Hoeflin Power Test Exposure Watch

Hoeflin Power Test: Public Exposure Audit

Version 2.0 — Revised August 4, 2026

DisclaimerThis document records exposure evidence only. It does not publish answer keys, verified answer values, solution steps, decisive methods, or answer drawings. Redacted content is indicated where applicable. The goal is to document exposure for research-validity purposes, not to facilitate further compromise.
Download:

The original audit document is the fixed 4 August 2026 audit, served exactly as issued. All later daily-search findings, classification changes, and corrections are consolidated in the separate Updates document. The page below renders the audit as HTML for reading; changes marked ▲ reflect post-audit updates and are not part of the original document.

Contents

  1. 1. Executive Report
  2. 2. 36-Item Classification Matrix
  3. 3. Predecessor Maps
  4. 4. Source Register
  5. 5. Search Log
  6. 6. Limitations & Quality Control
  7. 7. Corrections
  8. Appendix A: Research Prompt
  9. Appendix B: Draft Notification Email

1. Executive Report

This observatory monitors publicly discoverable online content for evidence that items of the Hoeflin Power Test have been exposed in ways that may compromise their psychometric utility. Four exposure classes are tracked: verbatim prompt text, method or hint exposures, candidate answers, and verified solutions. Each finding is assessed for confidence, source type, and whether it warrants reclassification of the affected items.

The primary exposure vectors identified are: (1) full-test text reproductions on document-hosting platforms (Scribd, CNBlogs, Calvo OMNI compilation) and the official Power Test page; (2) predecessor-test carryover from the Mega, Titan, and Ultra tests — 28 of the 36 Power Test items have identifiable predecessors in these earlier Hoeflin instruments, and predecessor materials are widely searchable; and (3) forum and Usenet discussions from 2001–2023 that address specific items at method-hint or candidate-answer level.

Baseline — 4 August 2026
High25 itemsMedium6 itemsLow5 items
Current — 13 August 2026
High27 itemsMedium6 itemsLow3 items

Three items changed classification since the 4 August baseline: P9 (Medium→High, BrainDen thread), P16 (Low→High, New Indian Express prompt reproduction), and P27 (Low→Medium, Math SE thread re-reviewed). No item decreased in classification.

2. 36-Item Classification Matrix

Redacted: No answer values, solution steps, or decisive methods are included. Evidence descriptions record only the existence and nature of exposure.

ItemPredecessorBaselineCurrentEvidence Summary (redacted)
P1MegaM27HighHighMega scan; cube-colouring discussion (2001); ScribbleHub (2026)
P2TitanT25HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P3MegaM28MediumMediumMega scan — predecessor carryover (medium confidence)
P4TitanT34HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P5UltraU43MediumMediumUltra material — predecessor carryover (medium confidence)
P6UltraU44MediumMediumUltra material — predecessor carryover (medium confidence)
P7TitanT28HighHighTitan material; Physics Forums method-hint thread; 2005 rec.puzzles
P8MegaM34HighHighMega scan; FerrariChat forum discussion (2026)
P9MegaM31MediumHigh▲ Upgraded 13 Aug: Mega scan + BrainDen substantial-hint thread
P10TitanT31HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P11TitanT32HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P12MegaM35HighHighMega scan; Metamerist blog candidate-answer analysis
P13TitanT33HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P14MegaM38HighHighMega scan; Bachet academic literature (method context)
P15MegaM41HighHighMega scan; FreeMathHelp substantial-hint thread
P16MegaM29LowHigh▲ Upgraded 13 Aug: Mega scan + New Indian Express prompt reproduction
P17MegaM37MediumMediumMega scan; FreeMathHelp method-hint thread
P18MegaM39HighHighMega scan — predecessor carryover
P19TitanT29HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P20MegaM32HighHighMega scan; ClassAce prompt reproduction (Aug 2026)
P21TitanT38HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P22TitanT39HighHighTitan material; FZ.se forum discussion (Aug 2026)
P23TitanT40HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P24TitanT41HighHighTitan material; 2005 rec.puzzles Titan-answer discussions
P25TitanT36HighHighTitan material; Physics Forums thread (Aug 2026)
P26MegaM36HighHighMega scan; HN/GameDev/Puzzling SE; BBO; Rick Rosner; In-Sight (Aug 2026)
P27TitanT35LowMedium▲ Upgraded 13 Aug: Math SE method-hint discussion (re-reviewed)
P28MegaM33MediumMediumMega scan; FreeMathHelp; FZ.se forum discussion
P29TitanT30HighHighTitan material; 2009 sequential-movement candidate-answer discussion
P30UltraU40HighHighUltra material; Math SE substantial method-hint discussion
P31TitanT37HighHighTitan material; Math SE candidate-answer thread; DeepMath dataset
P32UltraU50HighHighUltra material; Reddit discussion; Math Doctors decisive hint; PM-request evidence
P33UltraU52LowLowUltra material only — predecessor carryover (low confidence)
P34UltraU54LowLowUltra material only — predecessor carryover (low confidence)
P35UltraU53LowLowUltra material only — predecessor carryover (low confidence)
P36—HighHighFull-test publications; Noesis 217; 1996 Noesis journal; Washington Post 1988

▲ = classification changed since 4 Aug 2026 baseline

3. Predecessor Maps

Items in the Mega, Titan, and Ultra tests that map directly to Power Test items. Prompt text for the predecessor item is identical or near-identical to the corresponding Power Test item; public exposure of predecessor material constitutes predecessor-carryover prompt exposure for the Power Test item.

Mega → Power

M27P1
M28P3
M29P16
M31P9
M32P20
M33P28
M34P8
M35P12
M36P26
M37P17
M38P14
M39P18
M41P15
13 items

Titan → Power

T25P2
T28P7
T29P19
T30P29
T31P10
T32P11
T33P13
T34P4
T35P27
T36P25
T37P31
T38P21
T39P22
T40P23
T41P24
15 items

Ultra → Power

U40P30
U43P5
U44P6
U50P32
U52P33
U53P35
U54P34
7 items

Power Test item P36 has no documented predecessor in the Mega, Titan, or Ultra tests.

4. Source Register

Public source titles, URLs, and dates only. No answer content is recorded. Where no single stable page could be verified, the source is recorded by title and a descriptive status instead of a link.

Source / TitleURL / IdentifierDateItemsTypeAccess
Official Hoeflin Power Test pagehttps://megasociety.org/admission/power/c. 1995P1–P36 (all)Official/primary sourcePublic
1996 Noesis (Mega Society Journal)https://megasociety.org/noesis/121.pdf1996P1–P36 (all)Official/primary; document repositoryPublic (PDF)
Washington Post feature articlehttps://www.washingtonpost.com/archive/local/1988/11/28/is-176-iq-enough-in-white-house/c0535981-650c-4c33-a233-78babc3006e0/1988Multiple itemsNews/newspaper; historical compromiseArchive
April 1985 Mega Test scanhttps://www.docdroid.net/file/download/DvwMtpk/megatest-april-1985-omni-mag-pdf.pdf1985P1,P3,P8,P9,P12,P14,P15,P16,P17,P18,P20,P26,P28Predecessor; document repository; archivedPublic (PDF)
Titan Test materialshttps://www.scribd.com/doc/253577266/Mega-Society-Titan-Testc. 1990P2,P4,P7,P10,P11,P13,P19,P21–P25,P27,P29,P31Predecessor; document repositoryGated/partial
Titan Test mirror (Scribd, secondary upload)https://www.scribd.com/document/372410805/74196672-Mega-Society-Titan-Test-pdfUnverifiedc. 1990P2,P4,P7,P10,P11,P13,P19,P21–P25,P27,P29,P31Predecessor; document repository; mirrorGated/partial
Ultra Test official page/normshttps://megasociety.org/admission/ultra/c. 1993P5,P6,P30,P32,P33,P34,P35Predecessor; official/primaryPublic
Noesis Issue 217https://megasociety.org/noesis/217.pdfc. 2010P36 (context)Official/primary; document repositoryPublic (PDF)
2005 rec.puzzles — Titan-answer discussionshttps://groups.google.com/g/rec.puzzles/c/_B_FTJxS_Ys2005Titan-predecessor items → P2,P4,P7,P10,P11,P13,P19,P21–P25,P27,P29,P31Usenet; candidate answer; public answerPublic (Google Groups)
2001 online cube-colouring discussionhttps://groups.google.com/g/rec.puzzles/c/Pk756QuOork2001P1Forum; method hintPublic (Google Groups)
2009 Mega Test discussionhttps://groups.google.com/g/rec.puzzles/c/hpC-_IgMmZw2009P8,P12,P17,P20,P26Forum; discussionPublic (Google Groups)
2009 sequential-movement discussionhttps://groups.google.com/g/rec.puzzles/c/6_LEKQOk5ks2009P29Forum; candidate answerPublic (Google Groups)
2009 Reddit Power Test discussionhttps://web.archive.org/web/20260814005758/https://www.reddit.com/r/math/comments/ai97q/the_hoeflin_power_test_very_difficult_iqlogicmath/2009P1,P7,P12,P26Reddit; discussionPublic (archived)
Physics Forums — P7 threadhttps://www.physicsforums.com/threads/octahedron-problem-how-many-color-patterns.288312/c. 2012P7Forum; method hintPublic
Metamerist blog — P12 analysishttps://metamerist.blogspot.com/2007/08/hard-puzzles-like-butter.htmlc. 2008P12Blog; candidate answerPublic/archive
FreeMathHelp forums — P15https://www.freemathhelp.com/forum/threads/volumetric-displacement-question-3-000-barrels-of-toxic-chemicals.109742/c. 2013P15Forum; method hint; substantial hintPublic
FreeMathHelp forums — P17https://www.freemathhelp.com/forum/threads/dodecahedron-scraps.129053/c. 2014P17Forum; method hintPublic
FreeMathHelp forums — P28https://www.freemathhelp.com/forum/threads/visualization-problem-dots-and-lines.109715/c. 2015P28Forum; method hintPublic
Math Doctors — plane-cutting articlehttps://www.themathdoctors.org/cutting-up-space-using-n-planes/c. 2006P32Academic source; decisive hintPublic
Mathematics Stack Exchange — P27https://math.stackexchange.com/questions/2424226/interpenetrating-solidsc. 2013P27Stack Exchange; method hintPublic
Mathematics Stack Exchange — P30https://math.stackexchange.com/questions/4689341/number-of-pair-of-edges-of-a-dodecahedron-up-to-rotational-symmetryc. 2015P30Stack Exchange; substantial hintPublic
Mathematics Stack Exchange — P31https://math.stackexchange.com/questions/1371864/ten-marbles-put-in-a-box-colour-of-each-set-by-toss-of-a-fair-coin-you-draw-wc. 2016P31Stack Exchange; candidate answerPublic
Reddit — P32 discussionhttps://www.reddit.com/r/cognitiveTesting/comments/1h9fou5/sequence_puzzles_help/c. 2020P32Reddit; discussionPublic
Hacker News — P26 discussionhttps://news.ycombinator.com/item?id=14472265c. 2017P26Forum; method hintPublic
GameDev.net forums — P26 discussionhttps://gamedev.net/forums/topic/408053-what-is-your-iq/3716929/c. 2014P26Forum; method hintPublic
Puzzling Stack Exchange — P26https://puzzling.stackexchange.com/questions/91881/what-is-the-three-cube-problem-in-the-mega-testc. 2018P26Puzzle site; Stack Exchange; candidate answerPublic
CNBlogs — full Power Test reposthttps://www.cnblogs.com/kakrat/archive/2008/09/18/1293224.htmlc. 2010P1–P36 (all)Blog; prompt reproduction; duplicatePublic
Scribd mirror — full-test reproductionhttps://www.scribd.com/document/109126131/Web-Archive-Org-4c. 2008P1–P36 (all)Document repository; archived/mirrorGated/partial
Scribd mirror — Ronald Hoeflin's Power Testhttps://www.scribd.com/document/1020486307/Ronald-Hoeflin-s-Power-TestUnverifiedc. 2008P1–P36 (all)Document repository; archived/mirrorGated/partial
Calvo OMNI compilationhttps://www.scribd.com/document/402818061/Notes-Collected-by-Gabriel-F-Calvo-OMNI-IQ-Test-problems-2017-pdfc. 2005P1–P36 (all)Document repository; duplicateGated/partial
Bachet de Méziriac — weighing problem literaturehttps://arxiv.org/abs/1010.54861612 ff.P14Academic source; decisive hint; common methodPublic (academic)
Balanced number system academic literaturehttps://doi.org/10.1007/s12045-018-0749-1c. 1950 ff.P9Academic source; method hint; common methodPublic (academic)
Redvaldsen context paperhttps://www.mdpi.com/2624-8611/2/2/10c. 2010P7,P12,P26Academic source; method hintPublic (academic)
DeepMath public datasethttps://huggingface.co/datasets/Ujan/deepmath_trace_6/viewer/default/train2019P31Academic source; AI-generated materialPublic (dataset)
Reddit/SlateStarCodex — Mega discussionhttps://www.reddit.com/r/slatestarcodex/comments/a38tik/the_mega_test_and_bullshitchristopher_langan/c. 2019Multiple Mega-predecessor itemsReddit; discussionPublic
Rick Rosner — P26 referencehttps://rickrosner.org/2025/09/22/ask-a-genius-1515-space-dread-crypto-skepticism-and-political-math/c. 2015P26DiscussionPublic
ClassAce (classace.io) — P20https://www.classace.io/answers/if-27-identical-cubical-chunks-of-cheese-are-piled-together-to-form-a-cubica-stack-as-depicted-below-what-is-the-maximum-number-of-these-cheese-chunks-through-which-a-mouse-of-negligible-size-could-mu2023P20Blog/educational; prompt reproductionPublic
FZ.se — P22, P28 threadshttps://www.fz.se/forum/trad/47508-tva-svara-uppgifter-kan-nagon2008P22, P28Forum; discussionPublic
Physics Forums — P25 thread (Aug 2026)https://www.physicsforums.com/threads/what-iq-does-solving-the-einstein-puzzle-indicate.380251/2010P25Forum; method hint; substantial hintPublic
Bridge Base Online — P26 discussionhttps://www.bridgebase.com/forums/topic/14816-puzzles/2006P26Forum; discussionPublic
Reddit — P32–P35 private-answer request evidencehttps://www.reddit.com/r/cognitiveTesting/comments/1pva0f0/core_results_and_analysis/c. 2023P32,P33,P34,P35Reddit; answer request; private answer request; evidence of private sharingPublic (request visible)
Scribd full-test reproduction (Aug 2026 specific upload)https://www.scribd.com/document/109126131/Web-Archive-Org-4c. 2015P1–P36 (all)Document repository; prompt reproduction; archivedGated/partial
In-Sight Publishing — P26 referencehttps://in-sightpublishing.com/2026/05/28/fumfer-physics-37-from-iq-puzzles-to-physics-breakthroughs/2026P26Blog/academic; discussionPublic
BrainDen — P9 puzzle threadhttps://brainden.com/geometry-puzzles.htmc. 2011P9Puzzle site; method hint; substantial hintPublic
New Indian Express — P16 prompt reproductionhttps://www.newindianexpress.com/magazine/voices/2017/Jan/21/sharper-than-cards--and-houdini-like-to-boot-1561375.html2017P16News/newspaper; prompt reproductionPublic
ScribbleHub — P1 referencehttps://www.scribblehubforum.com/threads/iq-test.10568/2022P1Blog/fiction site; discussionPublic
FerrariChat forums — P8 discussionhttps://www.ferrarichat.com/forum/threads/critical-thinking.86230/Unverified2005P8Forum; discussionPublic
rec.puzzles, May 2023 — collaboration requesthttps://groups.google.com/g/rec.puzzles/c/ooxuvYQv8BEMay 2023Multiple (Power Test)Usenet; collaboration/assistance requestPublic (Google Groups)

5. Search Log

12 August 2026

Baseline Watch Established

  • • Formal monitoring commenced. All sources documented in the 4 August 2026 audit snapshot were logged into the observatory database.
  • • Baseline classification: 25 items High / 6 Medium / 5 Low.
  • • Search engines surveyed: Google, Bing, DuckDuckGo. Primary search terms: "Hoeflin Power Test" + item-type keywords and predecessor test names.
13 August 2026 — First Run

New Sources: P20, P22, P25, P26, P28, P32–P35

  • • ClassAce (classace.io): P20 prompt reproduction identified. Classified prompt/high.
  • • FZ.se Swedish forum: P22 and P28 discussion threads. Classified discussion/medium.
  • • Physics Forums: P25 substantial method-hint thread. Classified method-hint/high.
  • • Rick Rosner public content: P26 reference. Classified discussion/high.
  • • Bridge Base Online (bridgebase.com): P26 combinatorial discussion. Classified discussion/high.
  • • In-Sight Publishing: P26 journal reference. Classified discussion/high.
  • • Scribd: Specific full-test upload identified (all items). Classified prompt/high.
  • • Reddit: Public posts evidencing private-answer requests for P32–P35. Classified as tracking flag (answer-request/evidence-of-private-sharing). Changes-classification flag set.
  • • Indexed leads for P25 and P26 noted but not fully reviewed — logged as medium-confidence pending items.
  • • Running classification unchanged: 25H / 6M / 5L.
13 August 2026 — Second Run

Classification Upgrades: P9 (High), P16 (High), P27 (Medium)

  • • BrainDen (brainDen.com): P9 puzzle thread with substantial method-hint content. Upgrades P9 from Medium to High.
  • • New Indian Express: P16 prompt reproduction in mass-circulation national newspaper. Upgrades P16 from Low to High.
  • • ScribbleHub (scribblehub.com): P1 creative-writing platform reference. Classified discussion/high (no classification change; P1 already High).
  • • FerrariChat (ferrarichat.com): P8 recreational puzzle discussion. Classified discussion/high (no change).
  • • rec.puzzles (Usenet, May 2023): Public collaboration/assistance request for Power Test items. Classified collaboration-request/high.
  • • P27 Math SE thread re-reviewed: engagement with solution method is more direct than initially assessed. Upgraded from Low to Medium.
  • • Running classification: 27H / 6M / 3L.

6. Limitations & Quality Control

Search coverage. Web search engines index only a fraction of the public web. Gated or paywalled sources, scanned PDFs, dynamically-loaded content, private Discord or Slack channels, and email lists are not accessible to this search methodology. The exposure record should be treated as a lower bound.
URL resolution. Where a URL could not be confirmed with high confidence, the source is recorded by title, approximate date, and a descriptive non-link status only. Such sources should be verified by a researcher with appropriate database access before being treated as confirmed.
Private communications. Direct-message or private-channel answer sharing cannot be detected by public search. Evidence of such activity (e.g. Reddit PM-request threads) is logged when publicly visible, but the actual extent of private sharing is unknown and likely underrepresented.
Confidence calibration. Confidence levels (Low / Medium / High) reflect how directly the source's content corresponds to the Power Test item text and how reliably the source–item correspondence was verified. Confidence does not reflect the correctness of any candidate answer that may appear in a source.
Non-English sources. Non-English sources (FZ.se in Swedish; CNBlogs in Chinese; others) may contain additional material not fully reviewed. Language-specific search runs are recommended to supplement this audit.
Predecessor-carryover assumptions. The predecessor maps are based on item-text comparison. A small number of predecessor relationships were established from published documentation rather than direct item-text verification; these are noted in the individual findings records.

7. Corrections

Corrections to source attributions, item mappings, predecessor assignments, or classification assessments should be submitted to the observatory for review. Any correction that materially affects a classification will generate a new dated classification record and a corresponding daily-update log entry. Corrections do not retroactively alter the 4 August 2026 baseline snapshot; they are recorded as dated amendments with full change history preserved.

No corrections have been applied to the 4 August 2026 baseline as of 13 August 2026.

Appendix A: Research Prompt

The following prompt was used to guide AI-assisted search research during the August 2026 search runs. It is reproduced here for transparency and reproducibility.

"Search for public online sources — forums, blog posts, academic papers, puzzle sites, Stack Exchange, Reddit, Usenet archives, and document repositories — that discuss, reproduce, or provide hints toward solving items from the Hoeflin Power Test. For each source found, record: (1) the URL or best-available identifier, (2) the Power Test item number(s) affected, (3) the nature of the exposure (verbatim prompt, method hint, candidate answer, or solution), and (4) the approximate date of publication. Do not record or infer specific answer values, solution steps, or verifiable correct answers — record only the existence and nature of the exposure. Mark any URL you cannot verify with high confidence as unresolved and record that source by title and approximate date only."

Appendix B: Draft Notification Email

Draft notification for IQ society administrators and test committees.

Subject: Hoeflin Power Test — Public Exposure Audit Report (August 2026)

Dear [Society Administrator / Test Committee],

This communication accompanies the August 2026 Public Exposure Audit for the Hoeflin Power Test, conducted by the Hoeflin Power Test Exposure Watch observatory (thehoeflinpowertest.info).

Summary of findings as of 13 August 2026:

  • • 27 of 36 items — High exposure via publicly accessible sources.
  • • 6 items — Medium exposure.
  • • 3 items (P33, P34, P35) — Low exposure (Ultra predecessor material only).

Primary exposure vectors: (1) full-test text reproductions on Scribd, CNBlogs, and similar platforms; (2) predecessor-test carryover from the Mega, Titan, and Ultra tests; (3) Usenet and forum discussions from 2001–2023 addressing specific items at method-hint or candidate-answer level.

The audit does not publish answer keys or solution values. The full audit document, source register, and searchable findings database are available at https://thehoeflinpowertest.info/audit.

We recommend that administrators consult the findings before administering the Power Test in competitive or high-stakes contexts, and consider item-by-item exposure risk when interpreting scores.

Yours sincerely,
Hoeflin Power Test Exposure Watch Observatory
https://thehoeflinpowertest.info

Findings Archive

Browse all 49 documented findings, filter by item, exposure type, confidence, tag, or predecessor test.

Browse Findings

Methodology & Limitations

Detailed exposure-category definitions, confidence-level rubric, and candid account of what this observatory cannot track.

Read Methodology