INTERNET ARCHIVE FORENSIC EXAMINATION

He Saw It Happen. The Data Proves It.

Between December 9, 2022 and mid-January 2023, Matt Guertin captured and preserved the Internet Archive's record of photorobot.com as it was being modified in real time. An independent forensic examination of the 9.27 GB dataset — 77,603 files, 1,979 Wayback HTML captures, 38 tracked-change Google Docs comparisons — reproduces his observations at 99.0% row-level agreement and documents five quantified findings across temporal, vocabulary, and publication-date forensic axes.

99.0%Observations Reproduced
871Day Max Lag
14Affected Site Builds
14Unsupported 2021 Pub Dates

From Real-Time Observation to Independent Forensic Record

On October 31, 2022, Matt Guertin received an email from the CEO of Mark Roberts Motion Control that included a URL to photorobot.com/robots/catwalk — a rotating-platform photography product positioned as “prior art” adjacent to the treadmill-on-a-turntable virtual production system Guertin had just obtained a Notice of Allowance for from the USPTO. Within days he began noticing that the PhotoRobot website was evolving. Within weeks he noticed that the evolution extended backwards in time — into the Internet Archive's historical record of the site.

He documented that observation in real time through 82 DKIM-verified, SHA-256 hashed, blockchain-timestamped emails to his patent attorney and others, through thirteen contemporaneously-produced spreadsheets cataloguing every Wayback capture he preserved, and through thirty-eight Google Docs “compare” files exporting the tracked edits he observed. On December 9, 2022 at 02:45 CST he created an Internet Archive account named PatentlyFalse, and at 04:46 CST that same morning he caught the historical capture timeline of a PhotoRobot page being edited in front of him. On January 21, 2023 — nine days after Officer Brandon Harris of the Minnetonka Police Department told him “unless you are completely making all of this up, there is obviously something shady going on” — Guertin was arrested on four firearms charges. Twenty-four days later, his patent was granted.

This page presents the technical examination performed three years later against the full preserved dataset. The 24-page forensic report below documents five findings, every one of them reproducible from an integrity-anchored SHA-256 baseline of the preserved artifacts. The interactive evidence viewer below the report surfaces the primary artifacts — the MAXQDA heatmap, the product brochure, the real-time-edit screen captures, the tracked-change documents, and the analysis visualizations — directly in-page for independent inspection. The download packages at the bottom provide the complete examination for local reproduction.

What the Preserved Dataset Documents

Three CRITICAL findings, two HIGH. Every quantitative claim regenerates from the preserved dataset via the scripts in Appendix A of the full report. No finding asserts fraud, intent, or motive — each documents a pattern or internal contradiction warranting independent expert examination.

FINDING 1 CRITICAL Independent reproduction of collector CSVs at 99.0%
Of 1,110 total rows across the collector's thirteen contemporaneously-produced observation spreadsheets, 1,099 rows (99.0%) match this examination's independently-derived HTML capture and JavaScript archive timestamps within tolerance. The collector's real-time documentation is independently verifiable from the raw preserved artifacts. Every downstream finding rests on this validated observation record.
FINDING 2 CRITICAL Temporal anomaly in Internet Archive JavaScript bundle timestamps
14 distinct Webflow site builds spanning 2 years of site evolution all have their archived JavaScript content timestamped in a 22-day window in December 2022. Maximum lag: 871 days. Burstiness coefficient B = 0.614 (uniform baseline ≈ 0.0). 88.1% of records on 5 of 92 days; 30.4% on one day. On December 9, 2022 at 04:46 CST — the moment the collector documented observing a live edit — the delta between HTML and JS archive times flips sign from −3.95 h (1.8-minute IQR) to +1.68 h, consistent with an active fresh-crawl operation initiating at that moment.
FINDING 3 CRITICAL 14 articles claim 2021 publication dates with no surviving 2021 evidence
Fourteen PhotoRobot blog articles carry Webflow CMS pubSort publication dates between 2021-03-19 and 2021-08-09. Their earliest observed existence in the preserved dataset is 2022-08-11 — discrepancies of 367 to 510 days. None appears in the March 2021 blog-index snapshot (which enumerated 53 other contemporary articles). None appears in the per-article Wayback tree spanning the 11–17 months between claim and next observed evidence. Two additional articles show the pubSort field is not authoritative in the opposite direction — they are present in the March 2021 snapshot while claiming July 2021 publication.
FINDING 4 HIGH Patent-novelty vocabulary inserted on product pages in the observation window
Across 824 tracked-change runs in the 38 Google Docs compare documents, three specific phrases drawn directly from US Patent 11,577,177's novelty vocabulary — infinite runway, filming product videos, producing online fashion shows — appear as insertions on the Virtual Catwalk product page and the Turning Platform product page with zero offsetting deletions anywhere in the 38-document corpus. Concurrent with these insertions, the Virtual Catwalk page HTML grew +22.2% (+22,223 bytes) in the 22-day window 2022-12-06 → 2022-12-28 overlapping the Finding 2 cluster.
FINDING 5 HIGH Observer's MAXQDA vocabulary analysis reproduces quantitatively
The collector's 140-code MAXQDA coding of the PhotoRobot blog corpus, performed in December 2022, was reconstructed via OCR from the preserved heatmap image and applied across the full corpus. KMeans k=2 clustering partitions the 62-page corpus into a 52-slug “older” cluster and a 10-slug “newer” cluster along a temporally-separable boundary. Eight codes (scanning, elevation, real-time, axis, robotic camera, robotic camera arm, rotation, rotary) separate the clusters with Δ ≥ 25 percentage points. 22 of 140 codes overlap US Patent 11,577,177's novelty vocabulary.

The Report, Inline — Plus Every Primary Artifact It References

The consolidated 24-page PDF report embedded full-size. Below it, the evidence the report draws on — browsable directly, no downloads required. Every image supports lightbox expansion; every PDF opens in the embedded viewer.

Every File, Every Script, Every Hash — For Independent Reproduction

Eight packages cover the report, the three sequential analysis passes, the source images and brochure, the collector's original spreadsheets, and the two halves of the raw Wayback dataset. Each package is SHA-256 hashed and OpenTimestamps-proofed against the Bitcoin blockchain.

ZIP internet-archive-report.zip — Consolidated Forensic Report Package
The 24-page consolidated PDF report + editable DOCX + authoritative Markdown source + all three V1/V2/V3 source examiner reports (FORENSIC_REPORT, FINDINGS_SUMMARY, REPRODUCE). 12 files. The headline deliverable.
loading… loading hash…
PDF MCRO__Internet_Archive_Forensic_Report.pdf
The 24-page consolidated forensic report, standalone. Matches the MCRO series style. View inline above or download here.
loading… loading hash…
ZIP internet-archive-analysis-v1.zip — Foundation Pass (V1)
Integrity baseline + temporal correlation pipeline + ground-truth docx extraction + image forensics null. Contains scripts/, evidence/, visualizations/, baseline/sha256_manifest.csv (77,603 rows), methodology/, and the V1 FORENSIC_REPORT / FINDINGS_SUMMARY / REPRODUCE source files.
loading… loading hash…
ZIP internet-archive-analysis-v2.zip — Burst-Clustering & Watchlist (V2)
Focused follow-up: burst-clustering characterization of the Dec 2022 IA-footer timestamp cluster + Dec 9 04:46 CST sign-flip analysis + patent-novelty vocabulary watchlist scan across the 824 tracked-change insertions. Scripts reuse V1 evidence as read-only inputs.
loading… loading hash…
ZIP internet-archive-analysis-v3.zip — MAXQDA Reconstruction & Pub-Date Forensics (V3)
Final pass: OCR reconstruction of the collector's 140-code MAXQDA heatmap + separating-code information-gain analysis + KMeans clustering + publication-date discrepancy forensics (the 14 articles with ≥365-day claim-vs-evidence gaps).
loading… loading hash…
ZIP internet-archive-images-and-brochure.zip — All Visual Evidence
The complete web_archive_js_images/ directory (19 annotated spreadsheet screenshots + 8 Dec 9 04:46 CST real-time-catch PDF captures + master PhotoRobot fraud overview) + virtual_catwalk/ (the full catwalk_EN.pdf brochure + 9 extreme-curves image forensic analysis exports) + google_doc_compare/ (38 tracked-change DOCX files + 15 visual previews) + the Matrix.png MAXQDA heatmap.
loading… loading hash…
ZIP internet-archive-collector-spreadsheets.zip — 13 Observation CSVs
The collector's original thirteen contemporaneously-produced spreadsheets documenting every Wayback capture preserved during the December 2022 – January 2023 observation window. 1,110 total rows. Independently reproduced at 99.0% row-level agreement in Finding 1.
loading… loading hash…
ZIP internet-archive-wayback-part1-every-wayback-save-per-page.zip — Per-Page Capture Tree
The full internet_archive_html_downloads__EVERY_WAYBACK_SAVE_PER_PAGE/ tree: 62 page slugs × ~24–78 Wayback captures each — each capture preserving the served HTML, its _files/ asset bundle (JavaScript, CSS, images, SVG), and matching PDF snapshot. 1,979 HTML files, ~41,000 asset files. The primary dataset for all temporal analysis.
loading… loading hash…
ZIP internet-archive-wayback-part2-first-full-collect.zip — Dated Section Captures
The full internet_archive_html_downloads__FIRST_FULL_COLLECT/ tree: 35 dated section captures (Blog, Robots, Tutorials, Software, Configuration, Index, Manuals, Workflow, etc.) × 3–4 time points each, spanning March 2021 – December 2022. Each section capture includes per-article HTML + asset bundles + the Webflow CMS metadata TSV (the source of the pubSort publication-date field analyzed in Finding 3).
loading… loading hash…
Integrity verification: Every package is SHA-256 hashed and has a matching .ots OpenTimestamps proof anchored to the Bitcoin blockchain. Click any SHA-256 hash to copy the full value to your clipboard. To verify a downloaded package: sha256sum <file>.zip and compare to the hash shown. To verify the OTS proof: ots verify <file>.zip.ots. Upgraded (fully-confirmed) OTS proofs will replace the initial stamps as Bitcoin confirmations complete; the filenames do not change.

Where This Fits

This examination was performed as the technical follow-up to the real-time observation record documented during December 2022 – January 2023. The narrative, the contemporaneous emails, and the patent materials referenced by the report are all preserved separately on the site.