September 8, 2026 · Yunus Emre Vurgun
Cognitive Biases: A Decision Checklist
Seven biases distort most technical decisions: what gets built, what gets funded, and which incidents get blamed on people instead of systems. Run important calls past this checklist before committing.
The seven biases
| Bias | What it does | Counter-move |
|---|---|---|
| Confirmation Bias | Favoring information that confirms existing beliefs. | Assign someone to argue the opposite case. |
| Anchoring | Over-relying on the first piece of information offered. | Get independent estimates before sharing numbers. |
| Availability Heuristic | Judging probability by how easily examples come to mind. | Ask for base rates, not anecdotes. |
| Dunning-Kruger Effect | Unskilled individuals overestimate their ability. | Require a worked example, not a confidence claim. |
| Sunk Cost Fallacy | Continuing an endeavor because of previously invested resources. | Ask: would we start this today, knowing what we know? |
| Hindsight Bias | Believing past events were more predictable than they were. | Write predictions down before outcomes are known. |
| Bandwagon Effect | Adopting beliefs or behaviors because others do. | Ask what the dissenters know that you don't. |
The one that kills projects
Sunk cost is the expensive one. Teams keep funding a failing migration, rewrite, or vendor because "we've already spent six months." The spent months are gone either way; the only question is whether the next month is better spent here or elsewhere. Make the "would we start this today" question a standing agenda item for long projects.
Fetch it as data
This checklist is the Cognitive Biases dataset — usable as a pre-decision prompt insert for agents that advise on plans:
curl "https://yjtoon.com/api/dataset/cognitive-biases?format=toon"Related: Learning Theories for how teams absorb this material, and verifying LLM output — models exhibit the same biases, borrowed from their training data.