Bias Guardrails for Better Decisions
Turn costly judgment errors into rules that block recurring bias
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 97%
Daryl Morey's decision process treats every important mistake as evidence about a flawed judgment mechanism. The team reconstructs what happened, identifies the irrelevant cue or cognitive shortcut that influenced evaluators, and then installs a narrow rule to prevent recurrence. A demeaning nickname, for example, can make one physical feature dominate an athlete's full record, while same-race comparisons can silently restrict the set of analogies scouts consider. Rather than demanding that people simply become less biased, the method changes the decision environment. Its output is a growing set of practical guardrails that preserve relevant evidence, interrupt predictable distortions, and make evaluations more consistent without pretending that human judgment can be eliminated.
Origin
Extracted from Masters in Business through Michael Lewis's account of Daryl Morey's Houston Rockets scouting operation and its attempts to control judgment errors.
Core principles
- 01Analyze mistakes instead of merely regretting them
- 02Convert recurring biases into explicit constraints
- 03Keep irrelevant labels from shaping evaluation
- 04Use structured comparison to challenge instinct
- 05Treat decisions as opportunities to learn
How to run it
- 1
Audit a decision error
Choose a meaningful miss and reconstruct how the team reached its decision. Separate the evidence available at the time from the story created afterward.
Pro tip Review both rejected winners and selected failures to avoid one-sided learning.
Watch out Do not assume the final outcome alone proves that the original decision was irrational.
- 2
Find the distorting cue
Identify the label, image, analogy, stereotype, or emotional reaction that displaced more relevant evidence.
Pro tip Listen for language that repeatedly frames how everyone discusses the subject.
Watch out A memorable cue may look harmless even when it anchors the whole conversation.
- 3
Create a narrow guardrail
Write an explicit rule that blocks or counterbalances the identified cue without banning legitimate analysis.
Pro tip Prefer observable rules such as prohibiting nicknames over vague instructions to be objective.
Watch out Overly broad restrictions can remove useful information along with the bias.
- 4
Apply it before deliberation
Introduce the guardrail before evaluators discuss the next candidate so the initial framing cannot contaminate the group.
Pro tip Include the rule in templates, scorecards, and meeting agendas.
Watch out Applying the rule after consensus forms is unlikely to reverse anchoring.
- 5
Measure and refine
Compare later decisions with earlier ones and determine whether the guardrail reduced the targeted error. Revise or retire rules that create new distortions.
Pro tip Maintain a decision log containing evidence, predictions, outcomes, and active guardrails.
Watch out Do not preserve a rule merely because it sounds principled.
In the wild
Houston scouts saw a shirtless photograph of Marc Gasol and began referring to him by a mocking physical nickname. Morey concluded that the label distorted every later discussion and contributed to the team passing on him. He subsequently prohibited nicknames in player evaluation.
→ The team removed a framing device that could overwhelm evidence about a prospect's basketball ability.
Morey noticed that scouts compared Black prospects with Black stars and white prospects with white stars. He allowed analogies only when the compared players were of different races, forcing scouts to examine playing style rather than defaulting to racial resemblance.
→ The stereotyped comparison habit stopped rather than continuing under the guise of expert intuition.
Common mistakes
Demanding unbiased intuition
Telling people to be objective does not remove automatic framing effects. Change the process around the judgment instead.
Writing rules before finding causes
A generic checklist may create bureaucracy without addressing the cue that produced the actual error.
Ignoring guardrail side effects
A constraint can introduce a new blind spot, so its results must be reviewed rather than treated as permanently correct.
Is it for you?
Best for
It is best for teams making repeated hiring, scouting, investment, or product-selection decisions.
Not ideal for
It is not ideal for unique creative choices where rigid constraints would suppress useful experimentation.
From the transcript
“they built in kind of rules to try to guard against errors and judgments.”
“So he banned nicknames, said, no nicknames, it distorts our judgment of the person.”
“he banned right analogy analogies within race, so you can compare a player to another player and say, oh, he plays a lot like X…”
From the episode
Michael Lewis Discusses the Culture of Finance (Podcast)