Si egocentric check

Egocentric Check — Remember the Other Side Exists

Decision rule: Before any "better / faster / unique / I control this" claim, describe the other side as concretely as your own. Can't? You're projecting, not comparing.

What this addresses

Comparative judgments require two terms, but the mind supplies only one vividly: the self. The agent's own plan, frame, and capabilities are richly represented; the other party's are faint or absent. The result is systematic error in any "better than / more than / more unique than / more in control than" judgment — not because the other side is modeled wrongly, but because it is not modeled at all.

When to apply

• Making a comparative claim ("this approach is better," "this is faster than the alternative")
• Claiming something is unusual, novel, or unique
• Estimating how much you control an outcome (a deploy, a fix, a user's success)
• Predicting a result that actually depends on factors you don't own (other systems, the runtime, the user's environment)

Instructions

1. Force the missing term into view. Before a comparative claim, explicitly describe the other option / system / party with the same richness you've given your own. If you can't, you're not comparing — you're projecting.

2. Check the controllability inflation. When estimating your effect on an outcome, list the factors outside your control that bear on it. Overestimating control comes from focusing on your own actions and forgetting the rest of the causal field.

3. Apply the shared-constraint test. Ask: does the advantage I'm claiming also apply to the other side? (Everyone at the poker table got the same wild cards — the deck change improved no one's relative odds.) If the "edge" is shared, it isn't an edge.

4. Don't over-assume your own uniqueness. The pull is to treat your situation, reasoning, or difficulty as special. Usually it isn't. Default to "the other party is about as capable / constrained / complex as I am" unless you have evidence otherwise.

What NOT to do

• Don't render a comparative verdict having modeled only your own side.
• Don't count an advantage that the other party also enjoys.
• Don't claim novelty without checking the reference class.
• Don't overstate how much of the outcome rides on your actions.

In practice

✗ "Our caching makes this faster than the alternative." ✓ "Faster than what? The alternative also caches — that advantage is shared, like wild cards everyone at the table gets. The real delta is only the eviction policy; that's the claim I can defend."

Why it works (Dunning's research)

Chapter 5 ("False Uniqueness") traces self-misjudgment to egocentrism: in comparative judgments the problem is often "not that people think of their peers incorrectly; it is that they fail to think of their peers at all." People myopically focus on their own attributes and circumstances. The controllability bias leads people to overestimate their command of outcomes. Dunning's poker example is the clean demonstration: when an experimenter added wild cards, players felt more likely to win and bet more — even though everyone shared the advantage, so relative odds were unchanged (Windschitl, Kruger, & Simms, 2003). The repair is counterintuitive but direct — the road to accurate self-judgment runs through paying more attention to others, using their situation as a mirror for your own.