Recent advances in the sciences have in no small part been made possible due to the use of high-capacity computers to run simulations on all the possible outcomes and scenarios resulting from the input data. For instance, in the life sciences, simulation of the functioning of microscopic entities like cells, proteins, amoebas, bacteria, algae, larvae, etc. has proven to be an effective tool for research.
Simulations have long been employed in the social sciences, humanities, and the arts as training and brainstorming devices. Role-playing exercises in workshops and stakeholder consultations are nothing but simulations run by groups of people. Computerised social simulations have also gained some traction in the social sciences. Data sets obtained from the study of social problems like traffic, housing, earthquakes, weather disasters, the spread of diseases, etc. are used for modeling solutions. Agent-based social simulations are most popular in studies on particular social problems and the outcome of corresponding policies, proposed or in effect. Flight simulators are the stock examples in any introduction to computerised simulations. AI-assisted VR-AR medical simulations are predicted to be the next big leap in medical technology. Simulations of chickens fighting dinosaurs have long been a running joke in online subcultures.
Considering our ability to run such computerised simulations, what is lacking is simulations of social systems themselves. Of course, there are stand-alone examples of structural simulations, but I am hinting at something more institutionalised. For the purpose of this discussion, by simulating social systems I simply mean creating computer-generated models of society to obtain answers to the following questions -
1)what are the potential ways in which existent technology can guide the future course of human society,
2)which of these will be the best ways forward,
3)and how to go about manifesting them in reality.
It is all well and good to trace current social issues with respect to historical events and turns in government policy. But this creates endless streams of data trapped in the appendices of reports and policy recommendations that get partially or wholly challenged when a government of a different ideological inclination comes to power and commissions its own set of reports and consultations. What might provide conclusive insight into the best future course of action is running simulations on all social problems simultaneously, with all the available data. This is required because running simulations on a particular social problem or a set of interrelated social problems gives us little to no idea about how to design social systems that are devoid of such problems to begin with. System-wide simulations may solve this. Is social democracy the best form of governance available? Do societies outgrow the need for social democracy and necessarily need minimal government for their prosperity? What are the factors and events that create such societies? Simulations at the scale of the world may be useful in providing so many answers to these questions that truly pluralist clarity may be obtained.
Most of the existing instances of using such mechanisms for social simulations are too tied up in highly conditional or microcosmic frameworks instead of looking at international social systems as a whole. For instance, the limitation of agent-based simulations is that they end up imagining the solutions to social problems by focusing on the individual units and groups that make up social systems. This strengthens the oft-invoked illusion that problems in social systems can always/only be eliminated by the conscientious and informed decisions of citizens if we incentivise this or that behavioural pattern. A limited simulation when a larger one is possible is unethical and opportunistic because it is designed to provide solutions that are acceptable to the prevailing social order on social problems it has consented to address. What I am trying to recommend is a simulation that can fault the system when no solutions exist in terms of constituent agents, as is possible in qualitative structural critiques.
Importantly, we need an institutional mechanism for running such simulations, built into our social systems. While the infrastructure for running large, comprehensive simulations exists, it is controlled by secretive military agencies or for-profit corporations. In theory, any enterprising social scientist could make use of complex algorithmic mechanisms on a cloud computing platform. But the problem is not merely of access to infrastructure but also one of scale. Any publicly available infrastructure is definitely not adequate for a system-wide simulation, let alone for the comparative analyses of multiple such simulations. But machines capable of performing such operations do exist, the issue is what uses they are put to.
There is also the problem of design. The possible smaller-scale substitutes available in the public domain, like cloud computing services, are easily available anywhere through the internet and some even have free basic plans with limited access (though for using these platforms for building a data model of any complexity one would need to subscribe to the higher payment plans). But these services are invariably designed as business management and market analysis tools. They generate revenue by offering businesses an outsourced IT plan, eliminating the need and cost for a permanent IT team on the payrolls, or significantly reducing those costs. This impacts the usability of said platforms. Most often tracking market trends and individual behaviours (of consumers, clients, employees, stakeholders, etc.) are the processes that the original, bare-bones design of the algorithm is geared towards. At best, some such softwares can be used by economists or other social scientists to design rudimentary models of social interactions which are closer to being interactive or automated visualisations of things we already know than to system-wide simulations of what the removal of all social problems could look like.
Such planning and calculations have always helped us evolve better social practices. As neurologist Gerald Smallberg writes, human beings "have the ability to try to contemplate the future that provided Homo sapiens its great evolutionary advantage. This talent to imagine a future before it occurs has been the engine of progress, the source of creativity." For instance, story-telling has long been a way for human beings to simulate ideas and transmit those simulations. The hold ancient epics and sacred texts have on human societies till today is a testament to the power of large-scale simulations that can also be applied to a host of microcosmic situations. But looking at the religious and cultural strife in the world, perhaps the dogma and didacticism of such narrative simulations no longer serve us as well as the guardians of these "traditional" social modelling would like us to believe.
As is obvious, there is no match between the computational capacities of computers and human beings, but there are stark cognitive differences. It is crucial that we zero in on those parts of human knowledge that computers have only learned to mimic so that we can translate our narrative understanding of complex social systems into terms or frameworks with which automated machines can work ethically and efficiently. Along the lines of our earlier examples from the natural sciences, can we design and manage elaborate social simulations via supercomputers, working on testable hypotheses from the social sciences, to arrive at helpful and constantly reviewed models for better social organisation like we simulate the structure of crystals, genes, proteins, and machines?