Simulation allows for large-scale searching of a possibility-space to occur with minimal human effort or attention. Smart optimization strategies can discover novel solutions which satisfy goals, as well as inventive means by which plans might fail. Taken together, simulation can help bulid a picture of not only what is possible, but what is likely, and therefore advisable, within a complex system with many moving parts.
HASH models allow probabilistic estimates to be expressed within simulations. Allowing prior beliefs to be combined with randomness, and running multiple replications of a simulation, it becomes possible to not only determine theoretical best strategies, but also the practical optimal approach in any given scenario, accounting for risks and uncertainties. HASH simulations can be used to uncover both technically perfect strategies, and provide an understanding of the probability terrain of a problem, allowing for the pursuit of safer strategies and paths.
More often used as an assistive tool to aide in strategy-development, rather than as a wholesale replacement for humans-in-the-loop, probabilistic HASH simulations are powerful tools for decision-making.
When not being used to generate novel strategies, simulation can be used to test the resiliency of plans under a wide range of real and theoretical scenarios.
HASH supports domain randomization, Monte Carlo simulation, and other risk analysis methods.
Even before simulations have been run, the very act of modeling can improve strategies. “Constructing a simulation requires stating and piecing together various assumptions, and sometimes their edges don’t line up neatly. When a model doesn’t behave the way I expect, I realize that there’s a gap in my understanding of either the structure/design of the system or the dynamics of the system”, says Nur Shlapobersky, a Simulation Architect at HASH and former Systems Engineer at MITRE.
Decision-making with HASH
Whether using HASH simulations to generate new strategies, or battle-test existing ones, it’s easy to get started:
- Pre-built simulations of common business functions and processes, as well as major world cities and social structures can be found on hIndex.
- Optimization experiments can be run at scale on powerful cloud infrastructure with just a few clicks to discover the best behaviors and parameters within any system.