Cognitive Biases: How the Brain Fools the Engineer

The brain is a remarkable organ. It allowed us to build civilization. But it also systematically deceives us. Cognitive biases are not bugs - they are features, optimized for survival on the savanna, not for making decisions in IT.

• **Confirmation Bias:** An 'objective' analysis that confirms a decision already made in advance
• **Sunk Cost:** Continuing a failing project because 'we've put so much in'
• **Authority Bias:** 'Google does it this way' without understanding their context
• **Bandwagon:** 'Everyone is moving to X' - so should we

Предварительные знания

• Critical Thinking in IT: Why Smart People Make Terrible Decisions

What Are Cognitive Biases

**Cognitive biases** are systematic thinking errors. Not random mistakes, but predictable patterns that repeat across all people.

**Why the brain works this way:** The brain optimizes for energy, not accuracy. Heuristics (shortcuts) work in 95% of cases. But in the remaining 5% - catastrophe.

**Key point:** Knowing about biases doesn't provide immunity. But it enables building systems of checks - checklists, reviews, processes.

Confirmation Bias: Seeing What We Want to See

**Confirmation Bias** - we seek out information that confirms our existing beliefs and ignore anything that contradicts them.

**How it manifests:** - Only blogs that agree with the existing position get read - Neutral data gets interpreted as confirmation - The wins of the chosen technology are remembered; the losses are forgotten - An 'objective' analysis confirms a decision already made in advance

**Antidote: Steel Man**

Sunk Cost Fallacy: The Investment Trap

**Sunk Cost Fallacy** - we continue a project because we've already invested resources, even when continuing makes no sense.

**Past investments must not drive future decisions.** The money and time are already spent. The only question is: what is the best thing to do NOW?

**Signs of sunk cost fallacy:** - 'We've already put so much in' - 'Almost done, just a bit more' - 'We can't admit we were wrong' - 'Investors/management won't understand'

Authority Bias and the Bandwagon Effect

**Authority Bias:** If an expert or leader said it - it must be true. **Bandwagon Effect:** If everyone does it - it must be right.

**Antidotes:** 1. **Ask 'why'** - not 'what to do' but 'why does this work' 2. **Check context** - does others' experience apply to the current situation 3. **Seek dissenters** - who disagrees, and why 4. **Demand evidence** - not opinions, but data

**Experts are valuable, but not infallible:**

• An expert in one field ≠ an expert in everything
• An expert may have conflicts of interest
• An expert operates in a different context
• Even experts are wrong in 30–40% of their predictions

Key Takeaways

• **Cognitive biases are systematic** - predictable thinking errors
• **Confirmation Bias:** We seek confirmation, ignore contradiction. Antidote: Steel Man
• **Sunk Cost:** Past investments must not drive future decisions
• **Authority/Bandwagon:** 'Everyone does it' and 'the expert said so' are not arguments without context
• **Knowledge doesn't immunize anyone** - systems of verification are essential

What's Next?

Biases are the diagnosis. Next come the tools:

• Evaluating Evidence — How to distinguish facts from opinions
• Trade-off Analysis — How to compare decisions
• Detector Questions — How to catch biases in real time

Вопросы для размышления

• Consider a decision where analysis 'proved' something that had already been decided beforehand. Confirmation bias?
• Which project dragged on longer than it should have because of sunk cost?
• Where were 'best practices' copied without understanding WHY?

Связанные уроки

• prob-04-bayes