Compare Two Scenarios
Scenario comparison lets readers juxtapose two agents and see how similar surface beliefs can arise from different epistemic structures.
Comparison Lab
Juxtapose two epistemic structures
Use the presets or adjust the sliders to compare two agents side by side. The point is not just who is right, but where each agent's belief state is breaking: evidence perception, credence assignment, or both.
Preset Pairings
Same rough conclusion, different epistemic structure: limited tools on one side, motivated override on the other.
Scenario A
Diagnosis
Mainly a Deep Rationality problem
Focus on evidence access, base rates, measurement quality, statistical tools, and better comparison classes.
Scenario B
Diagnosis
Mainly a Core Rationality problem
Focus on fear, identity, incentives, public commitment, and the conditions under which the agent could honestly update.
Interpretation Prompts
Which agent needs better evidence or better tools? Scenario A looks more like a skill-side problem.
Which agent needs motivational repair? Scenario B has the larger excess Core Irrationality (IC).
Could both agents reach the same conclusion while only one is well aligned?
What new evidence, reflection, or social condition would move each agent responsibly?
Use the Comparison Lab First
Start with a preset pairing, then adjust the sliders on either side. The important move is to compare structures rather than personalities: where is Objective Evidence (E0), what does each agent perceive, what credence do they assign, and how much Deep Rationality (SD) do they bring to the case?
The comparison table turns the visual difference into diagnostic questions. A larger Objective Evidence to Perceived Evidence gap (E0-EP) suggests an evidence-reading or skill problem. A larger Perceived Evidence to Assigned Credence gap (EP-CA), especially beyond warranted slack, suggests motivated override or belief-integrity pressure.
Why Compare
Two agents can hold the same belief for very different reasons. One may be an honest novice; another may be a biased expert. One may have bad data; another may have good data and bad faith. Comparing scenarios prevents superficial diagnosis.
Comparison Questions
Compare where the gaps occur. Is one case mostly Objective Evidence to Perceived Evidence (E0-EP) and the other mostly Perceived Evidence to Assigned Credence (EP-CA)? Does one agent have lower Deep Rationality (SD) but better internal alignment? Does one agent become more accurate when given better evidence, while the other resists?
Also compare repair paths. If the problem is mainly Deep Rationality (SD), better information, base rates, and statistical tools may help. If the problem is mainly Core Irrationality (IC), the repair may require safer update conditions, identity detachment, incentive changes, or courage to say what one already sees.
What Comparison Teaches
Comparison teaches that the same conclusion can have different epistemic quality. Being right is not enough; being right for evidence-responsive reasons matters. Being wrong is not always culpable; the structure of the error matters.