The guest list, checked / Where she has a point
The places the record is on her side, or open, and every attack we wrote against our own work.
Two of her Hansell charts agree with each other. Her July 9 stream and the chart she showed on Stew Peters put "Bradley Hansell" on the same dates. Google says the same request made on the same day returns the same sample, so two matching screenshots can also mean they were taken the same day. Nobody, including us, knows when she took each one. We have asked for the capture dates.
She said "sampling" herself. On the July 16 stream with @danksterintel she described Trends as a sampling method, on air. That is correct. It is also why a blip that vanishes the next day is a sample wobble, not a deletion.
She disclosed the Ondas timing on her own slide. Her slide says nobody on that list worked at Ondas in the 2025 window. The open question is why "ONDAS" is in the search term and on the 2025 rows anyway.
She hedges, often. "Allegedly", "in my opinion", "not proof", "I'm not accusing anyone" are all on tape. This site quotes the hedge and the claim together wherever both exist, because a hedge does not change what a slide says.
She does name outliers and warn about common names. Five outliers out of 43, a July 9 warning that a plain common name proves nothing. The filter removes people from her list; it never tests the method.
Google's written warning about noise is recent. The passage this site quotes most, that one-off spikes "should not be interpreted as real search activity or evidence of individual searches", is not in archived copies of Google's Trends FAQ from August 13, 2025 or April 20, 2026. It is in the copy from May 16, 2026. Anyone who says Google added that wording after September 2025 is right about the date. The earlier page already said Trends is a sample and that low-volume terms show as zero, and a Google spokesperson gave the same explanation to ABC News in Australia in December 2025.
Multiple dates for one person are not automatically a contradiction. Two rows can be two strings or two loads. The point made here is narrower: each date was presented as exact and meaningful.
Stew Peters' words are not hers. "Hard data that has been cross-checked" and "irrefutable" are his. They are labelled as his wherever they appear.
@danksterintel is on several of these clips. See What @danksterintel did in July.
And the limit that cuts both ways: nobody can show her total number of tries. The tests on this site show the method; they do not give her count.
, on camera. In July he typed her Stew chart in live and it did NOT match (collab frame 1:04:40, P8.2). Ana herself said "it is done through a sampling method" (~1:05:45). He kept checking; that's the job. Show the frame.
"I said that in July. I was wrong, and here's how I found out": Daffy Duck vanished the same way (P10 #1).
Attack the method, not people. Give the 60-second do-it-yourself so viewers don't need to trust @danksterintel. "People she named are real people" - good-place close.
The cartoons vanish too. Daffy Duck Sep 5 (02:52) -> "not enough data" (17:59), 7 of 9 cartoon blips gone, 2 moved (P2 R4). 42 blips -> 24 gone, 17 moved in ~2 h. Researcher: "exactly what to expect with Google Trends data." Her own "within 24 hours" matches Google's own once-a-day sample description (2015).
That report is about "blog posts or YouTube videos"; the word Trends isn't on it. The Trends FAQ: "Unlike Google Trends, Autocomplete is subject to Google's removal policies" (since 2019). The Google AI answer on her own screen said Trends "does not delete historical search data". And no takedown order explains Daffy Duck.
(a) The 43 are the survivors of "hundreds and hundreds" of names - the 38/43 is calculated AFTER the sieve. (b) She picked the three days after looking ("as I narrowed in on these three dates"). (c) Each name was tried with titles, misspellings, other spellings, other windows - every try is another ticket (P9). (d) Our public numbers: 468 tries -> 5 exactly on her three days.
We also ran her exact strings and real Army-documented visitors: 0 hits on their visit days (P2 R3). Re-running her exact strings gives other dates (Hansell: her Sep 8 -> Aug 17 x3; May 25 live in July).
Same request, same day, same sample (Google, 2015) - checking the same day shows the same draw, not a real search. The test is checking a day or more later: July 16 and Oct 4 both failed (P8.2).
Our captures are on the classic Explore page (the "Try the new experience" banner is visible). Repeated loads within minutes give identical answers; differences follow TIME, not device (04 section C). @danksterintel's July run and her Stew chart were one day apart and different.
Trends measures "searches people make on Google," and a search is dated the day it's made. Nothing typed in 2026 can land on a 2025 date. Our cartoon blips are on 2025 dates.
Yes - that's the demonstration. We publish the whole count: 468 tried, 177 blipped, 5 on her three days. We show our denominator; she doesn't show hers.
Search hundreds and one in three still blips somewhere - that's dozens. Real visitors' names don't land on their real visit days. And a blip on the same name moves between loads (P8.2).
Google spokespeople said it to ABC and AAP in Dec 2025; Google's own economists wrote in 2009 that sampling "adds some additional noise"; researchers documented it in 2010, 2022, 2023, 2024. The "Autocomplete is subject to removal policies, unlike Trends" line has been there since 2019.
Then you can't trust Google Trends either - her whole case is Google's numbers. Plus independent sources: Univ. of Sydney biostatistician, Central Bank of Chile working paper, peer-reviewed journals, two fact-checkers.
Our control "Utah Valley University to Timpanogos hospital" blipped May 13, 2025 in the Tucson market. By her logic, Arizona had foreknowledge in May too. Hospitals' addresses are things people look up.
"Tucson (Sierra Vista) AZ" is a TV market of ~1.27M people, not a base. Visitors are ID-checked at the gate (REAL ID + NCIC-III, photo taken there); VIP bios come from the visitor's own office. Real visitors don't show.
GenAI.mil launched Dec 9 2025, after her 2025 dates. Google's deployment guide tells government customers to switch the Google-Search grounding off; the approved version uses Google Cloud's stored web index. Trends filters searches made by Google products.
Separate question, separate video. Even if every word he said were true, a blip can't add a name - the same blip shows up for Daffy Duck. (Harpole's complaint alleges travel records put Harpole in Texas - an allegation.)
Her: "this isn't just random noise ... so far past random" (wmLljWb4bcM 1:17:58); she named "the inside man" (1:28:26). Stew: "irrefutable" (02:07:41 / 02:09:07 / 02:14:52), "hard data that has been cross-checked. It has been verified." (00:49:48), "absolutely 0%" coincidence (00:50:53).
Correct her record precisely: she did search live (5 Sep 2026) and said "before it gets wiped ... I know it's going to disappear." She expects the blips to vanish. Never say "she never searches live."
The same five names gave three different date sets on three loads: Jul 10/15 (her chart), ~Jul 13/~Aug 2 (July 16 live), ~May 31 (Oct 4). Which one lines up?
No claims about her motive or money.