To Improve The Real World, Simulate It

Lux Partner Bilal Zuberi and Lux Scientist-in-Residence Sam Arbesman share their view on why we need to move everything we do into the realm of simulation in order to efficiently build the future.

The real world is a mess. It’s slow, it’s complicated, and it’s tricky to deal with. Want to make a new medicine? Dealing with the real world might mean pipetting a large variety of chemicals over and over and trying to see which ones work. Want to build a better airplane? You can make models and play with wind tunnels. Want to train self-driving cars? You might just have to test them on millions of miles of road, yet still not know if they can handle every edge case. Want to develop new batteries? It might require conducting thousands of experiments to test for charging capacity and durability.

Simply put, the real world is not enough for us to be able to efficiently build the future world: we need to move everything we do into the realm of simulation. We can experiment so much more, so much faster, and much more safely in computers than in the real world. Drug discovery must move into computers. Industrial design should happen in machines. Strength and degradation of our infrastructure should be modeled to avoid disastrous accidents. Autonomous vehicles on our streets and in the air should use simulation for safer design and development. And even how we reopen our sports facilities safely during a pandemic must be examined virtually. 

Everything that is designed, engineered, modeled, manufactured, and tested should be done in silico first.

From agent-based modeling in urban planning and epidemiological scenarios to the sophisticated chemical and physical models of protein folding, to the mechanics of how a bridge bears weight or a bicycle manages the many forces to which it is subjected, simulation is possible in every field. The vanguard of this is in self-driving and manufacturing—and even in gaming—but simulation is quickly reaching every domain, from supply chains to drug design. There are even endeavors to combine many of these models in order to examine the complex feedback between humanity and our environment, from climate change to resource management.

These simulation tools are already around us, and yet too many people still view simulation as a sideshow. Because that is how change often feels: slow and incremental at first, then sudden and everywhere. We are at that inflection point, where simulation has begun to surround everything we build, from building computational models in industrial manufacturing to molecular dynamics simulations in protein folding, autonomous car development, and digital twins for clinical trials. 

Lux has already made huge bets on this space: Applied Intuition, Desktop Metal, Runway, Shiru, Drone Racing League, and Recursion Pharma. Applied Intuition is able to rapidly simulate the countless number of possibilities involved in the world of self-driving cars, from the differences in sensors to the algorithms that control the cars themselves. Desktop Metal has a suite of tools for generative design, to allow for manufacturing to move into the organic-looking and the strange, for design should only be limited by the laws of physics, not by considerations of what has previously been done. Runway makes it easy for creators to easily generate images and text, putting the tools of machine-based creation into the hands of artists and designers. Shiru computationally explores the massive space of non-animal proteins for foods, combing through this landscape of chemicals for the ones that could provide a way of sustainably feeding the planet. Drone Racing League provides true-to-life simulator for high speed aerial drones to be using in racing, as well as for training aircraft pilots. And Recursion Pharma simulates phenotypes before conducting actual wet lab activities in their high throughput drug-discovery work.

But there is so much more to create in the realm of simulation. While some scientists and engineers involved in trial-and-error experimentation in the real world might see their jobs disappear, as what they do will be viewed as confirmatory rather than the actual work of discovery, there will be new skills to be learned and new talent to foster, as creators and researchers cultivate software-focused skills rather than those of the wet and messy real world. We are going to need to develop new experimental designs and new types of software and tools, and entirely new product development environments, when everything is being built at a breakneck speed inside machines. And we will have to rethink computing architectures and even how data is shared and regulated. Consider this: instead of consultants designing automation for factories, simulators would optimize factories to utilize best available technologies for humans and robots to work together. Similarly, car-makers would simulate the design, engineering, manufacturing, and road safety performance of entire vehicles before 3D-printing the first prototype parts. Effects of Climate Change on human lives, infrastructure, and agriculture could be simulated at local or global level under various scenarios. And vaccines and drugs would not only get designed within minutes of identifying new pathogens but also get put on an accelerated clinical testing path by using synthetic data and digital twins to design clinical trials.

Fourteen years ago, Amazon Web Services made cloud computing a reality, and it’s hard to imagine the world of technology operating any other way. Gone are the days of starting a company by filling a massive room with rows and rows of servers. The same thing is now true of simulation. These are the early innings, but the game is picking up pace rapidly. We are on the cusp of a massive shift in how we imagine building our future. 

If software is eating the world, simulation is building it anew.

ABOUT THE AUTHORS

Bilal Zuberi PhD, is a partner at Lux Capital. He backs startups that solve big, practical problems with technically ambitious solutions, with founders who are driven to build large companies and envision a better more prosperous future than today. Find him on Twitter at @bznotes.

Sam Arbesman PhD, is Scientist-in-Residence at Lux Capital. As a complexity scientist, Sam is passionate about bringing together seemingly unrelated ideas from science and technology. He works with founders that recognize the future happens at these boundaries, in such areas as open science, managing massive technological complexity, artificial intelligence, and infusing computation into everything from biology to manufacturing. Find him online at arbesman.net, or subscribe to his newsletter: Cabinet of Wonders.