Explicit — Formulation and framing

The human dimension: why we frame badly without noticing

Biases choose the frame before we do. Anchoring makes "initial exposure to a number — even an irrelevant one — disproportionately influence subsequent estimates," and the availability heuristic makes "vivid, recent, or emotionally charged events be systematically overweighted" (Intention Architecture, Module I: Human Intention and Decision-Making · Biases and Heuristics). The client who just lost a big order will frame their whole problem around that episode, even if it is atypical.

Confirmation bias contaminates the formulation itself. The syllabus flags it as "arguably the most consequential bias": "when a user approaches with a predetermined conclusion, they will frame their query in ways that bias the output toward confirmation" (Module I · Biases and Heuristics). The client does not ask "why are sales falling?"; they ask "we need more advertising, right?".

Gains and losses change the decision. The classic experiment demonstrated that people "are risk-averse when outcomes are framed as gains and risk-seeking when framed as losses — even when the expected values are mathematically identical" (Module II: Correct Problem Formulation · Framing Errors; expanded in Module V: Behavioral Economics · Framing: "95% lean" vs. "5% fat"). How the problem is worded already tilts the solution that will be chosen.

Words mean according to the game they are played in. Wittgenstein showed that "meaning is not an inherent property of words but a function of their use": "check means one thing in chess, another in banking, another in medicine" (Module IV: Language as Action · Context). "Optimize," "urgent," and "mess" mean different things to the owner, the accountant, and the warehouse worker — and an explicit formulation must fix which language game is being played.

Pointing words need context. Deictic expressions ("this," "here," "now") "depend entirely on context"; a request like "fix this" or "make it better" has no referent without contextual grounding (Module IV · Pragmatics). Being explicit is, to a large degree, resolving the hidden deictics of the request.

The discipline of the method

Recognizing wicked problems. Rittel and Webber identify problems "whose definition is itself contested; every attempt to define the problem is simultaneously a commitment to a particular solution space" (Module II · Framing Errors). In those cases, choosing the frame is the most important decision of the project.

Framing operates at two levels. At the linguistic level, word choice shapes perception; at the structural level, "the boundaries drawn around a problem determine which variables are considered relevant" (Module II · Framing Errors). The syllabus example is exactly that of the company framing declining sales as a "marketing problem": "the frame itself prevents the correct diagnosis."

The antidote is deliberate reframing: "restating the problem from multiple perspectives before committing to a formulation," presenting alternative formulations and asking which one best captures the intent (Module II · Framing Errors).

An additional warning for the AI era: language models "are highly sensitive to how inputs are phrased: the same question, asked in two different ways, can produce substantially different outputs" (Module II · Framing Errors). A poor frame does not only confuse humans: it confuses the tools.

Running case: Marino Distributors

With the end already clear (reduce returns caused by wrong purchases, Topic 1), Posada R&D generates three alternative frames before accepting the "messy catalog" one:

• (a) Data problem: product information arrives inconsistent from suppliers; the mess is a symptom.
• (b) Process problem: there is no defined workflow for adding products; everyone uploads them however they can.
• (c) Tool problem: the current platform does not support technical attributes (thread, diameter, material), so key information ends up crammed into the product title.

The client's anchoring bias is also detected: the manager framed everything as "mess" because an important customer complained last week about a duplicated product — a vivid but peripheral episode. Analyzing 40 recent returns, the frame that survives the evidence is (b) with a component of (c): without an intake process, any cleanup will get dirty again. The problem is formulated as:

"Products are added without a process or an attribute structure, producing incomplete product sheets that cause wrong purchases and returns."

Double exercise

Filter question

Is it clearly defined what problem must be solved, and was that definition chosen among alternatives or accepted without examination?

Syllabus references

• Module I: Human Intention and Decision-Making · Biases and Heuristics (anchoring, availability, confirmation)
• Module II: Correct Problem Formulation · Framing Errors (levels of framing, wicked problems, reframing, LLM sensitivity)
• Module IV: Language as Action · Pragmatics (deixis)
• Module IV: Language as Action · Context (Wittgenstein's language games)
• Module V: Behavioral Economics · Framing (attribute, goal, and risky choice framing)

Bibliography: Kahneman (2011); Kahneman, Slovic & Tversky (1982); Rittel & Webber (1973); Wittgenstein (1953); Thaler & Sunstein (2008).