outcomes with known results and future outcomes with unknown results in order to predict and thus avoid negative effects. Thus, it is not sufficient to merely remember past outcomes, the agent must summarize past experience in a manner informative about the future.
The scenario that I have described here is precisely the state of affairs in the theoretical field of machine learning, where the agents in question are learning algorithms. Substantial progress has been made in understanding the mechanisms of learning in the past decade, and the foundations of learning theory, at least in some simple concept spaces, are quite clear. In the thought experiment that I have constructed, a learning algorithm would behave in exactly the manner a rational individual would, except that its methods for finding similarities between outcomes would be completely statistical and correlational. How would its activity appear to an outside observer?
When initialized, it may or may not have some prior knowledge about the space of outcomes . In the former case (let us call it 'Bayesian Faith') the algorithm's prior knowledge may be inaccurate, in which case, it is going to get knocked around by a lot of negative outcomes until it amends its view of the space to be in conformity with what it observes. In case its prior knowledge is accurate, it will make fewer mistakes and will be relatively more successful than any of its peers so long as it exists in the limited subspace of outcomes for which it possesses strong prior knowledge. In the latter case, which we call 'Popperian Agnosticism', the algorithm will begin from a state of complete ignorance and learn from its mistakes incrementally. While it will take longer than Bayesian Faith in forming its preference relations, it will be able to explore a larger subspace of outcomes and will thereby generalize and predict better. The perpetual trade-off for learning algorithms is one between power and generalizability. Essentially, an algorithm can either choose to have great predictive ability over a set of outcomes that it has a strong hypothesis for, or it can choose to be able to represent all outcomes, but not necessarily be correct in its predictions about them, or be able to interpret and explain them to others.
We now come to the nub: consider a scenario where the space of outcomes can either not be determined by a single preference relation, or the complexity of this preference relation is outside the computational power of individual agents. In several cases, it has been seen that multiple weak learners, allowed to vote using their individual predictions using a simple or weighted majority scheme can learn preference relations well beyond the representational power of each one of them singly. This phenomenon is known in the machine learning community as 'boosting'. Extending our Devil's advocacy further, however, what if the preference relation is beyond the boosted collective learner just as well? Boost the boosters? Hierarchical boosting? It has been shown that such measures will have diminishing returns and that at some point, there will be no advantage gleaned from trying to improve and at that point, the best an algorithm will be able to do will be to make predictions as best it can, with occasional mistakes.
Without entering upon a semantic discussion, I suggest that it is not too outlandish to describe the first stage, where the agent learns using either prior knowledge or through expansive explorations of the space of outcomes as the brahmacharya phase of the agent's existence. Further, the second stage, where the agent (a) combines its preference relations with others in order to make better predictions and (b) uses its learned preference relation to make choices actively outside its domain of learned examples, can be called the grihastha phase of its existence without placing too superogatory an epistemological burden upon either concept.
Where our thought experiment departs from this well-known machine learning framework is in adding the additional constraint of self-preservation to the agents' behavior. Thus, while the unconstrained learning algorithm can settle down into a state of imperfect optimality and continue to function with its best possible preference relation, an algorithm constrained to self-preservation can no longer safely do so under all circumstances. Should the algorithm stop learning or reach an asymptote in performance, the number of mistakes it makes on subsequent decisions stops declining. Given enough time and outcome trials, the agent must necessarily die. Since perception of the viability of self-preservation can be thought of as a function of the magnitude of the agent's fitness, it is natural to see that the agent will have a pre-concept of this eventuality and must do something to ward it off.
At this stage, we must step out from the mathematical confines of learning theory into a more philosophical realm. Certain metaphors and intuitions, however, continue to hold. Assuming that the domain does not allow perfect certainty, the learning agent must either continue to improve its performance or attempt to reduce the number of trials it is making as a function of its life-span. The former case, as several mathematical results show, follows a law of diminishing returns unless the ideal preference relation is within the domain of the learner's capability. This does not exclude it from the possible range of an agent's options, it merely makes it a somewhat inefficient option. The other possibility is to gradually withdraw from the informational game that the space of outcomes plays with the learner, or, at the least, endeavor to do so. This will result in a stage where the agent, now in possession of the optimal hypothesis with respect to its preference relation, can choose to essentially 'play', make random learning excursions across the space of outcomes with little expectation of improvement. The decrease in the prospect of improving upon one's performance will cause a corresponding decrease in the transactional initiatives of the learner. Collective functions learned will gradually decline in value, yet not altogether so; the ones that improve performance significantly will be perpetuated. As a matter of fact, the possibility of sparsification in beneficial collective functions will make it likely that they will be preferred over individual functions learned, since it is in the agent's interests to minimize the number of transactions it makes. For example, if a boosted learner has 20 components and predicts the value of a particular subspace of outcomes with a precision that is almost entirely retained if only 15 components vote every time, effective scheduling will allow all 20 weak learners to make fewer decisions by sitting out rounds in a prescribed order. Again, and this is a major point of emphasis, we claim that this stage will correspond with the vanaprastha phase of the agent's existence.
The final phase, sannyasa, does not fit very well within the learning framework we have devised. Teleological considerations arise and affect the reasoning process (Why do the agents exist? Should the self-perceived cost of the death of an agent tend to zero or infinity as it comes to the end of its existence?) so that the connections to learning theory can no longer be effectively made. Statements made regarding this realm are almost certain to be tinged with ideology, and are hence avoided at this juncture. The only one that we will hazard is the suggestion that, free of philosophical positions, an agent with a mandate for self-preservation, operating under conditions that make its destruction imminent, can no longer derive any self-perceived benefit by operating logically. An open question remains as to whether it should persist in doing so, or attempt to make one of the two extra-logical operations feasible; sublimating either (a) its desire for self-preservation or (b) its perceptual and cognitive facilities that cause it to be affected by the consequences of outcomes. This stage, with its motivationally ambiguous characterization, we assert, may be identified with the sannyasa phase of an agent's existence.
The human condition
Psychiatrists, psychologists and various religious traditions have, at various times and with varying degrees of success, attempted to formulate cardinal principles for describing human behavior and motivations. The fact that all of these hypotheses, several of which are wildly divergent and some of which are quite definitely irrational, have met with substantial success in explaining aspects of human behavior (which is why they survive and flourish) leads one to believe that one compact and rigorous formulation might be as good (or bad) as the next. Here, we suggest that postulating humans as individual and collective agents in our learning experiment leads to a plausible and adequate description of the human condition. We further claim that instantiating these agents with representatives of various belief systems (religious, scientific, cultural) brings forth a large series of social phenomena in an emergent manner, leading us to believe that the isomorphism that we have discovered does actually characterize learning systems meaningfully. We now address each of these claims separately,
Humans as agents
If we posit that humans are primarily driven by the desire for self-preservation and that they do so by interacting with their environment and learning from examples of various outcomes that they encounter, we are claiming no more than a well-established scientific view of the functionality of humans. This part of the argument must, then, pass muster. The sequence of phases that we assigned to the operations of learning algorithms earlier are also claimed to apply here. It is possible, at this stage, to sink into a morass of semantic argument over the incompatibility of the ashram paradigm with some existing system of understanding the human personality. As I suggest above, it is not necessary for different explanatory frameworks to merge, if they make equivalent predictions and have equivalent explanatory power. Representational issues need not (and should not be allowed to) obstruct the flow of reason. Thus, if we are confident that the learning framework is internally consistent, and of the assumption that humans, like all living systems, interact with their environment based on what they learn, then we should be able to proceed further without seeking validation from existing literature. The ashram system was originally developed, as described earlier, to represent stages in human development. From the heuristic description inherent in their definition, it should not require a stretch of credulity to assign the corresponding four phases to humans operating in natural environments.
At this stage, we must make an important distinction. Specifically, the four phases represent behavior of an agent that is rational and more importantly self-aware. Thus, if an agent is oblivious to the existence of its desire for self-preservation, or learning from past negative outcomes, then its behavior will no longer match the expectations of the model. This statement is unfortunate , since it leaves us open to ideological ad hominem attacks (``So Mr Srivastava, not only are you presenting a new science for understanding humanity, you are giving us a new religion into the bargain!''). However, it does follow from the logic of the framework and hence, must be made. An exploration of how the four phases are manifest in people of different psychological persuasions and degrees of self-awareness is a fascinating exercise that I must leave here as an open problem, since discussing it in detail will draw us too far away from our primary arguments. Let me summarize, however, by stating that a framework of information-processing for self-preservation need not be rejected out of hand as a plausible mechanism for understanding human behavior.
Belief systems as agents
We now arrive at the raison d'etre for this essay - the case of the quibbling communities. Predictions about the future of humanity are chiefly predicated on the trajectory of evolution of society, since biological evolution has been rendered almost irrelevant by technological and social innovations. More descriptively, the mating fitness landscape of human existence is now governed predominantly by cultural as opposed to biological factors, which means that biological natural selection is not likely to play a significant role in the scheme of human affairs in times to come. Since memes rather than genes will determine the course of humanity, studies concerning the future of humanity are now dominated by predictions about the interplay between society and technology.
By using belief systems (or communities and/or identities defined based on belief systems) as agents in our learning framework, we recover a process that has, in recent years, been called 'creative destruction', when viewed from the economic paradigm of capitalism. Ideas emerge, if profitable, they are adopted very rapidly until the competitive advantage thereof is eliminated. Ideas that are obsolete are discarded, allowing a socio-economic `survival of the fittest' ecosystem to emerge. Furthermore, fitting the ashram system's four phases to the domain of belief systems can recover Kuhn's idea of progress through the shifting of paradigms. In our case, the traction an agent gains in the space of all agents may be measured by the number of transactions it consummates. As this number declines, the paradigm associated with the belief system the agent represents will gradually fade from view to be supplanted by newer ones.
Let us endeavor to describe the behavior of belief systems in terms of our learning framework. In the first phase, having been initialized with some prior preference relation, the agent will naturally seek to grow through interpreting and learning from social experiments that occur within its purview. The second phase emerges when the agent has reached a critical mass, whereupon it appears in the form of a community with fixed canons and traditions. This is the period when its social utility (measured using number of social transactions) reaches its apex. The third phase commences once the agent's predictive ability begins to reach an asymptote. The ashram intuition would suggest that it gradually decrease its social involvement and expend its existing momentum in exploring nooks and crannies in the edifice of its preference relation already established or in exploring schema of exploration it has heretofore rejected as non-canonical. The crucial element in understanding the subsequent evolution of the agent is in realizing that there is no fixed trajectory that it must necessarily follow. Whereas in the first two phases, a Cartesian detached view of the fitness of the preference relation (with relatively little self-examination) was sufficient for teleological guidance, such objective clarity is no longer optimal behavior for an agent in the vanaprastha phase.
Notice further that, whereas it is not hard to develop an intuition regarding the first two phases of the activities of this class of agent, it is much harder to assess the normative probabilities of its behavior in the third phase. I suggest that this fact is strong corroborative evidence for the assertion that the belief system that we are most familiar with - modern society - is currently at a transition period between the grihastha and vanaprastha phases. Our collective consciousness has not experienced and accepted the vanaprastha phase, rendering the wells of intuition dry in this regard. This is where the abstract theoretical framework I have proposed will assist us in making principled arguments regarding the future, assuming ideal rational behavior on the part of all participants.
The world turns fifty-five
It is perhaps presumptive for one who is less than half that age to assess the existential circumstances and motivations of a fifty-five year old human agent that has successfully navigated its local information domain. Unfortunately, that is precisely what I must now do, thankfully in very general terms. He has developed a strong and well-generalized preference relation and is associated with several boosted communities of agents with varying degrees of commitment. He is likely to be past his prime in terms of economic and intellectual productivity, but will retain much power and influence in the relevant communities nevertheless by virtue of his past performance. Beyond this point, he has several options, all relatively tranquil and all involving a gradual unwinding of his material transactions and a withdrawal into introspective, playful and abstract activities.
Based on the learning hypothesis, I postulate that, among others, the belief system of natural materialism has arrived at a congruent existential state. This, I further claim, manifests itself primarily through the rise to prominence of the law of diminishing returns in two essential domains (a) the social benefit of scientific inquiry and (b) the social benefit of economic productivity. From the description of the vanaprastha phase of the learning framework, it follows ineluctably that the corresponding belief systems will decline in social utility as they enter this stage of existence. Let us now briefly examine each of the two subsidiary claims in light of the physical evidence.
The first claim - decline in the social benefit of scientific inquiry - is bound to ruffle feathers if analyzed deductively. I will, therefore, merely present an empirical argument and pass lightly on. It is a widely held view that economic pricing is a good approximation of social utility in the limit of efficient markets and freely transferable resources. Assuming this to be true (or partially true), let us view the following scenario objectively: the Clay Mathematical Institute has offered a prize of $1 million each for the solution of seven of the most significant mathematical problems extant. These include the recently proved Poincare Conjecture, the metaphysically profound P=NP problem, the incredibly deep Riemann hypothesis and the fiendishly intractable Navier-Stokes analysis. Each one of these, of itself, addresses a foundational issue in our understanding of the physical world and the models we use to describe it. Realize, however, that a mathematician who, by dint of genius and labor, solves all of them and expands the frontiers of our knowledge by an unimaginable quantum would win a purse of $7 million in all (We ignore for the moment, any economic benefit he may derive by assuming that the level of mathematical abstraction required to solve these problems would render the mathematician incapable of functioning very efficiently where his economic self-interest is concerned). Realize further, that the newborn offspring of Brad Pitt and Angelina Jolie were paid twice this amount ($14 million) for the privilege of publishing their first photographs. Without wishing to pontificate in any way, we merely observe that a society that prices its social transactions in such a manner demonstrates a rather low probability of setting great store by scientific inquiry!
We turn now to the second claim - decline in the social benefit of economic productivity. This is a subject of intense debate in the current political climate, where environmental sustainability concerns are finally being allowed a hearing after nearly a century of suppression in favor of capitalistic fervor and economies of scale. The lack of correlation of income with happiness beyond a threshold value and the high happiness rates in some countries with low per capita income are observations that have been made and commented upon by several authors far more erudite than I. I will merely state, therefore, that the claim has significant support from a broad spectrum of empirical evidence, social activism and environmental sustainability theories. In point of fact, with the vicissitudes of the financial markets of the world in the past year or so, mainstream acceptance of the implausibility of sustaining economic growth is growing and alternative philosophies of consumption and living are being examined with wider interest than ever before. The realization that it is important to treat non-renewable natural materials as capital instead of resource has presented much cause for deliberation in public policy in the Western world in recent years. It is not that the circumstances that the modern economy faces are unforeseen and unpredictable, it is simply that the theories that explain economic behavior in such conditions as have now emerged (globalization in an unequal world causing lop-sided supply-demand, decline in availability of natural materials) have not found traction in the past. In short, while there is nothing radically new in the idea that the world's economic models must change to adapt to the change in their environment, the change, in itself, is radical in terms of its consequences upon how people behave and the things that (should) motivate them. It is important to raise this distinction for the following reason: one of the chief responses of capitalistic thought to sustainability theories is the claim that they are proposing nothing new, equating value with originality. The assumption, which follows from the efficient market hypothesis, is that the theory would have been adopted earlier if it had any value. Since it has not, it need not. It is important, therefore, to point out that change in the economic environment represents a change in the space of outcomes, rendering the efficient markets argument and hence the appeal to tradition irrelevant. We thus conclude that there is substantial reason to believe that our second claim is correct.
Validating our two claims in turn suggests the validity of our primary claim, that the belief system of natural materialism, inasmuch as it has governed the manner of thinking of industrial society, is past its prime. Again, there is no originality in this claim, it has been made before and with great vigor by proponents of ecological, environmental, primitivistic and anarchistic points of view. Our contribution lies merely in showing that such behavior is entirely expected of all systems that grow through learning from responses to their environment, and that modern society falls firmly within this framework. Thus, appeals to the glorious future of humanity must be tempered with an acknowledgement of the reality of its ephemeral character compared with the length and time-scales of existence in general.
Just as it is difficult for adolescents to appreciate the existential worries that affect middle age, it is difficult for minds indoctrinated with heuristics designed for the grihastha phase of social existence to attach much value to vanaprastha ruminations. The pop psychology expression for this phenomenon is `mid-life crisis', a mismatch between the desire for continued social utility and an awareness of its diminishing prospects. At this stage, there are several methods of handling one's predicament. One used at times is denial: witness the perpetual quest for immortality, now occasionally modified and expressed as a desire for mind-uploading. Another, more common, is dispersal of identity; the understanding that those who will come later, whether in one's genetic progression or another's will experience and learn what oneself could not. The reasons for this strategy can be explored further in our framework, but are passed over here in the interests of the continued cogency of our main argument. A third attitude, and one seen more frequently in the contemplative cultures of the East than in the extroverted milieu of industrial society is that of explicit vanaprastha, a conscious gradual disengagement from social transactions. The first two mechanisms, while not explicitly vanaprastha in nature, eventually follow in its stead, or are countermanded by external factors and forced along in its general direction.
Conclusion
A large proportion of writing that deals with the future delves into action imperatives and eventually deteriorates into genteel ideological brinkmanship. In this essay we have scrupulously steered clear of imposing any ideological preferences upon the choices that belief systems have made in the past or will make in the future. The primitivist view expressed in support of the claim of diminishing returns on scientific and economic endeavor is an exception in this regard. We qualify its use with the argument that this is indeed a prominent view in the public policy debates of the moment, so notwithstanding its ultimate validity, it remains an accepted alternative point of view. Apart from this, I have tried to keep this exposition free of my own preference relations, in order to address the matter in as general a manner as possible.
The necessity of striving to do so is further accentuated by our conclusion that the belief systems that we have come to accept as normal are gradually transitioning, or are at a stage where there is no logical reason why they need not transition, into an ideologically ambiguous vanaprastha state of low social involvement. Recall again that this does not mean that their activity is diminishing, it means that the social utility of their activity is decreasing, causing society to seek out other paradigms of thought that affect to carry greater social utility. Since much of humanity's existence is now inextricably woven up in the fabric of the existing system, it is essential that a transition, should it occur (or be occurring), be made with a degree of awareness as to its antecedents and possible consequences. Since it is a difficult task to view the evolution of belief systems shorn of individual and collective preference relations, it becomes all the more important to develop a neutral abstract framework to facilitate discussions on such matters. Should my thesis pass muster, I feel that it will make it simpler to understand and interpret the existential choices faced by our unique and complex industrial social structure.
There are two important general conclusions that I draw from this exercise. One is the understanding that deciding the future course of action for humanity is as free and as constrained as the set of choices a hypothetical gentleman approaching his golden years faces. Almost all prospects which avoid needless conflict and stress are pleasant, but it is advisable to choose soon and choose wisely. This is not a particularly insightful observation on face value, but appears to negate the possibility of discovering moral imperatives in favor of any one futuristic paradigm. The second conclusion, if the ashram framework be considered to have some epistemological validity, lies in the observation that both extraverted (technology-driven) and intraverted (sustainability-driven) modes of progress hereon will eventually lead to an equivalent state of dispersal of human identity. While one side of the technology-simplicity argument traditionally derides the other's vision of Utopia as a world of 'soulless automata', the other responds in kind by equating the goal of the former to be the domestication of the human species by stamping out its spark of innovation. In either scenario, loss of identity, the sannyasa phase must necessarily ensue. The same argument can be made for almost every other futuristic scheme (eudaimonia, thought control, cybernetic society) within foresight at the moment. The only difference between all these paradigms, insofar as I can judge, lies in the degree of inward exploration and self-awareness that the respective belief systems will choose to motivate among their adherents. That is the question predictors of the future should be looking to resolve, with the awareness that multiple answers will, in material terms at least, eventually find the same solution.
As an avid reader of science fiction during my younger days, I have often remarked at the pedestrian nature of the concepts and gadgetry (though not the literary value) deployed in contemporary science fiction writing. As our understanding of the near impossibility of human extra-solar space travel has gradually consolidated, and as advances in scientific discoveries have declined in cultural significance, writers increasingly have to look to realms of fantasy or sociology to construct their story-lines. An unfortunate consequence of this shift is that the delighful art of peering into the future has been decoupled from artistic imagination and is now the domain of sober scientific speculation either through fitting curves to empirical evidence from the modern industrial period or through seeking predictive power in the present by devising sociological theories of history.
To me, either methodology appears to result in an unfortunate narrowing of focus. In the first case, the predictors tend to be stalwarts of the avant garde of technological innovation (Kurzweil, Joy, Moravec etc), who look upon evolution as a best-case scenario, i.e., they operate on the premise that once technology has been developed, nothing can stop it from being adopted. Their statistical analysis of 50 years of technological discovery leads them to believe that advances will continue to accelerate, leading, in the next few decades, to a situation where, depending on the degree of sanguinity of the predictor (a) scarcity and the need to work will have been eliminated or (b) humans will have been marginalized or exterminated. I suggest that these observers suffer from a sampling bias, i.e., they consider human history in the period before 1800 to have made no contribution to ensuing developments. This arises from the assumption that progress must be physically present and demonstrable to count. A second fallacy lies in the assumption that technological feasibility will inevitably cause adoption. A stark counterexample is the fate of the space program, which has seen no progress in the last three decades through the lack of social incentive. I feel that their predictions, while quite specific, are unlikely to be accurate, since they operate entirely using inductive reasoning and cannot account for feedback effects and other social variables. In the second case, that of sociological analyses, I feel that these observers suffer from the opposite problem, i.e., their models are not flexible enough to be able to make good predictions about the volatile period of human existence that we are living through. While observations regarding social disruption, anomie and commentaries on crime rates, urbanization etc are certainly apposite, these analyses do not have predictive power, since they are based entirely on analogous as opposed to inductive reasoning.
I title this essay `Cycles of life' because the basic thesis herein is a characterization of four stages of existential behavior in all systems that learn to make fewer errors in prediction based on their interactions with their environment. This framework, I then claim, presents a good description not just of an individual human's actions while in existence, but also those of a society's and lastly humanity's. I assign to these four phases of existence names assigned in the Indian Classical system of ashrams since the original system emerges as a special case of the more general model I propose. I first define the ashram framework and show characteristics of systems to which it will apply. Then, I examine the degrees to which it may be applied to entities that we recognize well - the liberal democratic Western culture, the scientific enterprise, less developed societies in other parts of the world, and finally humanity as a whole. By hypothesizing and providing evidence for a parallel between the existential trajectories of humans and societies it becomes possible to draw specific conclusions about the future of humanity in a heuristic and socially interpretable manner, as opposed to empirical methods which, as described above, make severe and dehumanizing assumptions about cultural homogeneity into the future. Furthermore, our mathematical formulation of the generative process affords our model the rigor and predictive ability that sociological approaches tend to lack.
A meta-comment before we begin: I am continually perplexed by what I formulate to myself as the 'dilemma of citation'. It is hard for me to keep track of the observations of technologists, social scientists and philosophers since my reading generally follows Brownian motion rather than a linear scholarly process. At the very outset, therefore, I disclaim all pretensions to originality and virtue in this work and apologize to those whose ideas I may appear to plagiarize throughout this exposition. Since I stand to gain nothing save some emotional satisfaction from the writing, I hereby consider myself to be ethically absolved.
The ashrams of Manu
The Manu Smriti is a Sanskrit text dated to the 2nd century B.C. that contains social commentaries and injunctions as preached (if not necessarily practiced) in India towards the end of the Maurya empire. It has attracted considerable controversy, since it defined and was used as spiritual sanction for the imposition of the infamous caste system. While post-colonially, it is possible to see that the practice was much caricatured by Western interpreters to increment the white man's putative burden, the fact remains that the Manu Smriti remains a controversial reference. Therefore, I wish to clarify that I borrow no ideology from the text, only some convenient nomenclature.
The Manu Smriti divides a man's life into four stages. The first, brahmacharya, is associated with the life of a student: minimal transactional involvement with society, exploratory expeditions and vocational and intellectual learning. The second, grihastha, is identified with the life of a man of the world: extreme transactional involvement, well-established social networks and relationships, sense of duty, responsibility and exaltation of the spirit of hard work. The third, vanaprastha, is identified with the lifestyle of a wanderer: great freedom of movement, lowered sense of transactional responsibility, greater social egalitarianism, increased affinity for simplicity. The fourth stage is called sannyasa, the life of the renunciate, which reflects a complete end to transactional as well as social involvement and a gradual dispersal of the sense of identity, preparatory to its physical cessation. The hypothesis of this essay is that the existential evolution of humans - both as individuals and as communities - proceeds along these broad stages of development. I further claim that this process is characteristic of systems that 'live', for reasons that I discuss next.
To see how the four-fold trajectory is an extremely plausible mechanism for the evolution of living systems, it becomes necessary for us to conduct a gedankenexperiment motivated by computational learning theory. Consider a space of possible outcomes, each of which can affect an agent that chooses them positively or negatively. The manner (positive or negative) in which the agent is affected is governed by the agent's preference relation, which is essentially a collection of statements that the agent stores in its memory to allow it to make choices intelligently. For example, the preference relation 'if hair on fire, then effect of going to movies strongly negative' will allow an agent to choose wisely between going to the movies and finding a bucket of water. It is not known whether all agents are initialized with the same preference relation, but the question is moot, since at every instant in an agent's existence span, it makes choices which affect its preference relation. Thus, it is almost impossible to find two agents with the same preference relation.
Now, if an agent is to survive, it is necessary for it to learn to avoid negative outcomes, a process that is accomplished using both prior knowledge as well as experience of outcomes. The greatest ability for an agent is to be able to generalize, that is, find similarities and analogies between past outcomes with known results and future outcomes with unknown results in order to predict and thus avoid negative effects. Thus, it is not sufficient to merely remember past outcomes, the agent must summarize past experience in a manner informative about the future.
The scenario that I have described here is precisely the state of affairs in the theoretical field of machine learning, where the agents in question are learning algorithms. Substantial progress has been made in understanding the mechanisms of learning in the past decade, and the foundations of learning theory, at least in some simple concept spaces, are quite clear. In the thought experiment that I have constructed, a learning algorithm would behave in exactly the manner a rational individual would, except that its methods for finding similarities between outcomes would be completely statistical and correlational. How would its activity appear to an outside observer?
When initialized, it may or may not have some prior knowledge about the space of outcomes . In the former case (let us call it 'Bayesian Faith') the algorithm's prior knowledge may be inaccurate, in which case, it is going to get knocked around by a lot of negative outcomes until it amends its view of the space to be in conformity with what it observes. In case its prior knowledge is accurate, it will make fewer mistakes and will be relatively more successful than any of its peers so long as it exists in the limited subspace of outcomes for which it possesses strong prior knowledge. In the latter case, which we call 'Popperian Agnosticism', the algorithm will begin from a state of complete ignorance and learn from its mistakes incrementally. While it will take longer than Bayesian Faith in forming its preference relations, it will be able to explore a larger subspace of outcomes and will thereby generalize and predict better. The perpetual trade-off for learning algorithms is one between power and generalizability. Essentially, an algorithm can either choose to have great predictive ability over a set of outcomes that it has a strong hypothesis for, or it can choose to be able to represent all outcomes, but not necessarily be correct in its predictions about them, or be able to interpret and explain them to others.
We now come to the nub: consider a scenario where the space of outcomes can either not be determined by a single preference relation, or the complexity of this preference relation is outside the computational power of individual agents. In several cases, it has been seen that multiple weak learners, allowed to vote using their individual predictions using a simple or weighted majority scheme can learn preference relations well beyond the representational power of each one of them singly. This phenomenon is known in the machine learning community as 'boosting'. Extending our Devil's advocacy further, however, what if the preference relation is beyond the boosted collective learner just as well? Boost the boosters? Hierarchical boosting? It has been shown that such measures will have diminishing returns and that at some point, there will be no advantage gleaned from trying to improve and at that point, the best an algorithm will be able to do will be to make predictions as best it can, with occasional mistakes.
Without entering upon a semantic discussion, I suggest that it is not too outlandish to describe the first stage, where the agent learns using either prior knowledge or through expansive explorations of the space of outcomes as the brahmacharya phase of the agent's existence. Further, the second stage, where the agent (a) combines its preference relations with others in order to make better predictions and (b) uses its learned preference relation to make choices actively outside its domain of learned examples, can be called the grihastha phase of its existence without placing too superogatory an epistemological burden upon either concept.
Where our thought experiment departs from this well-known machine learning framework is in adding the additional constraint of self-preservation to the agents' behavior. Thus, while the unconstrained learning algorithm can settle down into a state of imperfect optimality and continue to function with its best possible preference relation, an algorithm constrained to self-preservation can no longer safely do so under all circumstances. Should the algorithm stop learning or reach an asymptote in performance, the number of mistakes it makes on subsequent decisions stops declining. Given enough time and outcome trials, the agent must necessarily die. Since perception of the viability of self-preservation can be thought of as a function of the magnitude of the agent's fitness, it is natural to see that the agent will have a pre-concept of this eventuality and must do something to ward it off.
At this stage, we must step out from the mathematical confines of learning theory into a more philosophical realm. Certain metaphors and intuitions, however, continue to hold. Assuming that the domain does not allow perfect certainty, the learning agent must either continue to improve its performance or attempt to reduce the number of trials it is making as a function of its life-span. The former case, as several mathematical results show, follows a law of diminishing returns unless the ideal preference relation is within the domain of the learner's capability. This does not exclude it from the possible range of an agent's options, it merely makes it a somewhat inefficient option. The other possibility is to gradually withdraw from the informational game that the space of outcomes plays with the learner, or, at the least, endeavor to do so. This will result in a stage where the agent, now in possession of the optimal hypothesis with respect to its preference relation, can choose to essentially 'play', make random learning excursions across the space of outcomes with little expectation of improvement. The decrease in the prospect of improving upon one's performance will cause a corresponding decrease in the transactional initiatives of the learner. Collective functions learned will gradually decline in value, yet not altogether so; the ones that improve performance significantly will be perpetuated. As a matter of fact, the possibility of sparsification in beneficial collective functions will make it likely that they will be preferred over individual functions learned, since it is in the agent's interests to minimize the number of transactions it makes. For example, if a boosted learner has 20 components and predicts the value of a particular subspace of outcomes with a precision that is almost entirely retained if only 15 components vote every time, effective scheduling will allow all 20 weak learners to make fewer decisions by sitting out rounds in a prescribed order. Again, and this is a major point of emphasis, we claim that this stage will correspond with the vanaprastha phase of the agent's existence.
The final phase, sannyasa, does not fit very well within the learning framework we have devised. Teleological considerations arise and affect the reasoning process (Why do the agents exist? Should the self-perceived cost of the death of an agent tend to zero or infinity as it comes to the end of its existence?) so that the connections to learning theory can no longer be effectively made. Statements made regarding this realm are almost certain to be tinged with ideology, and are hence avoided at this juncture. The only one that we will hazard is the suggestion that, free of philosophical positions, an agent with a mandate for self-preservation, operating under conditions that make its destruction imminent, can no longer derive any self-perceived benefit by operating logically. An open question remains as to whether it should persist in doing so, or attempt to make one of the two extra-logical operations feasible; sublimating either (a) its desire for self-preservation or (b) its perceptual and cognitive facilities that cause it to be affected by the consequences of outcomes. This stage, with its motivationally ambiguous characterization, we assert, may be identified with the sannyasa phase of an agent's existence.
The human condition
Psychiatrists, psychologists and various religious traditions have, at various times and with varying degrees of success, attempted to formulate cardinal principles for describing human behavior and motivations. The fact that all of these hypotheses, several of which are wildly divergent and some of which are quite definitely irrational, have met with substantial success in explaining aspects of human behavior (which is why they survive and flourish) leads one to believe that one compact and rigorous formulation might be as good (or bad) as the next. Here, we suggest that postulating humans as individual and collective agents in our learning experiment leads to a plausible and adequate description of the human condition. We further claim that instantiating these agents with representatives of various belief systems (religious, scientific, cultural) brings forth a large series of social phenomena in an emergent manner, leading us to believe that the isomorphism that we have discovered does actually characterize learning systems meaningfully. We now address each of these claims separately,
Humans as agents
If we posit that humans are primarily driven by the desire for self-preservation and that they do so by interacting with their environment and learning from examples of various outcomes that they encounter, we are claiming no more than a well-established scientific view of the functionality of humans. This part of the argument must, then, pass muster. The sequence of phases that we assigned to the operations of learning algorithms earlier are also claimed to apply here. It is possible, at this stage, to sink into a morass of semantic argument over the incompatibility of the ashram paradigm with some existing system of understanding the human personality. As I suggest above, it is not necessary for different explanatory frameworks to merge, if they make equivalent predictions and have equivalent explanatory power. Representational issues need not (and should not be allowed to) obstruct the flow of reason. Thus, if we are confident that the learning framework is internally consistent, and of the assumption that humans, like all living systems, interact with their environment based on what they learn, then we should be able to proceed further without seeking validation from existing literature. The ashram system was originally developed, as described earlier, to represent stages in human development. From the heuristic description inherent in their definition, it should not require a stretch of credulity to assign the corresponding four phases to humans operating in natural environments.
At this stage, we must make an important distinction. Specifically, the four phases represent behavior of an agent that is rational and more importantly self-aware. Thus, if an agent is oblivious to the existence of its desire for self-preservation, or learning from past negative outcomes, then its behavior will no longer match the expectations of the model. This statement is unfortunate , since it leaves us open to ideological ad hominem attacks (``So Mr Srivastava, not only are you presenting a new science for understanding humanity, you are giving us a new religion into the bargain!''). However, it does follow from the logic of the framework and hence, must be made. An exploration of how the four phases are manifest in people of different psychological persuasions and degrees of self-awareness is a fascinating exercise that I must leave here as an open problem, since discussing it in detail will draw us too far away from our primary arguments. Let me summarize, however, by stating that a framework of information-processing for self-preservation need not be rejected out of hand as a plausible mechanism for understanding human behavior.
Belief systems as agents
We now arrive at the raison d'etre for this essay - the case of the quibbling communities. Predictions about the future of humanity are chiefly predicated on the trajectory of evolution of society, since biological evolution has been rendered almost irrelevant by technological and social innovations. More descriptively, the mating fitness landscape of human existence is now governed predominantly by cultural as opposed to biological factors, which means that biological natural selection is not likely to play a significant role in the scheme of human affairs in times to come. Since memes rather than genes will determine the course of humanity, studies concerning the future of humanity are now dominated by predictions about the interplay between society and technology.
By using belief systems (or communities and/or identities defined based on belief systems) as agents in our learning framework, we recover a process that has, in recent years, been called 'creative destruction', when viewed from the economic paradigm of capitalism. Ideas emerge, if profitable, they are adopted very rapidly until the competitive advantage thereof is eliminated. Ideas that are obsolete are discarded, allowing a socio-economic `survival of the fittest' ecosystem to emerge. Furthermore, fitting the ashram system's four phases to the domain of belief systems can recover Kuhn's idea of progress through the shifting of paradigms. In our case, the traction an agent gains in the space of all agents may be measured by the number of transactions it consummates. As this number declines, the paradigm associated with the belief system the agent represents will gradually fade from view to be supplanted by newer ones.
Let us endeavor to describe the behavior of belief systems in terms of our learning framework. In the first phase, having been initialized with some prior preference relation, the agent will naturally seek to grow through interpreting and learning from social experiments that occur within its purview. The second phase emerges when the agent has reached a critical mass, whereupon it appears in the form of a community with fixed canons and traditions. This is the period when its social utility (measured using number of social transactions) reaches its apex. The third phase commences once the agent's predictive ability begins to reach an asymptote. The ashram intuition would suggest that it gradually decrease its social involvement and expend its existing momentum in exploring nooks and crannies in the edifice of its preference relation already established or in exploring schema of exploration it has heretofore rejected as non-canonical. The crucial element in understanding the subsequent evolution of the agent is in realizing that there is no fixed trajectory that it must necessarily follow. Whereas in the first two phases, a Cartesian detached view of the fitness of the preference relation (with relatively little self-examination) was sufficient for teleological guidance, such objective clarity is no longer optimal behavior for an agent in the vanaprastha phase.
Notice further that, whereas it is not hard to develop an intuition regarding the first two phases of the activities of this class of agent, it is much harder to assess the normative probabilities of its behavior in the third phase. I suggest that this fact is strong corroborative evidence for the assertion that the belief system that we are most familiar with - modern society - is currently at a transition period between the grihastha and vanaprastha phases. Our collective consciousness has not experienced and accepted the vanaprastha phase, rendering the wells of intuition dry in this regard. This is where the abstract theoretical framework I have proposed will assist us in making principled arguments regarding the future, assuming ideal rational behavior on the part of all participants.
The world turns fifty-five
It is perhaps presumptive for one who is less than half that age to assess the existential circumstances and motivations of a fifty-five year old human agent that has successfully navigated its local information domain. Unfortunately, that is precisely what I must now do, thankfully in very general terms. He has developed a strong and well-generalized preference relation and is associated with several boosted communities of agents with varying degrees of commitment. He is likely to be past his prime in terms of economic and intellectual productivity, but will retain much power and influence in the relevant communities nevertheless by virtue of his past performance. Beyond this point, he has several options, all relatively tranquil and all involving a gradual unwinding of his material transactions and a withdrawal into introspective, playful and abstract activities.
Based on the learning hypothesis, I postulate that, among others, the belief system of natural materialism has arrived at a congruent existential state. This, I further claim, manifests itself primarily through the rise to prominence of the law of diminishing returns in two essential domains (a) the social benefit of scientific inquiry and (b) the social benefit of economic productivity. From the description of the vanaprastha phase of the learning framework, it follows ineluctably that the corresponding belief systems will decline in social utility as they enter this stage of existence. Let us now briefly examine each of the two subsidiary claims in light of the physical evidence.
The first claim - decline in the social benefit of scientific inquiry - is bound to ruffle feathers if analyzed deductively. I will, therefore, merely present an empirical argument and pass lightly on. It is a widely held view that economic pricing is a good approximation of social utility in the limit of efficient markets and freely transferable resources. Assuming this to be true (or partially true), let us view the following scenario objectively: the Clay Mathematical Institute has offered a prize of $1 million each for the solution of seven of the most significant mathematical problems extant. These include the recently proved Poincare Conjecture, the metaphysically profound P=NP problem, the incredibly deep Riemann hypothesis and the fiendishly intractable Navier-Stokes analysis. Each one of these, of itself, addresses a foundational issue in our understanding of the physical world and the models we use to describe it. Realize, however, that a mathematician who, by dint of genius and labor, solves all of them and expands the frontiers of our knowledge by an unimaginable quantum would win a purse of $7 million in all (We ignore for the moment, any economic benefit he may derive by assuming that the level of mathematical abstraction required to solve these problems would render the mathematician incapable of functioning very efficiently where his economic self-interest is concerned). Realize further, that the newborn offspring of Brad Pitt and Angelina Jolie were paid twice this amount ($14 million) for the privilege of publishing their first photographs. Without wishing to pontificate in any way, we merely observe that a society that prices its social transactions in such a manner demonstrates a rather low probability of setting great store by scientific inquiry!
We turn now to the second claim - decline in the social benefit of economic productivity. This is a subject of intense debate in the current political climate, where environmental sustainability concerns are finally being allowed a hearing after nearly a century of suppression in favor of capitalistic fervor and economies of scale. The lack of correlation of income with happiness beyond a threshold value and the high happiness rates in some countries with low per capita income are observations that have been made and commented upon by several authors far more erudite than I. I will merely state, therefore, that the claim has significant support from a broad spectrum of empirical evidence, social activism and environmental sustainability theories. In point of fact, with the vicissitudes of the financial markets of the world in the past year or so, mainstream acceptance of the implausibility of sustaining economic growth is growing and alternative philosophies of consumption and living are being examined with wider interest than ever before. The realization that it is important to treat non-renewable natural materials as capital instead of resource has presented much cause for deliberation in public policy in the Western world in recent years. It is not that the circumstances that the modern economy faces are unforeseen and unpredictable, it is simply that the theories that explain economic behavior in such conditions as have now emerged (globalization in an unequal world causing lop-sided supply-demand, decline in availability of natural materials) have not found traction in the past. In short, while there is nothing radically new in the idea that the world's economic models must change to adapt to the change in their environment, the change, in itself, is radical in terms of its consequences upon how people behave and the things that (should) motivate them. It is important to raise this distinction for the following reason: one of the chief responses of capitalistic thought to sustainability theories is the claim that they are proposing nothing new, equating value with originality. The assumption, which follows from the efficient market hypothesis, is that the theory would have been adopted earlier if it had any value. Since it has not, it need not. It is important, therefore, to point out that change in the economic environment represents a change in the space of outcomes, rendering the efficient markets argument and hence the appeal to tradition irrelevant. We thus conclude that there is substantial reason to believe that our second claim is correct.
Validating our two claims in turn suggests the validity of our primary claim, that the belief system of natural materialism, inasmuch as it has governed the manner of thinking of industrial society, is past its prime. Again, there is no originality in this claim, it has been made before and with great vigor by proponents of ecological, environmental, primitivistic and anarchistic points of view. Our contribution lies merely in showing that such behavior is entirely expected of all systems that grow through learning from responses to their environment, and that modern society falls firmly within this framework. Thus, appeals to the glorious future of humanity must be tempered with an acknowledgement of the reality of its ephemeral character compared with the length and time-scales of existence in general.
Just as it is difficult for adolescents to appreciate the existential worries that affect middle age, it is difficult for minds indoctrinated with heuristics designed for the grihastha phase of social existence to attach much value to vanaprastha ruminations. The pop psychology expression for this phenomenon is `mid-life crisis', a mismatch between the desire for continued social utility and an awareness of its diminishing prospects. At this stage, there are several methods of handling one's predicament. One used at times is denial: witness the perpetual quest for immortality, now occasionally modified and expressed as a desire for mind-uploading. Another, more common, is dispersal of identity; the understanding that those who will come later, whether in one's genetic progression or another's will experience and learn what oneself could not. The reasons for this strategy can be explored further in our framework, but are passed over here in the interests of the continued cogency of our main argument. A third attitude, and one seen more frequently in the contemplative cultures of the East than in the extroverted milieu of industrial society is that of explicit vanaprastha, a conscious gradual disengagement from social transactions. The first two mechanisms, while not explicitly vanaprastha in nature, eventually follow in its stead, or are countermanded by external factors and forced along in its general direction.
Conclusion
A large proportion of writing that deals with the future delves into action imperatives and eventually deteriorates into genteel ideological brinkmanship. In this essay we have scrupulously steered clear of imposing any ideological preferences upon the choices that belief systems have made in the past or will make in the future. The primitivist view expressed in support of the claim of diminishing returns on scientific and economic endeavor is an exception in this regard. We qualify its use with the argument that this is indeed a prominent view in the public policy debates of the moment, so notwithstanding its ultimate validity, it remains an accepted alternative point of view. Apart from this, I have tried to keep this exposition free of my own preference relations, in order to address the matter in as general a manner as possible.
The necessity of striving to do so is further accentuated by our conclusion that the belief systems that we have come to accept as normal are gradually transitioning, or are at a stage where there is no logical reason why they need not transition, into an ideologically ambiguous vanaprastha state of low social involvement. Recall again that this does not mean that their activity is diminishing, it means that the social utility of their activity is decreasing, causing society to seek out other paradigms of thought that affect to carry greater social utility. Since much of humanity's existence is now inextricably woven up in the fabric of the existing system, it is essential that a transition, should it occur (or be occurring), be made with a degree of awareness as to its antecedents and possible consequences. Since it is a difficult task to view the evolution of belief systems shorn of individual and collective preference relations, it becomes all the more important to develop a neutral abstract framework to facilitate discussions on such matters. Should my thesis pass muster, I feel that it will make it simpler to understand and interpret the existential choices faced by our unique and complex industrial social structure.
There are two important general conclusions that I draw from this exercise. One is the understanding that deciding the future course of action for humanity is as free and as constrained as the set of choices a hypothetical gentleman approaching his golden years faces. Almost all prospects which avoid needless conflict and stress are pleasant, but it is advisable to choose soon and choose wisely. This is not a particularly insightful observation on face value, but appears to negate the possibility of discovering moral imperatives in favor of any one futuristic paradigm. The second conclusion, if the ashram framework be considered to have some epistemological validity, lies in the observation that both extraverted (technology-driven) and intraverted (sustainability-driven) modes of progress hereon will eventually lead to an equivalent state of dispersal of human identity. While one side of the technology-simplicity argument traditionally derides the other's vision of Utopia as a world of 'soulless automata', the other responds in kind by equating the goal of the former to be the domestication of the human species by stamping out its spark of innovation. In either scenario, loss of identity, the sannyasa phase must necessarily ensue. The same argument can be made for almost every other futuristic scheme (eudaimonia, thought control, cybernetic society) within foresight at the moment. The only difference between all these paradigms, insofar as I can judge, lies in the degree of inward exploration and self-awareness that the respective belief systems will choose to motivate among their adherents. That is the question predictors of the future should be looking to resolve, with the awareness that multiple answers will, in material terms at least, eventually find the same solution.
