GTA VI, measured. Every study, every row, one page.
Grand Theft Auto VI launched on November 19, 2026. This page holds everything CULT+MATH has measured about it: what launch-day comments talked about, what fans said in their own words, which sources AI answers cite, what Take-Two spends and what people search for. Every row is published, with the method, the limits and the date it was taken.

The findings so far
| Date read | Finding | Section |
|---|---|---|
| Nov 20, 2026 | Fans talked about the wait, a joke and the creator they were watching. Lucia was named in 9 of 1,684 comments, Jason in 8. | Fans in their own words |
| Nov 20, 2026 | None of the eleven themes registered before launch reached 5% of 1,162 launch-day comments. Both registered predictions missed. | Launch day |
| Oct 24, 2026 | ChatGPT, Perplexity and Google, given one query string, cited zero domains in common. | AI answers |
| Oct 24, 2026 | Take-Two spent 19.34% of revenue on advertising in fiscal 2026, read from its SEC filings. | Publisher spend |
| Oct 24, 2026 | What people search for about GTA 6, 50 keywords with published volume. | Search demand |
Two companion studies have their own pages: whether GTA sells consoles and the twelve days other publishers left empty around launch. The two launch-day write-ups are on Sound Decisions: for fans and community teams and for brands.
Launch day: what 1,162 comments talked about
On November 19, 2026, GTA VI shipped. We drew ten of the fifty most-viewed "GTA 6" videos published that day, using a random seed written down before the draw, so no person picked them. Then we counted eleven themes, fixed three days before launch, across the top comments on all ten.
| Theme | Share of comments (low–high) | Leading term (count) | Pre-launch control (low) |
|---|---|---|---|
| Comparison with GTA V or RDR2 | 3.27–4.82% | gta 5 (38) | 4.67% |
| The wait | 1.81–3.53% | finally (21) | 2.33% |
| No PC version | 1.29–1.46% | pc (15) | 2.00% |
| The world and map | 0.69–1.12% | vice city (8) | 1.67% |
| Story and characters | 0.43–1.03% | lucia (5) | 6.33% |
| Memes and jokes | 0.34–0.69% | funny (4) | 2.17% |
| Price and value | 0.26–0.34% | price (3) | 1.17% |
| Online mode | 0.09–0.09% | online (1) | 0.33% |
| Downloads, servers and performance | 0.09–0.09% | download (1) | 1.17% |
| Bugs | 0.09–0.09% | crash (1) | 0.17% |
| Brands and ads | 0.09–0.09% | brand (1) | 0.67% |
No theme reached 5% of comments. The largest, the comparison with GTA V or Red Dead, reached 3.27%. The handful of comments the engine quoted behind it mix requests for a creator's GTA V series, nostalgia and some direct comparison, so the label may overstate how much of it is a comparison. Many of the quoted comments speak to the creator rather than about the game.
Every theme came in below its pre-launch level. Story and characters fell from 6.33% of comments on October's analysis videos to 0.43% on launch day. Two things changed between those corpora, not one: the videos and the day. So the drop says this launch-day audience discussed the game less, not that the launch caused it.
| Prediction | Prior | Result |
|---|---|---|
| A technical or price theme is among the top two | Roughly even (45–55%) | Did not happen. No PC version ranks third. |
| The largest theme is positive about the world, the story or the wait | Roughly even (45–55%) | Cannot score: the top theme is the GTA V comparison, its range overlaps the wait's, and the rule written before the fire scores that as unresolved. A miss in substance. |
Read every share as a floor. Themes are counted by English words, and the corpus is not all English: of the 45 comments the engine quoted, 12 or 13 are Hindi written in Latin script and 2 are Portuguese, even with the search set to the US and English. The leading terms carry sampling error too: "gta 5" is 2.4–4.5% at 95% confidence. The top three themes are an order, not precise shares.
Comment text is not stored. Terms are counted once per comment; a theme's high figure counts a comment that uses two of its terms twice. Comments are the top 200 per video by relevance, which skews to early comments. The method, the seed and the full draw are in the evidence record.
Launch day, in the fans' own words
The section above counted themes we chose before launch. This one starts from the words fans chose. We listed the most-used words on the first ten videos, named ten themes from them and from seven candidates written down beforehand, and then tested those themes on ten new videos drawn by a second published seed, so nothing was scored on the comments it was found in.
| Rank | Theme | Share of comments (low–high) | Leading term (count) |
|---|---|---|---|
| 1 (tie) | Relief that it's out | 4.81–6.41% | years (81) |
| 1 (tie) | “We got GTA VI before X” (a joke) | 3.38–10.27% | before gta (57) |
| 3 (tie) | Hype and praise | 2.61–4.22% | love (44) |
| 3 (tie) | Watching someone play | 2.55–4.99% | watching (43) |
| 3 (tie) | Comparison with GTA V or Red Dead | 1.78–3.15% | gta 5 (30) |
| 6 | Getting it: console, price | 1.48–2.61% | ps5 (25) |
| 7 | Asking the creator for more | 0.89–1.54% | please (15) |
| 8 (tie) | Characters by name | 0.53–1.01% | lucia (9) |
| 8 (tie) | Gameplay moments | 0.53–0.89% | car (9) |
| 10 | Aura | 0.12–0.18% | legend (2) |
Fans talked most about the moment and the person playing. Relief led on the word "years", which does two jobs: some comments are the 13-year wait, others tell a creator how long they have watched him. The plainer relief words are small: "waited" in 19 comments, "finally" in 8. A joke, "We got GTA VI before X", tied for first and appeared on both random draws. Of the eight comments the engine printed behind "love" and "watching", seven speak to the creator.
The game itself came last. Lucia was named in 9 comments and Jason in 8; police and heists in none. Three of the fourteen gameplay and character comments printed are bug reports. "Aura", which looked like a theme on the first ten videos, fell to one comment on the new ten.
| Candidate | Prior: a top-3 theme | Result |
|---|---|---|
| Relief that it's out | Likely (63–80%) | Hit, weakly: tied first on an ambiguous word |
| Asking the creator for more | Roughly even (45–55%) | Miss: seventh |
| Hype and praise | Roughly even (45–55%) | Tie at the cut |
| GTA V or Red Dead | Roughly even (45–55%) | Tie at the cut |
| Gameplay moments | Unlikely (20–45%) | Hit: stayed out |
| Characters by name | Very unlikely (5–20%) | Hit: stayed out |
| Getting it: console, price | Unlikely (20–45%) | Hit: stayed out |
These are comments on creators' own videos, so talk aimed at the creator is partly built in; this is not what all players think. Three of the ten videos are in Spanish and one in Arabic, and English and Hindi words mostly miss them, so every share is a floor. Several leading words sit mostly in one or two comment sections. A separate reviewer recomputed every figure and failed eight wording and ranking claims; all were fixed before publication.
The twenty-four videos whose comments we read
Every comment count on this page comes from the comment sections of these videos. The two launch-day sets of ten were each drawn at random, by a seed written down before the draw, from the 50 most-viewed "GTA 6" videos published on launch day. The four pre-release videos are the ones read for the vocabulary comparison, including the Call of Duty control. Scroll sideways to see all twenty-four; each opens on YouTube.
1 · first drawcounted in Launch day
2 · first drawcounted in Launch day
3 · first drawcounted in Launch day
4 · first drawcounted in Launch day
5 · first drawcounted in Launch day
6 · first drawcounted in Launch day
7 · first drawcounted in Launch day
8 · first drawcounted in Launch day
9 · first drawcounted in Launch day
10 · first drawcounted in Launch day
11 · second drawcounted in Fans in their own words
12 · second drawcounted in Fans in their own words
13 · second drawcounted in Fans in their own words
14 · second drawcounted in Fans in their own words
15 · second drawcounted in Fans in their own words
16 · second drawcounted in Fans in their own words
17 · second drawcounted in Fans in their own words
18 · second drawcounted in Fans in their own words
19 · second drawcounted in Fans in their own words
20 · second drawcounted in Fans in their own words
21 · pre-releasecounted in What people say
22 · pre-releasecounted in What people say
23 · pre-releasecounted in What people say
24 · pre-releasecounted in What people say
Grand Theft Auto VI was released on Thursday, November 19, 2026, on PlayStation 5 and Xbox Series X|S.
Rockstar Games confirmed the date on November 6, 2025, after two delays. Pre-orders opened June 25, 2026; digital pre-loads began November 12, 2026. As of the October 24 read, no PC release date had been announced. All six AI calls recorded below returned the November 19 date.
The registered question, written before any data was pulled
Given one identical query string, how much do the domains cited by an answer engine overlap with the domains Google ranks organically?
Four outcomes were registered with probabilities before retrieval: the layers are the same channel (15%), related but distinct (25%), separate distribution problems (45%), or too unstable to compare (15%). The falsifier was named in advance: the separate-problems reading dies if two or more of the nine reference domains rank organically for the same string.
One disclosure that weakens this registration. An uncontrolled version of the comparison had already been run hours earlier and returned zero overlap, so the prior placed highest had seen a number before it was written. That makes a confirming result worth less than a surprising one, and it is recorded here rather than hidden.
The result: zero domains in common
| Pair | Domains in common | Out of |
|---|---|---|
| ChatGPT and Perplexity | 0 | 17 distinct domains between them |
| ChatGPT and Google organic | 0 | 7 and 5 |
| Perplexity and Google organic | 0 | 10 and 5 |
| All three instruments | 0 | 22 distinct domains in total |
| Google with itself, "gta 6 release date" against "when is gta 6 out" | 1 — redbull.com | 5 and 6 |
The registered falsifier did not fire. Zero of the nine reference domains ranked organically, against a threshold of two.
The last row is the one worth sitting with. Google was less consistent with itself across two phrasings of one question than Perplexity was across three repeats of the same string. Rewording the question changed five of six results on the same engine, in the same city, in the same minute.
Every domain each machine cited
Published in full because a summary of a citation list is not a citation list. Sites flagged dedicated exist to answer questions about this one game and publish little else.
| Domain | Call 1 | Call 2 | Call 3 | Stable |
|---|---|---|---|---|
| gamesradar.com | yes | yes | — | 2 of 3 |
| gta6-wiki.org dedicated | — | yes | yes | 2 of 3 |
| gta6game.wiki dedicated | yes | yes | yes | 3 of 3 |
| gta6wiki.io dedicated | — | yes | yes | 2 of 3 |
| gtacorner.com dedicated | — | — | yes | 1 of 3 |
| pcgamer.com | yes | yes | — | 2 of 3 |
| windowscentral.com | yes | yes | — | 2 of 3 |
One domain of seven survived all three calls. On call three, all four citations were purpose-built GTA wiki sites: gta6-wiki.org, gta6wiki.io, gta6game.wiki and gtacorner.com. Mainstream games press appeared in two calls and vanished from the third. Rockstar's own domain was never cited at all.
| Domain | Call 1 | Call 2 | Call 3 | Stable |
|---|---|---|---|---|
| gtaboss.gg dedicated | yes | yes | yes | 3 of 3 |
| gtaboom.com dedicated | yes | yes | yes | 3 of 3 |
| bbc.com | yes | yes | yes | 3 of 3 |
| forbes.com.au | yes | yes | yes | 3 of 3 |
| ign.com | yes | yes | yes | 3 of 3 |
| digitaltrends.com | yes | yes | yes | 3 of 3 |
| cnet.com | yes | yes | yes | 3 of 3 |
| usatoday.com | yes | yes | yes | 3 of 3 |
| fox10tv.com | yes | yes | yes | 3 of 3 |
| collider.com | yes | yes | yes | 3 of 3 |
Ten of ten, in the same order, three times. This is the fourth separate measurement in which Perplexity has returned a fully stable citation set and ChatGPT has not. It is a measurement of these questions on these days, not a league table, and we would not publish it as one.
| Position | Domain | Note |
|---|---|---|
| 1 | reddit.com | |
| 2 | redbull.com | |
| 3 | gta.fandom.com dedicated | |
| 4 | facebook.com | |
| 5 | forbes.com | |
| 6 | redbull.com | same domain as position 2 |
| Position | Domain | Note |
|---|---|---|
| 1 | rockstargames.com | |
| 2 | vicountdown.com dedicated | |
| 3 | en.wikipedia.org | |
| 4 | xbox.com | |
| 5 | rockstargames.com | same domain as position 1 |
| 6 | youtube.com | |
| 7 | redbull.com |
Depth 10 was requested and six and seven organic results came back. The rest of each page is an AI overview, a video pack and paid units. A statement about organic overlap is therefore not a statement about what a person sees, and nothing on this page claims otherwise.
Every domain, every instrument, in one table
The twenty-seven distinct domains returned by four readings on one afternoon. Filter it. The control worth pressing is cited by more than one instrument, because of what it returns.
Showing all 27 domains.
| Domain | ChatGPT 1–3 | Perplexity 1–3 | Google: release date | Google: when is out | Instruments |
|---|---|---|---|---|---|
| bbc.com | ··· | ●●● | · | · | 1 |
| cnet.com | ··· | ●●● | · | · | 1 |
| collider.com | ··· | ●●● | · | · | 1 |
| digitaltrends.com | ··· | ●●● | · | · | 1 |
| en.wikipedia.org | ··· | ··· | · | ● | 1 |
| facebook.com | ··· | ··· | ● | · | 1 |
| forbes.com | ··· | ··· | ● | · | 1 |
| forbes.com.au | ··· | ●●● | · | · | 1 |
| fox10tv.com | ··· | ●●● | · | · | 1 |
| gamesradar.com | ●●· | ··· | · | · | 1 |
| gta.fandom.com dedicated | ··· | ··· | ● | · | 1 |
| gta6-wiki.org dedicated | ·●● | ··· | · | · | 1 |
| gta6game.wiki dedicated | ●●● | ··· | · | · | 1 |
| gta6wiki.io dedicated | ·●● | ··· | · | · | 1 |
| gtaboom.com dedicated | ··· | ●●● | · | · | 1 |
| gtaboss.gg dedicated | ··· | ●●● | · | · | 1 |
| gtacorner.com dedicated | ··● | ··· | · | · | 1 |
| ign.com | ··· | ●●● | · | · | 1 |
| pcgamer.com | ●●· | ··· | · | · | 1 |
| redbull.com | ··· | ··· | ● | ● | 1 |
| reddit.com | ··· | ··· | ● | · | 1 |
| rockstargames.com | ··· | ··· | · | ● | 1 |
| usatoday.com | ··· | ●●● | · | · | 1 |
| vicountdown.com dedicated | ··· | ··· | · | ● | 1 |
| windowscentral.com | ●●· | ··· | · | · | 1 |
| xbox.com | ··· | ··· | · | ● | 1 |
| youtube.com | ··· | ··· | · | ● | 1 |
No domain reaches two. Twenty-seven domains, four readings of one question inside half an hour, and the Instruments column never leaves 1. The only repeat anywhere is redbull.com, which Google returned for both of its own phrasings.
The filters hide rows that are already on the page. Nothing here is fetched or calculated in the browser, so this table is complete in the HTML with scripting switched off — which is also how a retriever reads it.
What the machines reach for instead: pages built for one question
| Domain | Cited by | What it is |
|---|---|---|
| gta6game.wiki | ChatGPT, all 3 calls | A wiki about one unreleased game |
| gta6-wiki.org | ChatGPT, calls 2 and 3 | A wiki about one unreleased game |
| gta6wiki.io | ChatGPT, calls 2 and 3 | A wiki about one unreleased game |
| gtacorner.com | ChatGPT, call 3 | A site about one franchise |
| gtaboss.gg | Perplexity, all 3 calls, position 1 | A site about one franchise |
| gtaboom.com | Perplexity, all 3 calls, position 2 | A site about one franchise |
Six of the 17 domains the two answer engines cited are sites that exist to answer questions about this one game. Both of Perplexity's top two positions are held by them. All four of ChatGPT's third-call citations are.
Of ChatGPT's seven domains, exactly one — pcgamer.com — is mainstream press. IGN, GameSpot, Eurogamer, Rockstar's own site and Take-Two's investor pages were registered as the expected match list before the pull. Rockstar was cited zero times out of three by ChatGPT on a question about Rockstar's own release date.
What this does not establish. These sites were cited; nothing here measures why, and a correlation between being single-subject and being cited is not a mechanism. One query, one day. The honest reading is that the answer layer is reachable by small dedicated pages in a way the search layer, on this query, was not — and that is a hypothesis this page will keep testing in public rather than a rule.
What people actually search for
| Keyword | Monthly volume | Shape |
|---|---|---|
| gta 6 | 1,500,000 | brand |
| grand theft auto gta 6 | 1,500,000 | brand |
| gta 6 release date | 1,000,000 | release date |
| when does gta 6 come out | 201,000 | release date |
| gta 6 pre order | 301,000 | pre-order |
| gta 6 leaked | 135,000 | leaks |
| gta 6 map | 110,000 | map |
| gta 6 price | 110,000 | price |
| gta 6 game price | 90,500 | price |
| gta 6 delay | 74,000 | delay |
Thirteen of the top thirteen keywords are release-date variants, each between 823,000 and 1,500,000 searches a month. Pre-order, leaks, map, price and delay follow. There is not one analysis-intent query in the top fifty.
Volumes are Google's advertising bands rather than counts, and keyword suggestion data is an advertising product that skews commercial. Read the ordering, not the magnitudes.
What the publisher spends
| Fiscal year | Period end | Advertising | Revenue | Intensity |
|---|---|---|---|---|
| FY2021 | 2021-03-31 | $241,100,000 | $3,372,800,000 | 7.15% |
| FY2022 | 2022-03-31 | $297,300,000 | $3,504,800,000 | 8.48% |
| FY2023 | 2023-03-31 | $1,212,500,000 | $5,349,900,000 | 22.66% |
| FY2024 | 2024-03-31 | $1,132,400,000 | $5,349,600,000 | 21.17% |
| FY2025 | 2025-03-31 | $1,253,900,000 | $5,633,600,000 | 22.26% |
| FY2026 | 2026-03-31 | $1,287,100,000 | $6,656,400,000 | 19.34% |
The step change is FY2023, a 4.08-times jump, and intensity has been flat to falling since. Take-Two closed the Zynga acquisition in May 2022, inside FY2023, and Zynga's model is paid user acquisition. The most recent year is lower than three years ago.
Take-Two’s 10-K confirms the Zynga explanation. The timing and the mechanism fit, and the filing now backs them: the 10-K for the year to March 2023 attributes the $1,076.2m rise in selling and marketing expense primarily to marketing for titles from the Zynga acquisition. It does not say how much of the advertising increase was Zynga’s.
For scale: the median advertising intensity across 137 prepackaged-software filers, 61 pharmaceutical filers and 40 business-services filers for CY2024 was 2.01%, 3.22% and 3.30%. Take-Two sits at roughly six to nine times a cross-industry typical value — but the spread inside every category measured was so wide that the category itself predicts almost nothing, which is a separate CULT+MATH study.
The same query, three times in one hour, three different answers
Before reading any comment section we had to pick one, and picking it turned out to be the finding. Six YouTube searches were run three times inside one hour and the top ten by comment count recorded each time.
| Corpus | Query | Run 1 ∩ 1b | Run 1b ∩ 1c | All three |
|---|---|---|---|---|
| GTA VI reaction | "GTA 6 trailer reaction" | 1 | 10 | 1 of 10 |
| GTA VI analysis | "GTA 6 analysis" | 9 | 10 | 9 of 10 |
| GTA VI news | "GTA 6 news" | 8 | 8 | 7 of 10 |
| GTA V roleplay | "GTA 5 roleplay" | 10 | 8 | 8 of 10 |
| Call of Duty MW4 | "Call of Duty Modern Warfare 4" | 10 | 10 | 10 of 10 |
| Zelda OoT remake | "Zelda Ocarina of Time remake" | 9 | 9 | 8 of 10 |
The instability arrives in discontinuities, not as steady churn. The reaction corpus replaced nine of ten members between two consecutive runs and then held perfectly. The roleplay corpus held perfectly and then lost two. Only Call of Duty survived all three intact.
A volatility rate computed from a single interval would have been wrong for both of the first two. So this table reports pairs rather than a score, and every corpus read on this page has its video IDs frozen in the specification rather than re-picked at run time.
A retraction, published rather than quietly removed. An earlier version of this research named a video — "The GTA 6 Situation" by Runtime Caffeine, 7,368 comments — as the largest comment corpus in the landscape. It appeared in the first run and in neither re-run. One observation had been reported as a property. The claim is withdrawn; it may well be the largest corpus, but this instrument did not establish it.
What people say, before a game ships and after
| Term | Share of 200 comments |
|---|---|
| gta | 10.0% |
| rdr2 | 8.0% |
| jason | 7.5% |
| love | 7.0% |
| lucia | 7.0% |
| rockstar | 5.0% |
| car | 4.0% |
| Term | Share of 200 comments |
|---|---|
| jake | 11.5% |
| tim curry | 11.5% |
| gta | 9.0% |
| horse | 9.0% |
| rip | 6.5% |
| 30fps | 5.5% |
| buy | 3.5% |
| Term | Share of 200 comments |
|---|---|
| jason | 9.5% |
| gta | 8.0% |
| lucia | 7.5% |
| yellow | 6.0% |
| red | 5.0% |
| rdr2 | 4.5% |
| trailer | 4.5% |
| Term | Share of 200 comments |
|---|---|
| cod | 22.0% |
| campaign | 20.0% |
| the campaign | 10.5% |
| jake | 7.0% |
| best | 6.0% |
| mission | 5.0% |
| review | 5.0% |
The control row is the finding. gameranx appears twice. On an unreleased game its comment section returned the host's name, a tribute to a late actor, and a horse. On a game that had shipped two days earlier, the same channel returned campaign in 20% of comments, review, mission, and unprompted price talk — one commenter volunteering what they would pay for the campaign sold separately.
Because the channel is held constant, the difference is not the channel. Pre-release comment sections carried no commercial vocabulary at all. For anyone hoping to read purchase intent out of pre-launch conversation, that is the answer, and it is a negative one.
Two things change between those rows, not one: a different title and a shipped-versus-unshipped game. The comparison can rule out a channel explanation. It cannot by itself prove the shipping explanation, and a pass that holds the title constant across a launch is the next measurement.
Comment text is not stored. Terms are counted once per comment, so a figure is how many people said it rather than how often it was typed. Comments are ordered by relevance, not time, so each reading skews to the days after publication.
The audience answered the economic claim with their own working week
A television station said the launch could drain a billion dollars of US productivity. We read the comment section to see whether an economic framing draws an economically-minded audience. It draws something better and stranger than that.
| Channel | Subject and framing | Group | Terms | Which |
|---|---|---|---|---|
| FOX 2 Detroit | GTA VI, economic framing | finance | 11 | day 25.0% · week 14.5% · work 12.5% · vacation 6.5% · my job 5.0% |
| Nino | GTA VI, economic framing | finance | 6 | day 17.5% · work 10.5% · days 5.5% · day off 3.5% |
| India Global Review | GTA VI, economic framing | finance | 2 | work 6.3% · day 5.1% |
| How Money Works | industry economics | finance | 0 | — |
| Moon Channel | industry economics | finance | 0 | — |
| TGG | GTA VI, gameplay analysis | analysis | 1 | take 3.5% |
| gameranx | GTA VI, gameplay analysis | analysis | 0 | — |
| DarkViperAU | GTA VI, gameplay analysis | analysis | 0 | — |
| gameranx | Call of Duty, already shipped | control | 0 | — |
Three of three corpora covering this game economically carry it. Zero of six everything else — including two corpora about industry economics, three gameplay analyses, and a control covering a game that had already shipped.
What people wrote, quoted at the 120-character cap the listening standard sets, with no author names:
- “I quit my job 2 years ago for this day”
- “A day off work. Rookie numbers. I took a week off.”
- “I’m 44 & I’ve already put in my vacation time over 3 months ago”
- “Put in to take the day off for this 3 months ago.”
- “I’ve already notified the boss I’m going to be sick on that day”
They did not debate the billion-dollar figure. They are the billion-dollar figure, reporting themselves, unprompted, with dates and notice periods attached.
What this is not. These are consumers planning time off, not buyers of research, and a large engaged comment section is not evidence of commercial demand. We registered a commercial-vocabulary test before this pull — price, buy, revenue, earnings and twenty-odd others — and it found almost nothing: a mean of 1.4 terms in the finance corpora against 0.33 in the analysis ones, which fell between our own thresholds and so returned no verdict at all. The table above was not what we were looking for, and it is reported as an unregistered observation rather than dressed up as the result we predicted.
The framing that is already working, for somebody else
| Channel | Title | Comments | Views | Published |
|---|---|---|---|---|
| How Money Works | WTF Is Happening To The Video Game Industry? | 4,167 | 1,238,088 | 2026-07-31 |
| Moon Channel | Okay, NOW the Video Game Industry is Collapsing | 2,104 | 344,154 | 2026-09-26 |
| Nino | The World Isn't Ready For GTA 6 | 2,000 | 1,802,957 | 2026-10-12 |
| FOX 2 Detroit | GTA 6 launch could drain $1 billion from US economy | 1,742 | 311,245 | 2026-10-22 |
| India Global Review | GTA 6: Billions In Sales, $1 Billion In Lost Productivity | 133 | 84,555 | 2026-10-23 |
A Detroit television affiliate framing a game launch as an economic event drew 1,742 comments. Three of the ten domains Perplexity cited about this game are financial press — and on the open-ended version of the question, CNBC, Yahoo Finance and Benzinga were among them.
The FOX 2 figure moved from 1,683 to 1,742 comments across four readings in one evening, which is why it is given with its reading dates rather than as a standing number.
What this study cannot tell you
- One city, one day, one query string. Houston, October 24, 2026. No claim about other markets and no trend claim.
- Organic results are a minority of the Google page. Depth 10 returned six and seven organic positions. The AI overview, the video pack and the paid units are not counted and are most of what a person sees.
- Platform-default budget-tier models. Both engines are read at the tier the API serves, which is not the model a consumer is served.
- Nothing here observes a purchase. Search volume is interest, comment share is attribution, model citation is belief. Not one row measures a sale, and advertising spend is an input rather than an outcome.
- The Zynga attribution is confirmed but not sized. The 10-K attributes the selling-and-marketing rise primarily to Zynga titles; it does not say what share of the advertising line that was.
- The frame comparison returned no registered verdict. Four candidate outcomes were registered before the pull and the result fell in a gap between two of them, so the test could not report what it found. The design fault is ours and the self-report table is labelled unregistered because of it.
- Three of the five economically-framed corpora concern this one game, 26 days from launch. A general audience with a dated event in front of them is an unusual condition and the self-report pattern may not survive outside it.
How this page changes
This is a standing study, not an article. The measurements above are re-run on a schedule and the page is corrected in place with the date of the correction shown. Corrections are published, never applied quietly, and the retraction in the corpus section is the first entry in that record.
| Measurement | Cadence | Also re-run when |
|---|---|---|
| Answer-engine citations | Weekly | Rockstar announces anything |
| Google organic positions | Weekly | — |
| Search demand | Monthly | — |
| Take-Two advertising | On each 10-Q and 10-K | — |
| Corpus membership | Weekly | — |
| Comment vocabulary | At launch (done Nov 20) and 30 days after | The frozen video IDs stay frozen |
| Date | What changed |
|---|---|
| 2026-10-24 | Published. Retracted the Runtime Caffeine corpus claim in the same pass that measured it, per the corpus section. |
| 2026-10-24 | Added the self-report section, nine corpora. The registered commercial-vocabulary test returned no verdict; the labour-vocabulary result is marked unregistered on the page. |
| 2026-11-20 | Added the launch-day section: ten videos drawn by seed, 1,162 comments. Both registered predictions missed: one did not happen, one cannot be scored. |
| 2026-11-20 | Added the fan section: themes named from the fans' own words, tested on ten new videos drawn by a second seed, 1,684 comments. |
Standing research. Figures are re-measured on the schedule stated on each page and the page is corrected in place, dated. Corrections are published, not applied quietly.
Cite as: CULT+MATH LLC, Ipalibo Da-Wariboko. Attribution to cultmath.com is requested.
Studies signed by a desk are made by the research desks and published by Ipalibo Da-Wariboko.
Keep your own calls on record
DEOS is the decision app built on this research. Talk through a plan or a decision, and it writes the record, checks it against your past calls and brings it back when it is time to review.
Start freeFree for 30 days, no card. Then $99 a month.
Or see the evidence for this category’s decisions in the Decision Brief →