Systems thinking: why cause and effect are rarely a straight line

The previous two articles in this series described what happens when an action produces the opposite of its intended result – the cobra effect, in various forms from colonial snake hunting to healthcare administration. This article is about why it happens. The answer lies in systems thinking: the study of how the parts of a system affect each other over time, often in ways that don't become visible until long after the decision was made.

The field that gives us the language for this is called system dynamics, founded at MIT by Professor Jay W. Forrester in the 1950s and later carried forward by John Sterman, now the Jay W. Forrester Professor of Management at MIT Sloan School of Management and director of MIT's System Dynamics Group. Sterman's standard text, Business Dynamics: Systems Thinking and Modeling for a Complex World, gave its name to the entire field this article draws on.

A system, not a chain

The most common trap in decision-making is treating cause and effect as a chain: A leads to B leads to C. A system rarely works that way. Instead, the parts are connected in loops, where C often feeds back into A – sometimes directly, sometimes through several intermediate steps and a delay that makes the connection hard to spot.

Two types of loops explain most of what's hard to predict:

Reinforcing loops make a change drive itself further in the same direction. More customers bring more revenue, which funds investment in the product, which attracts more customers. Left unchecked, the loop becomes either a growth spiral or, if the direction is negative, a downward spiral that accelerates.

Balancing loops instead resist change and seek an equilibrium. A thermostat is the simplest example: temperature rises, the system reacts, temperature falls back. Much of what people call an organization's "immune system" against change – why new initiatives so often fizzle out – is a balancing loop in practice.

The cobra effect is, at its core, a reinforcing loop that got built in by accident. The bounty for dead cobras created an incentive (breeding) that pushed more cobras into the system, not fewer – a loop that reinforced exactly what it was meant to reduce.

The delay that makes loops dangerous

What makes loops hard to manage in practice isn't the loop itself, but the delay between cause and visible effect. In the healthcare example from the previous article, it takes time before increased administrative burden shows up as longer queues, and even longer before those queues show up as falling satisfaction scores. Whoever made the original decision to measure satisfaction more aggressively is rarely still in the same role, or even aware of the connection, by the time the effect finally becomes visible.

The delay creates two recurring problems. The first is overreaction: because the effect of an action isn't immediately visible, the lack of change gets read as the action not being enough, so more gets added – until all the delayed effects hit at once and surprise everyone involved. The second is misdiagnosis: by the time the problem is visible, enough else has happened that the original action is no longer the obvious explanation, and the search for the cause starts in the wrong place.

A recurring pattern: "Fixes that Fail"

Systems thinking includes a set of recurring patterns, known as system archetypes, that describe how organizations repeatedly fall into the same type of trap regardless of industry. The one that best describes what this series has covered so far is called "Fixes that Fail" – a quick fix that relieves the symptom in the short term but, through a delayed side effect, makes the underlying problem worse in the long term.

The pattern follows the same shape every time: a problem appears, a quick fix is applied, the symptom eases temporarily, but the fix has an unintended side effect that, with a delay, reinforces the root problem. Because the relief comes fast and the side effect comes late, the same quick fix often gets applied again and again – each time from a slightly worse starting point.

A closely related pattern is "Shifting the Burden": the easy, symptom-relieving solution gets chosen over the harder root-cause solution, and every time it is, the organization's actual capacity to address the root cause gets a little weaker – which makes the easy solution even more tempting next time. Hiring administrative staff to handle the symptoms of an underlying process problem, instead of fixing the process, is an example that fits that pattern exactly.

Why this matters for decisions, not just diagrams

The point of learning to see loops, delays, and archetypes isn't to draw nice diagrams. It's to change the question you ask before a decision. Instead of "does this solve the problem," the question becomes: "what loop does this set in motion, and where in that loop does the effect come back around to hit us – and when?"

That question also explains something otherwise hard to accept: that the best solution to a deeply rooted problem often makes things visibly worse before it makes them better, while the quick, symptom-relieving fix often delivers a sense of success right away – one that later reverses. That's exactly the dynamic the next article in the series takes on.