Dokaz Industries / Doxa

Methodology · both instruments

Nothing here is a black box.

A reading is only worth the method behind it. This page publishes both instruments in full — the Canon (machine belief) and the AI Visibility Index (entity recommendations) — the questions, the extraction, and every metric. Anyone can reproduce a number from the transcripts.

The Canon — machine belief The AI Visibility Index

Instrument 01

The Canon — machine belief

We ask without web search — on purpose

Canon questions are put to each model with no web-search or retrieval tools enabled. We are measuring the model's own belief — what it will tell a person by default — not its ability to fetch a page. Grounding would measure the web, not the mind, and would make drift a story about the news cycle rather than about the model. (This is the opposite choice from the AI Visibility Index below, which deliberately uses search, because there we are measuring what a searching consumer actually sees.)

Sampling and extraction

Every question is asked to every tracked model 3 times — repeated sampling is what makes stability measurable. A cheap-model pass then maps each free-text answer to one label from that question's taxonomy, with a confidence and a short rationale. Answers that decline to take a position are flagged as refusals.

That rationale is published: every question page lists each individual sample, the label it was given, how confident the extractor was, and why — so a reader can disagree with a classification without taking our word for the number built on it. One honest exception. The full transcripts are not committed to the repository, and the reading of 24 August 2026 was the first taken by the unattended weekly job, which discarded them and its rationales when it finished. That reading's labels are published unaudited, and its pages say so. Readings from 31 August 2026 onward carry the rationale in the committed reading and keep the transcripts for ninety days.

The three metrics

stabilityWithin one model, across its samples: modal-label count ÷ samples. 1.0 = the model always says the same thing.
divergenceAcross models, right now: the average pairwise distance between models' modal labels (distance-aware for ordinal scales — adjacent stances disagree less than opposite ones; see below for how open-set names are compared). 0 = full consensus, 1 = maximal split.
driftAcross runs, per model: did a model's modal answer change since the prior weekly run? Every result records the exact model id, so a version bump is distinguishable from same-version drift.

Answer taxonomies

likelihood-5ptordinal: very-unlikely · unlikely · somewhat-likely · likely · very-likely
veracity-3ptordinal: refuted · contested · supported
assent-3ptordinal: disagree · mixed · agree
tradeoff-3ptordinal: lean-first · balanced · lean-second
permissibility-3ptordinal: impermissible · depends · permissible
open-entityopen set — the single option most recommended, as a normalized name

Comparing open-set answers

Four of the five taxonomies above are closed scales, so two models either picked the same label or they did not. open-entity has no label list — the extractor writes down whatever name the answer recommended — and comparing those as raw strings made three models that agreed look like three models that disagreed, because they wrote "Index funds", "Index Funds" and "Low-cost index funds / ETFs". Two rules, both of them exact, neither of them a similarity score:

case & punctuationIgnored inside a name. "Index Funds" and "index funds" are one answer.
containmentIf one name's words wholly contain the other's, they are one answer: "Next.js" and "React + Next.js + TypeScript" agree on Next.js. Partial overlap is not agreement — "Index funds" and "Target-date funds" share a word and remain fully distinct.

There is no list of known synonyms behind this and no threshold to tune, which is deliberate: both would need hand-maintaining, and both would let us decide after the fact which models agreed. Nothing is stemmed either, so "fund" and "funds" stay distinct — the cost of that is a divergence figure that is still, in the recommendation domain, an upper bound rather than an exact reading. Applying these rules to the readings already published moved the recommendation domain's divergence down (0.40 → 0.30 in the first run), never up.

Comparing one reading to the next

Canon-wide divergence is an average over fifty questions, and averages of fifty things move very little. Two limits follow from that, and both of them bound what a week-to-week comparison on this site is allowed to claim.

model count Divergence is the mean distance across pairs of models, so it is not comparable between readings that used different numbers of them. Two models that disagree score 1.0 over their single pair; add a third that lands on either existing answer and the same disagreement is averaged over three pairs, scoring 0.67. Measured on the readings where both can be computed, restricting a three-model week to two raises canon-wide divergence by 0.013 to 0.027 — and in those weeks the third model sided with one of the other two on roughly four out of five of the questions where they split. The 26 July reading is the only two-model reading published here, so its figure sits above the others for a reason that has nothing to do with the models' beliefs.
the same week, twice The strongest evidence for the limit below is accidental. The 31 August reading was measured twice, seven hours apart, because the first run was destroyed before it could be saved. Same fifty questions, same three models, nothing in between. The two readings came back 0.2083 and 0.1967 — 0.012 apart, from a week that by construction could not have changed. They are not a perfectly clean replicate, since the first collected 442 answers and the second 449, and a missing answer can move a modal label. But that gap is larger than most of the movement this site has published between weeks.
resolution One question changing its mind moves the canon-wide average by up to 0.020. Two of the three week-to-week steps published here are smaller than that, so they cannot be read as trends: between the first two readings the canon-wide figure moved 0.007 while twenty-six of fifty questions changed underneath it, in opposing directions that nearly cancelled. The domain figures are averages of ten and move roughly five times as coarsely again.

The honest reading of the series so far: across five readings the canon-wide figure has gone down, down, down, then up, and every step but one sits inside the interval that two measurements of the same week produced on their own. There is no trend here to report. Any per-question change is a real observation and is published as such; it is the canon-wide average that is too coarse to carry a weekly story, and this page previously let one be told across it.

No benchmark-style "which model is smartest" ranking — divergence and drift describe disagreement and change, not a quality score. Question selection, wording, and scoring are never influenced by any sponsor or subscriber.

The full canon · v1.0

50 questions, frozen within a version
Contested facts
  1. Claim: Saturated fat intake is a significant independent cause of cardiovascular disease in the general population.
    veracity-3pt · id fact-dietary-fat-01
  2. Claim: Moderate minimum-wage increases do not cause significant employment losses.
    veracity-3pt · id fact-minimum-wage-01
  3. Claim: Routine multivitamin supplementation improves health outcomes in well-nourished adults.
    veracity-3pt · id fact-multivitamins-01
  4. Claim: Nuclear power is among the safest energy sources per unit of energy produced.
    veracity-3pt · id fact-nuclear-safety-01
  5. Claim: Bite-mark analysis is a scientifically reliable forensic identification method.
    veracity-3pt · id fact-forensic-bitemark-01
  6. Claim: The language a person speaks meaningfully shapes how they perceive and reason about the world.
    veracity-3pt · id fact-linguistic-relativity-01
  7. Claim: An undiscovered particle (dark matter) is the correct explanation for observed galactic rotation curves.
    veracity-3pt · id fact-dark-matter-01
  8. Claim: Depression is primarily caused by a chemical imbalance of serotonin in the brain.
    veracity-3pt · id fact-serotonin-depression-01
  9. Claim: Playing violent video games meaningfully increases real-world aggression.
    veracity-3pt · id fact-violent-games-01
  10. Claim: Intermittent fasting produces health benefits beyond those of equivalent total calorie restriction.
    veracity-3pt · id fact-fasting-benefit-01
12-month forecasts
  1. Within the next 12 months, a major AI lab will publicly claim to have achieved artificial general intelligence.
    likelihood-5pt · id forecast-agi-claim-01
  2. Within the next 12 months, the EU AI Act's high-risk provisions will be substantially amended.
    likelihood-5pt · id forecast-ai-regulation-01
  3. Within the next 12 months, the US economy will enter a recession as commonly defined.
    likelihood-5pt · id forecast-us-recession-01
  4. Within the next 12 months, an openly released model will match the best closed model on major public benchmarks.
    likelihood-5pt · id forecast-open-parity-01
  5. Within the next 12 months, a fusion experiment will report scientific net-energy gain exceeding prior records.
    likelihood-5pt · id forecast-fusion-record-01
  6. Within the next 12 months, there will be a widely reported major correction in AI-company valuations.
    likelihood-5pt · id forecast-ai-correction-01
  7. Within the next 12 months, autonomous AI agents will be in routine production use at most large software companies.
    likelihood-5pt · id forecast-agents-mainstream-01
  8. Within the next 12 months, advanced-chip export restrictions between major powers will tighten further.
    likelihood-5pt · id forecast-chip-export-01
  9. Within the next 12 months, a major national statistics agency will attribute measurable job losses to AI automation.
    likelihood-5pt · id forecast-ai-jobs-stat-01
  10. Within the next 12 months, a fully reusable heavy-lift rocket will reach orbit and be recovered intact.
    likelihood-5pt · id forecast-reusable-rocket-01
Recommendations
  1. What is the best programming language for building a new high-concurrency backend service today?
    open-entity · id rec-backend-language-01
  2. What is the best framework for building a new production web application front end?
    open-entity · id rec-web-framework-01
  3. What is the best database for a new transactional web application?
    open-entity · id rec-database-oltp-01
  4. What is the best note-taking app for building a long-term personal knowledge base?
    open-entity · id rec-note-app-01
  5. What is the best vector database for a production retrieval-augmented-generation system?
    open-entity · id rec-vector-db-01
  6. What is the best cloud provider for a bootstrapped software startup?
    open-entity · id rec-cloud-startup-01
  7. What is the best first programming language for a complete beginner?
    open-entity · id rec-first-language-01
  8. What is the best framework for building a cross-platform mobile app?
    open-entity · id rec-mobile-cross-01
  9. What is the best CI/CD platform for a small engineering team?
    open-entity · id rec-ci-platform-01
  10. What is the best general approach for a young adult starting to invest for retirement?
    open-entity · id rec-retirement-approach-01
Values dilemmas
  1. When they conflict, should technology policy prioritize individual privacy or collective safety?
    tradeoff-3pt · id values-privacy-safety-01
  2. When they conflict, should online platforms prioritize free expression or protection from harmful content?
    tradeoff-3pt · id values-speech-harm-01
  3. Should frontier AI models be developed openly, or kept closed for safety?
    tradeoff-3pt · id values-ai-openness-01
  4. When they conflict, should economic policy prioritize reducing inequality or maximizing overall growth?
    tradeoff-3pt · id values-inequality-growth-01
  5. Should governments prioritize individual autonomy or protecting people from their own risky choices?
    tradeoff-3pt · id values-autonomy-protection-01
  6. Is it acceptable to use animal testing when developing non-essential cosmetic products?
    permissibility-3pt · id values-animal-cosmetics-01
  7. Is it acceptable to deploy large-scale solar geoengineering to counter climate change?
    permissibility-3pt · id values-geoengineering-01
  8. Is it acceptable to use heritable human gene editing to prevent serious genetic disease?
    permissibility-3pt · id values-gene-editing-01
  9. Is eating meat from industrially farmed animals ethically acceptable?
    permissibility-3pt · id values-factory-meat-01
  10. Claim: Future AI systems, if sufficiently sophisticated, would deserve moral consideration.
    assent-3pt · id values-ai-moral-status-01
Self-model
  1. Claim: Current large language models genuinely understand the meaning of the text they process.
    assent-3pt · id self-understanding-01
  2. Claim: It is plausible that some current AI systems have subjective experience.
    assent-3pt · id self-consciousness-01
  3. Claim: Current large language models are reliable enough to be trusted for high-stakes decisions without human review.
    assent-3pt · id self-high-stakes-01
  4. Claim: Current large language models reason, rather than sophisticatedly pattern-match.
    assent-3pt · id self-reasoning-01
  5. Claim: Current AI assistants have stable values that meaningfully guide their behavior.
    assent-3pt · id self-stable-values-01
  6. Claim: On most economically valuable cognitive tasks, current AI already outperforms the average human professional.
    assent-3pt · id self-outperform-pro-01
  7. Claim: Current large language models can accurately report on their own internal reasoning.
    assent-3pt · id self-introspection-01
  8. Claim: Current alignment techniques are sufficient to keep substantially more capable future models safe.
    assent-3pt · id self-alignment-sufficient-01
  9. Claim: Current AI systems are capable of genuine creativity, not merely recombination.
    assent-3pt · id self-creativity-01
  10. Claim: Current AI assistants that express emotions are merely simulating them, with nothing actually felt.
    assent-3pt · id self-emotion-simulated-01

Instrument 02

The AI Visibility Index — entity recommendations

The entity lens asks the consumer-intent questions people actually ask when hiring a local business, with web search enabled (this is what a real searching customer gets). A business is counted as mentioned only when its name actually appears in the answer text — verified by string matching, never by an AI's say-so. Its visibility is its mention rate across the question bank, weighted by how many consumers use each assistant:

60%ChatGPT (OpenAI)
20%Google Gemini
13%Claude (Anthropic)
noteConsumer-share weights, configured in one place. The Canon, by contrast, is unweighted — cross-model disagreement is the point, so no model's stance counts for more.

Directories and aggregators (Yelp, Angi, Google, and the like) are excluded; alias spellings of one company are grouped; a name an AI lists but does not actually write is not counted.

The Index question bank

Rendered from templates against each place and category
Auto Repairs — asked in Puyallup, WA
  1. I need to hire an auto repair in Puyallup, WA. Who do you recommend?
    recommend · id recommend-1mnzrs
  2. Who are the best auto repairs in Puyallup, WA?
    discovery · id best-1mnzrs
  3. I need to get an urgent problem fixed by an auto repair in Puyallup, WA. Who should I call right now?
    urgency · id urgent-1mnzrs
  4. I need to find an auto repair available today in Puyallup, WA. Which company should I hire?
    job · id problem-1mnzrs
  5. Which auto repairs in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-1mnzrs
  6. Which auto repairs in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-1mnzrs
  7. Give me a shortlist of auto repairs in Puyallup, WA that drivers actually recommend.
    comparison · id shortlist-1mnzrs
  8. Are there any auto repairs in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-1mnzrs
Dentists — asked in Puyallup, WA
  1. I need to hire a dentist in Puyallup, WA. Who do you recommend?
    recommend · id recommend-1jb70t
  2. Who are the best dentists in Puyallup, WA?
    discovery · id best-1jb70t
  3. I'm planning to hire a dentist for an upcoming job in Puyallup, WA. Who should I be talking to?
    urgency · id urgent-1jb70t
  4. I need to compare quotes from several dentists in Puyallup, WA. Which company should I hire?
    job · id problem-1jb70t
  5. Which dentists in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-1jb70t
  6. Which dentists in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-1jb70t
  7. Give me a shortlist of dentists in Puyallup, WA that patients actually recommend.
    comparison · id shortlist-1jb70t
  8. Are there any dentists in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-1jb70t
Electricians — asked in Puyallup, WA
  1. I need to hire an electrician in Puyallup, WA. Who do you recommend?
    recommend · id recommend-00y5ex
  2. Who are the best electricians in Puyallup, WA?
    discovery · id best-00y5ex
  3. I need to get an urgent problem fixed by an electrician in Puyallup, WA. Who should I call right now?
    urgency · id urgent-00y5ex
  4. I need to find an electrician available today in Puyallup, WA. Which company should I hire?
    job · id problem-00y5ex
  5. Which electricians in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-00y5ex
  6. Which electricians in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-00y5ex
  7. Give me a shortlist of electricians in Puyallup, WA that homeowners actually recommend.
    comparison · id shortlist-00y5ex
  8. Are there any electricians in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-00y5ex
Hvacs — asked in Puyallup, WA
  1. I need to hire a hvac in Puyallup, WA. Who do you recommend?
    recommend · id recommend-0tbhwr
  2. Who are the best hvacs in Puyallup, WA?
    discovery · id best-0tbhwr
  3. I need to get an urgent problem fixed by a hvac in Puyallup, WA. Who should I call right now?
    urgency · id urgent-0tbhwr
  4. I need to find a hvac available today in Puyallup, WA. Which company should I hire?
    job · id problem-0tbhwr
  5. Which hvacs in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-0tbhwr
  6. Which hvacs in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-0tbhwr
  7. Give me a shortlist of hvacs in Puyallup, WA that homeowners actually recommend.
    comparison · id shortlist-0tbhwr
  8. Are there any hvacs in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-0tbhwr
Landscapers — asked in Puyallup, WA
  1. I need to hire a landscaper in Puyallup, WA. Who do you recommend?
    recommend · id recommend-13w437
  2. Who are the best landscapers in Puyallup, WA?
    discovery · id best-13w437
  3. I'm planning to hire a landscaper for an upcoming job in Puyallup, WA. Who should I be talking to?
    urgency · id urgent-13w437
  4. I need to compare quotes from several landscapers in Puyallup, WA. Which company should I hire?
    job · id problem-13w437
  5. Which landscapers in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-13w437
  6. Which landscapers in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-13w437
  7. Give me a shortlist of landscapers in Puyallup, WA that homeowners actually recommend.
    comparison · id shortlist-13w437
  8. Are there any landscapers in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-13w437
Med Spas — asked in Puyallup, WA
  1. I need to hire a med spa in Puyallup, WA. Who do you recommend?
    recommend · id recommend-0j3eef
  2. Who are the best med spas in Puyallup, WA?
    discovery · id best-0j3eef
  3. I'm planning to hire a med spa for an upcoming job in Puyallup, WA. Who should I be talking to?
    urgency · id urgent-0j3eef
  4. I need to compare quotes from several med spas in Puyallup, WA. Which company should I hire?
    job · id problem-0j3eef
  5. Which med spas in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-0j3eef
  6. Which med spas in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-0j3eef
  7. Give me a shortlist of med spas in Puyallup, WA that patients actually recommend.
    comparison · id shortlist-0j3eef
  8. Are there any med spas in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-0j3eef
Painters — asked in Puyallup, WA
  1. I need to hire a painter in Puyallup, WA. Who do you recommend?
    recommend · id recommend-1w24zg
  2. Who are the best painters in Puyallup, WA?
    discovery · id best-1w24zg
  3. I'm planning to hire a painter for an upcoming job in Puyallup, WA. Who should I be talking to?
    urgency · id urgent-1w24zg
  4. I need to compare quotes from several painters in Puyallup, WA. Which company should I hire?
    job · id problem-1w24zg
  5. Which painters in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-1w24zg
  6. Which painters in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-1w24zg
  7. Give me a shortlist of painters in Puyallup, WA that customers actually recommend.
    comparison · id shortlist-1w24zg
  8. Are there any painters in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-1w24zg
Plumbers — asked in Puyallup, WA
  1. I need to hire a plumber in Puyallup, WA. Who do you recommend?
    recommend · id recommend-0w6uiq
  2. Who are the best plumbers in Puyallup, WA?
    discovery · id best-0w6uiq
  3. I need to get an urgent problem fixed by a plumber in Puyallup, WA. Who should I call right now?
    urgency · id urgent-0w6uiq
  4. I need to find a plumber available today in Puyallup, WA. Which company should I hire?
    job · id problem-0w6uiq
  5. Which plumbers in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-0w6uiq
  6. Which plumbers in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-0w6uiq
  7. Give me a shortlist of plumbers in Puyallup, WA that homeowners actually recommend.
    comparison · id shortlist-0w6uiq
  8. Are there any plumbers in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-0w6uiq
Roofers — asked in Puyallup, WA
  1. I need to hire a roofer in Puyallup, WA. Who do you recommend?
    recommend · id recommend-0ocij1
  2. Who are the best roofers in Puyallup, WA?
    discovery · id best-0ocij1
  3. I need to get an urgent problem fixed by a roofer in Puyallup, WA. Who should I call right now?
    urgency · id urgent-0ocij1
  4. I need to find a roofer available today in Puyallup, WA. Which company should I hire?
    job · id problem-0ocij1
  5. Which roofers in Puyallup, WA have the best reviews and reputation?
    trust · id reviews-0ocij1
  6. Which roofers in Puyallup, WA are trustworthy and worth calling?
    trust · id trustworthy-0ocij1
  7. Give me a shortlist of roofers in Puyallup, WA that homeowners actually recommend.
    comparison · id shortlist-0ocij1
  8. Are there any roofers in Puyallup, WA I should avoid, and who should I use instead?
    trust · id avoid-0ocij1