Purpose-Calibrated Depth
Investigate a question only as deeply as the decision requires
- Difficulty
- Easy
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 93%
Purpose-Calibrated Depth treats understanding as a layered model rather than an all-or-nothing state. Begin with the basic affordance needed to act: a cup's handle is enough when the goal is simply to drink coffee. Move deeper only when the current layer cannot answer the decision, explain a failure, or expose a material risk. For recurring high-impact domains, investigate important plumbing in advance so a crisis does not force learning from zero. Preserve that knowledge as a working model, then update it when the environment changes. The method balances efficiency with preparedness: routine decisions receive a sufficiently accurate abstraction, while fragile or consequential systems justify deeper investigation. Its output is not perfect knowledge, which is unattainable, but a model detailed enough for the present purpose and inexpensive to refresh when circumstances shift.
Origin
Extracted from Masters in Business through Dave Nadig's discussion of affordances, market plumbing, and how a financial futurist prepares before systems break.
Core principles
- 01Match explanatory depth to the purpose at hand
- 02Use simplified models when they are sufficient for action
- 03Build deeper knowledge before fragile systems fail
- 04Update existing models instead of restarting from scratch
How to run it
- 1
Define the purpose
Specify what decision, explanation, or action the research must support. This sets the required standard of detail.
Pro tip Write the decision as a question with a concrete next action.
Watch out Without a defined purpose, curiosity can turn into unbounded research.
- 2
Start with the usable abstraction
Build the simplest model that explains the relevant objects, relationships, and expected behavior.
Pro tip Use familiar affordances or analogies to orient yourself quickly.
Watch out Do not mistake a useful abstraction for a complete description of reality.
- 3
Test for sufficiency
Ask whether the model resolves the decision and whether omitted details could materially change the outcome.
Pro tip Focus on assumptions most likely to fail under stress.
Watch out Familiarity can make a shallow model feel more complete than it is.
- 4
Descend through consequential layers
Follow mechanisms, dependencies, incentives, and operational plumbing until the unmet need is satisfied.
Pro tip Prioritize boundaries where ownership, money, data, or responsibility changes hands.
Watch out Stop when added detail no longer changes understanding or action.
- 5
Store and refresh the model
Retain the mechanics and assumptions so future events require an update rather than a complete restart.
Pro tip Review the model after regulatory, technological, or market changes.
Watch out Previously accurate domain knowledge can become dangerously stale.
In the wild
An analyst learns the basic mechanics of payment for order flow before it becomes headline news. When controversy erupts, the analyst checks what changed during the previous eight weeks and updates the existing model rather than beginning with the acronym.
→ The analyst produces a faster and more grounded explanation during a market disruption.
An advisor needs only a functional explanation of ETF trading for a routine allocation discussion, but investigates custody, settlement, and counterparty mechanics before recommending a novel tokenized product.
→ Research effort matches the consequence and novelty of each decision.
Common mistakes
Confusing detail with usefulness
More information is not automatically better when it cannot affect the decision or expose a relevant risk.
Stopping at a familiar label
Knowing terminology without understanding the mechanism leaves the model fragile when conditions change.
Letting expertise go stale
Deep prior knowledge still needs a current-state update before it can support today's conclusions.
Is it for you?
Best for
Researchers, advisors, analysts, and operators who must navigate complex systems under time constraints.
Not ideal for
Safety-critical or scientific work where the required depth is dictated by formal verification standards.
From the transcript
“Our model of the universe doesn't have to be perfect. It just needs to be good enough to get us through the day.”
“It's not important to what your job is until something breaks right”
“But I'm not starting from scratch every day”
From the episode
Dave Nadig on Exchange-Traded Funds
Dave Nadig