Version 0.3.2 · Experimental · Source available, noncommercial
A framework for
readings that can lose.
What does a question take for granted? What would prove your reading wrong? What do you gain by being the person who sees through it?
The Zizkian turns those questions into a small, inspectable record. Its Prolog checker flags missing support, defeated readings and interpretation traps. You keep the choice to test, change direction or retain the original brief.
A project inspired by ideology critique. “What would prove that reading wrong?” is an inherited question. The name does not imply a theory or endorsement by Slavoj Žižek.
One word: refutable.
Inspect a fictional reading in three states. These are recorded outputs from the included Prolog checker. The buttons select reports. Edit these records and run Prolog in the browser lab.
The record declares support
Flow complexity prevents first-task completion. The fictional record includes an observation and a tested ordinary alternative.
Inspect the complete fictional input
{
"id": "onboarding-supported",
"mode": "consulting",
"coaching_invited": false,
"brief": "Improve onboarding",
"desired_result": "New users complete the first task",
"owner": "Product lead",
"lenses": [
"constraint_hunter",
"veil_piercer"
],
"ordinary": {
"claim": "Clear instruction may be sufficient",
"check": "Observe first-task completion after instruction",
"status": "checked",
"evidence": [
"e1"
]
},
"return_cut": {
"observer": "Analyst",
"statement": "I may favor redesign because it gives my analysis a larger role",
"defeater": "When a product-lead-owned comparison finds instruction sufficient, I recommend keeping instruction and closing this analysis",
"status": "hypothesis"
},
"evidence": [
{
"id": "e1",
"statement": "In this fictional comparison, users still failed after clear instruction",
"source": "Fictional comparison A; no real participants",
"kind": "observation"
}
],
"readings": [
{
"id": "r1",
"claim": "Flow complexity prevents first-task completion",
"status": "supported",
"response": "unasked",
"support": [
"e1"
],
"counterevidence": [],
"defeat_test": {
"method": "Compare instruction with a simpler flow",
"observation": "Instruction alone performs as well as the simpler flow",
"status": "survived",
"evidence": [
"e1"
]
}
}
],
"outcome": {
"kind": "change",
"action": "Trial a simpler flow",
"reason": "Declared failures after instruction warrant a reversible trial. To limit the analyst's redesign preference, the product lead reviews the comparison and can choose to retain instruction instead.",
"reading_ids": [
"r1"
],
"stop_rule": "Reconsider if instruction alone performs as well"
}
}{
"evidence_verified": false,
"human_review_required": [
"source_authenticity",
"evidence_relevance",
"test_discrimination",
"ordinary_explanation_strength",
"participant_authority",
"analyst_payoff",
"stopping_decision"
],
"schema_version": "0.3.2",
"scope": "declared_record_only",
"status": "pass",
"verdict": "awaiting_change_review",
"violations": [],
"warnings": []
}This is a finite witness: adding the declared defeating observation blocks an active reading; withdrawing it permits closure. It proves a behavior of this checker. It does not establish real-world truth or better decisions.
swipl -q -s proof/refutable.pl reproduces all three states after extracting the source package.
Enter where the question is.
Understand the method
Start with the inquiry and its limits. The historical essay is a separate conceptual appendix.
The method and its invariantsHistorical conceptual essay (optional)Where this came fromPut a reading at risk
Follow nine diagrams, run fictional cases and complete ten exercises.
The field guide and exercisesThe executable record contractHow the framework itself could failUse the questions
A small panel for a business brief, one invited lens for personal reflection.
Consulting question libraryCoaching question libraryWWZS? provocation modeA pass still requires human review.
The checker audits declared relationships. It cannot determine whether a source exists, an observation supports a claim, a test was honest or consent was real. Fabricated prose can pass. No finding remains a valid conclusion.
The weights in the attention matrix are authored priorities, excluded from the checker and from decision authority. This release has running code and test cases, with no controlled outcome benchmark.
Read the limits before using the resultWhy the instrument is not for sale.
The restriction protects a question from becoming a credential.
If a framework examines who gains authority from an interpretation, its own packaging belongs in that examination. This public release offers noncommercial permissions and no paid commercial licensing tier. That choice can itself fail: unpaid work can exploit people, restrictions can entrench powerful firms, and refusing profit can produce prestige.
Code: PolyForm Noncommercial 1.0.0. Prose, prompts, diagrams and fictional examples: CC BY-NC-SA 4.0. Standard terms govern permissions; the polemic is an ethical argument.