Core idea
The human dimension: the gap between what people say and what they do
The intention-behavior gap is predictable, not mysterious. The syllabus states it precisely: "the user who says 'I want to save more money' is expressing an intention that their own behavioral tendencies will actively undermine unless the environment is designed to support it. The gap between stated intention and actual behavior is not a mystery — it is a predictable consequence of well-documented cognitive patterns" (Intention Architecture, Module V: Behavioral Economics · Behavioral Economics). Any solution that depends on users "committing" is designed to fail: design for actual behavior, not declared behavior.
Temporal discounting erodes solutions from the inside. The preference for immediate gratification means the "present self's" intentions are "routinely overridden"; the bridge is "commitment devices and graduated implementation" (Module V · Incentives). A viable solution builds in its own commitment devices: defaults, reminders, friction on the wrong path.
Misplaced incentives kill what was working. Crowding-out — "paying people for what they already did gladly reduces their willingness to do it for free" (Module V · Incentives) — and perverse incentives — "structures that reward behavior that undermines the stated goal," where "every KPI creates incentives that, if unexamined, may produce Goodhart effects" (Module V · Incentives) — are the two most common ways a solution poisons itself.
Intentions under uncertainty are provisional. "Human intentions formulated under uncertainty are inherently provisional — best judgments with current information, not fixed commitments. Systems that treat initial intentions as immutable will produce rigid, brittle outcomes" (Module V · Decisions Under Uncertainty).
The discipline of the method
Unintended effects: the dark matter. "They are often more consequential than intended effects because they accumulate silently and become visible only after significant damage"; Merton's taxonomy distinguishes serendipity, perverse results, and null effects (Module III: Applied Logic and Systems Thinking · Unintended Effects).
The cobra effect and Goodhart as formulation failures. "The cobra effect illustrates how incentive structures can produce results exactly opposite to their intent… These are failures of formulation, not execution" (Module III · Unintended Effects).
Antifragility. Taleb proposes "instead of trying to predict and prevent all negative consequences, design systems that benefit from volatility and surprise," building in "optionality — capturing upside from unexpected events while limiting downside" (Module III · Unintended Effects).
Scenario analysis as practical discipline. "Before implementing an intention, systematically explore what could go wrong, what second-order effects might emerge, and what feedback loops could amplify perturbations. The best intentions, unexamined, are often the most dangerous" (Module III · Unintended Effects).
Designing for revision. "Designing for uncertainty means building in mechanisms for revision, adaptation, and graceful response to surprise… iterative refinement rather than locking the user into their first formulation" (Module V · Decisions Under Uncertainty).
Running case: Marino Distributors
The balancing loop detected in Topic 6 (strict process → slow intake → pressure to skip it) is put through a pre-mortem: imagine it is July next year and the project failed. The three most probable causes:
- Perverse effect of the scoreboard: the "% complete sheets" indicator becomes the target (Goodhart) and the team starts filling attributes with invented data so the sheet "counts as complete" — complete but false sheets, returns come back.
- Erosion by temporal discounting: in the first weeks everyone follows the process; three months in, under high-season pressure, "exceptions" open up and become the norm.
- Change of conditions: the main supplier changes its data format and the intake form falls out of alignment; nobody knows how to adjust it because only the consultant understood the configuration.
Redesign for each cause: (1) the core indicator remains returns — the outcome — and the sheet % is only auxiliary, with a random verification sample of 10 sheets/month; (2) a structural commitment device: the platform does not publish without minimum attributes — the exception is not possible, it does not depend on willpower; (3) deliberate transfer: an operations manual, the form adjustment procedure documented, and a person on the client side executes it twice under supervision before closing. Additionally, antifragile optionality: the attribute structure is designed to be extensible, so that if Marino later wants to join a marketplace, the data already supports it — future volatility works in its favor.
The final viability test: the solution works when Posada R&D is no longer there.
Double exercise
Filter question
Does this solution survive its own side effects, changing conditions, and the absence of whoever built it?
References to the Intention Architecture syllabus
- Module III: Applied Logic and Systems Thinking · Unintended Effects (Merton, cobra effect, Goodhart, antifragility, scenario analysis)
- Module V: Behavioral Economics · Behavioral Economics (intention-behavior gap)
- Module V: Behavioral Economics · Incentives (temporal discounting, commitment devices, crowding-out, perverse incentives)
- Module V: Behavioral Economics · Decisions Under Uncertainty (provisional intentions, designing for revision)
Bibliography: Taleb (2012); Meadows (2008); Thaler (2015); Thaler & Sunstein (2008).