It’s all just a simulation


A double pendulum simulation made with Python
Source: https://commons.wikimedia.org/wiki/File:Double_pendulum_simulation_python.gif

What’s new?

In 2006, the British magazine New Scientist published an article by philosopher Nick Bostrom titled “Is Reality a Simulation?” in which he repeated an argument he first made in 2003: that we may all be living in a computer simulation. The article continues to reverberate and create discussion so New Scientist republished the 2006 article in its June 6-12, 2020, issue and included it in its new Essential Guide: The Nature of Reality.

What does it mean?

A simulation is a model of a real object. A computer simulation is a model implemented in computer code. Examples of simulations include computer games (such as Sim City), training exercises (such as disaster planning exercises), and some engineering tools (such as Solidworks),

A model or simulation is useful for performing experiments that would be costly, time consuming, and disruptive if performed on the real world system. In my field, industrial engineering, discrete event simulations are used to model the flow of products through production facilities, enabling the performance of experiments to determine, for example, the increase in product flow with the addition of a new machine or more staff (the phrase “discrete event” means the program steps through time simulating each event as it occurs). Many simulations incorporate random factors, so the simulation is run many times to estimate the probabilities of the various outcomes of the simulation (these are called “Monte Carlo simulations”, after the famed casino).

Bostrom’s argument is sophisticated, but put simply he argues (1) that it is likely that somewhere in the universe civilizations have developed technologically beyond our current state to the point where they can create sophisticated simulations including simulated minds that are conscious, (2) that such civilizations would be interested in creating simulations of their ancestors’ lives, and thus, (3) given the vast size and time scale of the universe, such simulations must have been created many times. Thus, Bostrom concludes, we are more likely to be living in a simulation than to be living in reality.

What does it mean for you?

I remember my college philosophy professor (many, many decades ago) telling my Introduction to Philosophy class that we might be the subject of an immersion experiment in which everything we experienced was simply the result of the artificially created world we lived in. He then posed the question: how would we know if we were in an experiment or the real world? I don’t remember the conclusion, or even any of the discussion, but obviously the question stayed with me. Over the years I have decided that, if the world is a simulation, it’s a very well done one and it’s all I have, so I might as well get on with “life.” I basically say to myself: I’m an engineer; let’s get on to practical topics.

Because simulations are practical, their use is growing. Using massive amounts of data to simulate the physical processes expressed in partial differential equations, numerical simulation is widely used to create weather forecasts that give probabilities over a range of possible outcomes. Simulations of the stock market are used to assess the viability of an investor’s portfolio. Simulation can be used to predict the spread of a disease in a population.

Simulation is widespread in gaming. I played an early game called Rogue (loosely based on Dungeons & Dragons), with treasure, monsters, and magic items, including the demonically named Boon of Genocide (the recipient can wipe out all of one kind of monster for the remainder of the game). Games are often at the forefront of the development of computer technology and these games have contributed to the development of simulations for learning and for decision making.

Simulations allow the user to perform experiments on the simulation, and more generally, simulation is a learning environment. Through repeated use of the simulation, the user may be able to develop intuition that normally would take a human many years to acquire. However, the intuition acquired is intuition about the results from the simulation, which, whether the user realizes it or not, may not always match the results from the real world. Daniel P. Huffman recently reminded us that in the game SimCity, crime is “treated very much as a natural consequence of population growth” and the solution is easy: pay for a police station and “all residents are happier and everything gets better.” Current events show that you can learn the wrong lessons from a simulation. The use of simulation in policy is especially fraught with this risk. From Industrial Dynamicsthrough Urban Dynamics and World Dynamics to Limits to Growth, policy makers have to be careful of built in assumptions of simulations, sometimes described as Malthus in, Malthus out or Malthus with a computer.

I tell my students that engineers use many models, but that we always must remember, “It’s only a model,” (said with a shrug of one’s shoulders). Of course a model cannot represent all aspects of the real world and sometimes a model will be wrong.

The opposite can also occur, in which the simulation is assumed to be wrong because the user rejects its actually accurate findings concerning the real world. I wrote a simulation many years ago to predict the pension costs for a company; I spent at least a week trying to find an error because the model predicted a seemingly high number of deaths among the employees. My boss and I both agreed there had to be an error. I finally called Human Resources and asked how many employees died last year; the number was close to what my model predicted. Our intuition about the real world was wrong and the simulation was right.

Simulation takes many forms. I use an online simulation of a Quincunx (or Plinko machine) to demonstrate the central limit theorem to engineering students. The animation of a simulation is always eye-catching and can give the user intuition about the system, but the important conclusions from a simulation usually come from analysis of the numbers generated by the simulation. Solidworks, a solid modeling package, can simulate the assembly of parts and can predict the stresses that parts will experience. A crop simulator simulates the growing of crops, enabling the user to test different crop management practices as climate changes. Simulation of the transportation system of a company can help with fleet management and logistics. Simulation plays an important role in Computer-Aided Drug Design.

Engineers use computer simulations as well as physical simulations such as a crash-test dummy, a shake table to test the effects of earthquakes on a building design, and the San Francisco Bay Model, used to study tidal flows. Physical simulations are also used in training, for example, in health care, aviation, and fire-fighting

The growth in computing power, the increasing use of sensors, and improvements in computer graphics have made simulations even more useful and seductive.  A digital twin is a simulation of a particular object, usually updated frequently with data from sensors on the actual object. A digital twin of an object can be used in a simulation of a larger system, interacting with other digital twins to allow experimentation and prediction. More generally, agent based simulation involves writing computer code to describe the behavior of the components of a system and then letting them interact in the larger system; these researchers used the method to plan responses to a zombie invasion of Chicago.  Virtual reality is an immersive simulated experience (my philosophy professor’s hypothetical “is it real?” situation).

The range of current and potential applications of simulation is staggering. No matter what your organization does, a simulation may exist already, although it will need to be adapted to your particular situation. Many simulation packages or general purpose simulation languages are available in free versions and some are open-source; you can even try out a simple simulation in a spreadsheet. Ask yourself what experiments you would like to perform on your organization; a simulation may be the way for you to do those experiments and deepen your understanding and intuition.

As simulations continue to improve, their use will spread. Perhaps some day soon we will be able to create simulations with conscious minds. Or maybe our descendants have already.

Where can you learn more?

Lists of simulation software are available at Wikipedia, Capterra, and SourceForge.

General purpose simulation companies often have case studies that may spark your thinking: see AnyLogic, Arena, and Simio, as examples.

While historical in focus and academic in style, this special issue of a journal from Springer-Verlag gives an overview of simulation.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.