Systems Thinking and behavioral change
Just finished reading the UK BIT's Manifesto for Applying Behavioral Science whilst dipping in and out of the Behavioral Economics Guides from the past 10 years.
Ten years ago, behavioral economics promised elegant solutions. Evidence-based, low-cost, measurable solutions to thorny issues. Reality delivered something different. A comprehensive analysis of 165 randomized trials from government nudge units found an average effect of 1.4 percentage points. Academic papers claimed 8.7%, but publication bias inflated those numbers (BE 2020). While nudges remain cost-effective, "a 1.4% shift in behavior is simply not enough to achieve the nudger's goal" for complex challenges (BE 2021).
Why such modest effects? The problem is as much cognitive as anything else. System 2 holds approximately seven objects in working memory (BE Guide, 2014). But actual systems require tracking dozens of interconnected variables. When ego depletion hits, people revert to System 1 heuristics (BE Guide, 2016). Cognitive limits compound with organizational realities. Implementation killers persist: capacity constraints, time pressure, resource limits, political barriers, and evaluation anxiety. "Many of the most important policy challenges occur as part of complex adaptive systems where interactions between the elements can cause unpredictable results" (Source: BE Guide 2023). Humans are fundamentally ill-equipped for systems thinking.
The Behavioural Insights Team's manifesto acknowledges what practitioners were discovering: we need to see the systems those interventions inhabit.
The BIT Manifesto offers three pathways:
Identify Leverage Points - Find where behavioral shifts produce wider system effects. Electricity initially replaced steam engines but eventually enabled complete factory redesign.
Model Collective Implications - When households copy neighbors' consumption, changing the copying speed creates tipping points, moving "beyond a focus on individual behaviour to embrace systemic change" (BE Guide 2023).
Change System Rules - UK sugar tax altered manufacturer incentives. Transport for London defaulted pedestrian crossings to green, nudging behavior while reshaping policies.
What should we do? First, zoom into specific tractable loops while recognizing the larger system. Next, diagnose complexity using tools like the Cynefin framework. Build adaptive capacity using the COM-B model (capability, opportunity, motivation) and measure resilience, not just outcomes. Finally, start small but think big. Map one feedback loop. Model one leverage point. Change one rule. Continue.
The guides document a decade of evolution: from 2014's excitement about "predictably irrational" interventions to 2024's pragmatic scaling frameworks. Coupled with the BIT manifesto it's clear we should embed systems thinking into organizational design, rather than just focus on individual capability.
The advice was to start small, map one loop. Adapt. Move on to the next.
But this assumes a system that tolerates such exploration. We don't appear to have that system.
Some thought provoking stuff from Azeem Azhar makes me concerned that the BIT's call to arms is doomed from the outset. Azeem points to a 2023 Nature analysis of 45 million scientific papers that concludes disruptive breakthroughs have declined universally across six decades. Not because we're less capable, but because we've built an infrastructure that penalizes the exploration required for systems work while rewarding incremental refinement.
Systems thinking requires synthesis across disciplines. But grant reviews demand feasibility demonstrations upfront. Complex systems are definitionally unpredictable until explored. But funding mechanisms bias against novelty. The "burden of knowledge" and race for tenure forces hyper-specialization, but systems thinking requires cross domain, broad understanding.
A researcher mapping feedback loops across behavioral science, policy design, and organizational systems would need funding from three skeptical silos. Their publications would satisfy none of the narrow journals that determine career progression. Early setbacks, inevitable in exploratory work, significantly increase attrition rates among scientists.
What have we actually created?
We've created parallel impossibilities. Behavioral science discovered 1.4% nudges can't solve complex problems. Science discovered incremental papers can't produce transformative breakthroughs. Both hit a similar wall: systems optimized for safe, measurable outputs that compound into systemic failure.
Scientific progress depends on failed experiments, but human careers cannot tolerate them. So we get safe nudges instead of systems transformation. Publishable findings instead of disruptive inquiry.
The BE guides show an evolution from 2014's excitement around behavioral science and the power of nudging to 2024's pragmatism and 2025's honesty. The infrastructure must change. We need to fund 'failure' more explicitly. We need to reward boundary crossing. We need to design tenure around exploration, not narrow, result-based exploitation.
But is systems thinking actually implementable within institutions that caused this stagnation? Have we created an incentive architecture that has made transformative work structurally impossible?
We appear to need systems thinking to fix systems that prevent systems thinking.
-
Behavioral Economics Guides 2014-2025
