Old Wine, New Barrel: Why Multi-Agent Simulations Matter More than Ever for Defense

Welcome back. This is post number three in our series on defense and simulation. Today, I want to make the case for sustainability – despite the fact that it’s no longer fashionable, assuming it ever was. Regardless, it remains essential. Ever more so.

Back when I was a fresh recruit in the Swiss Armed Forces, the old hands had a simple metaphor: armed forces are an insurance policy. You pay today because tomorrow is unpredictable.

Entire industries as well as academia thrived on the future’s very nature, namely that it is uncertain, yet governments, presumably hoping for eternal peace dividends, allowed their militaries to gradually dismantle their own analytical muscle. Thanks to a few isolated islands of expertise, the knowledge didn’t disappear entirely within the military.

Echoes borne from an earlier world. These publications grew out of work conducted around the turn of the millennium and the early 2000s at the Swiss Military Academy at ETH Zurich. The authors either worked full time at the Swiss Military Academy or were fulfilling their active reserve duties in the Swiss Armed Forces while working in relevant fields in their civilian lives.

We’re now sailing into a perfect storm, often using our rear mirror, in which advances in military technology reshape the battlefield and the battlefield reshapes technology. Multi-domain operations present themselves as complex systems in which convenient linearity is no more.

Dietrich Dörner warned in his book The Logic of Failure that humans fail in complex environments not because they’re impossible, but because we insist on being overconfident, short‑sighted, and simplistic. (Fun fact: he demonstrated this using simulation.)

Multi-agent simulations are perfectly suited to address the uncertainty inherent in complex systems and to support human decision-making. They force systemic thinking, enable the study of dynamics of complex systems that fall beyond the reach of purely analytical techniques, and enable future scenario exploration.

History is unkind to latecomers. And doing nothing is the most reliable way to earn that punishment. The upside? Simulation offers a “safe to fail” sandbox. Just start. Use the people you have, the tools you have, and the knowledge you have – and begin iterating through plausible futures. We don’t have the time or resources to chase perfection. “Good enough, fast” beats “perfect, too late” every single time.

And here’s the bonus the defense world still tiptoes around: multi‑agent systems aren’t just analytical tools. They’re cognitive warfare instruments. Agentic systems shape (see the recent plan to kill Ali Khamenei), control, and optimize decision cycles. In an era of multi-domain operations, multi-agent simulation is not support tech; it has become a foundational technology for maintaining decision superiority and ensuring sustained strategic advantage.

But tools don’t institutionalize themselves magically. Without a stable ecosystem – industry to engineer, academia to innovate, government to anchor – we build on sand.

At Scensei, we have been helping to build the ecosystem uninterruptedly since 2009 through living the partnership model: industry plus academia plus government. Because the alternative is a landscape of disconnected islands drifting toward irrelevance.