Social Media
TikTok
Nothing was taken from you. Everything was added on top of you.
Every other social assessment in this index tells a story of repossession: a product that once did what the customer chose, and gradually stopped. TikTok does not fit that story, and pretending it does would be the easiest way to get this page wrong. TikTok launched in 2018 as a full-screen, autoplaying, infinitely scrolling, real-time personalised feed. There was no chronological era, no friend feed, no user-ordered timeline to repossess. The attention architecture is not what changed — it is what customers signed up for. Established What changed is what TikTok subsequently built on top of it: non-refusable behavioural ad targeting (2021), advertising that wears the creator's own face (2021), a virtual-currency economy whose conversion rate TikTok reserves to its "absolute and sole discretion" (2019–), a shopping mall inserted into ordinary discovery with no off switch (2023), search sold to a guaranteed top position (2026), and interruption itself packaged as an advertising product (2026). Seven years of monetisation loaded onto an engine that never needed to get more compelling — because it already was.
CFS = CVI − CHI (81 − 74 = +7), placing TikTok in the index's finely balanced band, alongside Instagram (+6) and Facebook (+13) — the only other assessments there. This is the highest CHI and the highest CVI in the Social Media category, and the combination is the point: TikTok extracts more than its peers and delivers more than its peers. Scores were calculated from the evidence under v2.0 and then sanity-checked against Facebook and Instagram, not copied from them. Overall CHI confidence: B+ — high on the mechanisms, deliberately lower on the quantities, because several of TikTok's most-quoted figures are single-sourced and are treated that way here. Overall CVI confidence: A−. Methodology →
The defining finding
TikTok built the screen-time tool the public asked for — then measured its success by the press coverage.
Internal documents unsealed in Kentucky's litigation record the 60-minute teen limit reducing daily use from about 108.5 minutes to roughly 107, and define success as "improving public trust in the TikTok platform via media coverage." A project manager wrote: "Our goal is not to reduce the time spent." Unsealed filing · disputed
This is the difference between "TikTok is engaging" and a CHI finding. Engagement is not hostility. A safeguard that the company itself scored on media coverage is.
The central distinction · Why TikTok is not Facebook or Instagram
There was no "before." That is the whole analysis.
CHI's core principle is that direction matters more than any single decision. That principle cuts both ways. Facebook and Instagram are scored against a documented earlier bargain that the company itself later withdrew. TikTok has no such baseline — and a page that manufactured one would be scoring a fiction. The correct question is not "what did TikTok take away?" but "what did TikTok add, to whose benefit, and what did the customer get in exchange?"
- The algorithm came firstAugust 2018: the Musical.ly merger produces one app whose home is a full-screen, autoplaying, algorithmically selected For You feed. Following is secondary from day one.
- Nothing chronological was ever takenThere is no year in which TikTok's landing feed was ordered by the customer and then wasn't. Claiming otherwise is the most common error made about this company.
- The engine never had to changeIt was already optimised for retention and time spent — TikTok's own internal "Algo 101" says the "ultimate goal is to add daily active users."
- What changed is what got attached to itAds (2019) → non-refusable targeting (2021) → creator-native ads (2021) → LIVE virtual currency at scale (2019–) → Shop in the feed (2023) → search sold to the top (2023–26) → interruption as a product (2026).
- CHI reads this as progressive commercial densificationNot a thing taken away — a series of things added, each one monetisable, none of them optional.
Pattern one · Infinite Engagement — supported, type specimen
The stopping cue was never removed. It was never built.
Infinite Engagement is defined in the Lexicon as the removal of natural stopping points through endless feeds, autoplay and continuous recommendation. TikTok is the index's type specimen — not because it removed the most, but because it is the purest expression of the design: a product in which the absence of an ending is the founding feature rather than a later optimisation.
How this is scored. The architecture is counted once, inside Behavioral Manipulation, and discounted for being original rather than repossessed — which is why Behavioral Manipulation is 20/25 rather than the ceiling. What is not discounted is the second half of the finding: the tools TikTok built to counteract its own engine.
The sharpest exhibit · Stated intent versus measured intent
"Our goal is not to reduce the time spent."
This is the strongest documentary evidence in the entire TikTok record, and it is worth being precise about why. It is not evidence that TikTok is addictive — that is a contested scientific question this page does not pretend to settle. It is evidence about the relationship between a publicly presented safeguard and the company's internal priorities, which is exactly what CHI's Trust & Transparency dimension exists to measure.
The spine of the assessment
One engine. Seven years of freight.
Read downward. The base layer is what customers accepted in 2018. Everything above it is what TikTok attached to that acceptance — each layer monetisable, each layer added without a corresponding customer control.
Seven of these nine layers are advertising, commerce or currency. None of them arrived with a control that lets a customer decline that specific layer — the one partial exception, the EU's non-personalised feed toggle, came from the Digital Services Act rather than from product judgement, and reaches only the EU and Kazakhstan.
Pattern two · Algorithmic Replacement — supported, in TikTok's specific form
The shopping mall does not stop at the feed you didn't choose.
Algorithmic Replacement usually means recommendation displacing content the customer intentionally chose. TikTok cannot be scored the way Instagram is, because For You was never the customer's choice to lose. The TikTok-specific form of the pattern is narrower and better evidenced: commercial content that the customer did not choose now reaches the two feeds the customer builds deliberately — and TikTok's own documentation says so.
The finding, stated exactly. TikTok inserted a transaction layer into ordinary discovery, tuned its density centrally, steered creator incentives toward selling, declared it would not dial back when customers complained, and — three years later — still ships no control that reduces it. That the commerce is increasingly closing in its own tab is real mitigation and is credited. That it reaches Following and Friends is not mitigated by anything.
Pattern three · Reference-Price Erasure & Loyalty Financialization — supported
Cash goes in as money. It comes out as "popularity."
TikTok LIVE is the part of this product least like Instagram and least like Facebook, and it is where the strongest revenue-extraction findings sit. A viewer buys Coins with real money, spends Coins on Gifts, and the creator receives Diamonds. Every step of that chain is priced except the one that determines what the creator actually gets.
The clause that makes it a CHI finding. TikTok's Virtual Items Policy, reproduced verbatim in Utah's June 2024 complaint, states that Diamonds accrue "at a rate of conversion to be determined by us from time to time in its absolute and sole discretion." The 2021 version quoted by BEUC goes further: "we have the absolute right to manage, regulate, control, modify and/or eliminate such exchange rate as we see fit in our sole discretion." Policy text, reproduced in litigation
The claim this page does not make. TikTok does not take 70% of gifts. That figure comes from a 2022 BBC investigation into Syrian refugee streams, where the 70% is the loss across the whole chain including app-store commission; TikTok told the BBC its own cut was "significantly less than 70%" but declined to name it. Nor is it true that TikTok has never published its share: Utah's complaint records a single short article from December 2022 stating 50%. The accurate statement is that TikTok does not disclose the economics anywhere a user or creator would ordinarily look — not in the Virtual Items Policy, not in the Gifts help article, not in the Diamonds help article. Closure-verified formulation
The arithmetic, with its uncertainty attached. A $10 in-app purchase buys roughly 660 coins after the store's cut. At roughly 50% of face value, about $3.30 of that $10 reaches the creator — around a third of what the viewer spent. Buy the same $10 of coins on the web and roughly $4.85 reaches them. The ~50%-of-face-value figure is well corroborated; the share-of-consumer-spend figure is arithmetic built on it, and is presented as such.
Pattern four · Intellectual Disrespect — supported, primary
The advertising is built to look exactly like the thing you came for.
TikTok says all paid placements carry a "Sponsored" or "Ad" marker, and this page takes that at face value. The finding is not that TikTok hides its ads. It is that TikTok's flagship ad format is architecturally designed to be indistinguishable from organic content in every respect except the label — and that TikTok waited four years to enforce the disclosure rules covering the rest.
So: the creator's handle, the creator's face, the creator's caption, the creator's profile and the creator's accumulated likes and comments all travel with the ad, and the social proof the ad generates flows back into the organic post. The paid content and the organic content are the same object. TikTok Business Help
Pattern five · Advertising Creep — supported as to mechanism, not as to density
Search stopped being a lookup and started being inventory.
This page will not claim that TikTok "dramatically increased the number of ads in the For You feed," because no independent 2020–2026 ad-frequency series exists and inventing the impression of one would be exactly the kind of claim CHI is supposed to refuse. The defensible and more interesting argument is about surfaces, formats and guarantees — and search is the clearest case.
The rest of the Advertising Creep case, stated at its true strength: formats proliferated continuously from a single skippable pre-roll in January 2019 to Logo Takeover, Prime Time and TopReach in 2026; targeting became non-refusable outside the EU in April 2021; commercial placement expanded into search and into creators' own posts. Commercial surfaces, formats, targeting and interruption mechanisms expanded progressively — even though a clean longitudinal ad-frequency series is unavailable, and none is asserted here.
The control surface · What actually works
TikTok gives you more controls than Instagram does. They are built to reweight, not to exclude.
This is where a lazy version of this page would say "TikTok users cannot control their feed," which is false. TikTok's control surface is broad, has expanded materially, and in the EU is the best in this index. The finding is about the ceiling of those controls, not their absence.
The pattern across the table is consistent and worth naming. Where a control governs content, TikTok tends to build it well and, in Europe, better than anyone. Where a control would govern the commercial layer — advertising frequency, shopping, gifting, the landing feed — it either does not exist or does not bind. That is not an accident of engineering effort. It is the shape of the business model rendered in settings.
TikTok's case for TikTok
Why CVI is 81.
This is the highest CVI in the Social Media category and the second-highest in the index. If this page leaves you thinking TikTok is simply an extraction machine, it has failed — because the measured customer verdict points the other way, and CHI is required to say so.
Measured satisfaction is not merely good — in 2024 TikTok became the first-ever category leader in the American Customer Satisfaction Index's social media segment at 78, and it sat at 76 in 2026, above the industry average every year since 2023. Crucially, that rise happened during the Shop, LIVE and search build-out this page documents. Independent survey index
Zero mandatory price, at global scale, for a product with genuine distributional novelty: the interest graph lets a creator with no followers reach an audience — something no follower-graph platform has ever offered. Around seven in ten teen users report mostly positive experiences. Fact
A complete free creation stack: editor, effects, licensed sounds, Duet and Stitch, auto-captions (2021), photosensitivity warnings (2020), ALT text (2025), Add to Music App. For an enormous number of people this is the only video production tool they will ever need, and it costs nothing.
Creator economics have moved consistently in creators' favour: the Creator Fund's ~$0.02–0.04 RPM was replaced by Creator Rewards at roughly $0.40–1.00, Tips pass through at 0%, Pulse splits 50/50, and the 2025 US/Canada subscription share was raised to a 70% base rising to 90% — above YouTube's. Fact
Teen protections that arrived one to three years before Meta's equivalents: private-by-default for 13–15s and DMs off under 16 (2021), notification curfews (2021), LIVE and gifting restricted to 18+, a 60-minute default (2023), wind-down and Time Away (2025) — plus zero personalised advertising to EEA teens since 2023 and broad-only targeting for US teens since 2024. The effectiveness critique is documented above; the delivery is credited here. Fact
The best feed opt-out in this index, and a notable refusal: TikTok's EU non-personalised toggle was rated above every peer by Sitra, and TikTok has not deployed consent-or-pay in the EU — the mechanism that cost Meta a €200M DMA fine. Its keyword filters apply to Following, which Meta's do not.
A documented record of retreating under pressure, which CHI credits under Fixed-Under-Pressure: TikTok Lite Rewards permanently withdrawn from the EU within weeks (2024), the EU legitimate-interest ad switch paused one day before launch (2022), "Refuse all" cookies added before the CNIL fine, a mandatory Shop logistics rule paused after seller backlash (2026), and the reported reduction of feed commerce density after user complaints. Fact
Real consumer subsidy during the commerce build: TikTok absorbed a reported $500M+ US loss in 2023 funding 50% coupons and free shipping, and its 6% US referral fee is well under half Amazon's ~15%. Buyers got cheap goods; the cost shift onto sellers came later, and is scored above rather than here.
Category-defining innovation, and TikTok Pro as proof of concept. TikTok invented the modern short-video interest feed; every major competitor copied it. And in Germany, Portugal and Spain, TikTok already ships a free variant with no ads, no LIVE and no Shop — which demonstrates that a de-densified TikTok is technically feasible, and confined to three markets.
Deductions applied, and why CVI is not higher: Trust, Safety & Reliability sits at 8/15 because there is no documented direct phone or email support channel and recourse runs through a single in-app funnel, remedy is routed to app stores and sellers, TikTok Shop has a documented fraud problem its own executive called "organised crime," and the regulatory record on minors — a €345M Irish DPC fine, two preliminary European Commission findings — is the worst in this category.
CVI 81/100 against CHI 74/100 produces CFS +7 — the finely balanced band. Overall CVI confidence: A−.
Score breakdown
Why CHI is 74 — and why it is not higher.
74 is the highest CHI in the Social Media category, and it needs a reason that is not "TikTok feels more intense." The reason is that TikTok operates one extraction channel its peers do not have at comparable scale — a real-money virtual currency with an undisclosed conversion — and one commercial layer its peers retreated from — in-feed commerce with no user control. Against that, the design-from-origin discount is applied deliberately and visibly, and it is the single largest reason this number is not in the eighties.
CFS = 81 − 74 = +7, inside the −15 to +15 band the methodology calls finely balanced. TikTok therefore still returns more measured value than it extracts — but by less than a rounding error's worth of comfort, and by a narrower margin than Facebook (+13). The interpretive point is that both halves are large. A reader who saw only the CHI would think TikTok is among the most hostile products in this index; a reader who saw only the CVI would think it is among the best. Both readers would be right, which is the entire reason CHI reports two numbers.
Anti-double-counting, stated plainly.
The Coins economy touches three dimensions and is deliberately split rather than repeated: the money TikTok takes is counted once in Revenue Extraction, the mechanics that escalate spending once in Behavioral Manipulation, and the non-disclosure of the conversion once in Trust & Transparency. These are genuinely distinct harms — the amount taken, the pressure to spend, and the inability to know the terms — not one harm counted three times. Likewise, the wellbeing-tool arc is scored once under Behavioral Manipulation and not re-counted under Trust & Transparency despite qualifying there; the absence of a Shop control is scored once under Customer Restriction; and TikTok Shop's seller cost-shifting is scored only to the extent it reaches consumers, not as a separate consumer harm.
Associated patterns · Final state
The patterns materially relevant to this verdict.
Primary
Supported
Qualified and partial
Falsification · What the evidence does NOT support
TikTok is the most folklore-encrusted product in this index.
More popular claims about TikTok failed this investigation than survived it. Publishing the failures is the price of being believed about the findings above — and several of the loudest allegations are not merely unproven but demonstrably a distortion of something true and more interesting.
No capability has been demonstrated and no evidence supports it. The accurate kernel is duller and worse: TikTok's non-audio behavioural inference is good enough to produce the eavesdropping feeling on its own, and outside the EU you cannot refuse it. That is an argument about the data perimeter, not about microphones.
No credible measurement supports it. The one reported allocation figure — a doubling from 5% to 10% in late 2023, cut back to around 2% after complaints — is single-sourced and points the other way. The real finding is not the volume of shopping content but the absence of any control over it, and the fact that it reaches the feeds you curate yourself.
The 70% figure comes from a 2022 BBC investigation and describes losses across the entire chain including app-store commission. TikTok's own share is best reconstructed at around 50% of a gift's coin face value; TikTok told the BBC its cut was "significantly less than 70%" and declined to give a figure. The finding is the non-disclosure and the reserved discretion, not an inflated percentage.
Specific, documented mechanisms exist — For You feed ineligibility for undisclosed commercial content is now a published sanction — but no evidence supports blanket keyword-level suppression. Publishing the specific mechanism is more useful than repeating the folk theory, because the published one is real and the folk one obscures it.
The core product is entirely free and always has been. The exception is precise and confined to one market: from 11 May 2026 TikTok offers UK adults an Ad-Free tier at £3.99/month, described by three independent outlets as a consent-or-pay arrangement in which free use is conditioned on consenting to personalised advertising, with adults actively prompted to choose. That converts a previously free privacy control into a paid option — the same mechanism CHI documented at Meta, which drew a €200M DMA fine — but it does not paywall any core functionality, and it rests on corroborated reporting rather than a TikTok primary document. A reported 11 November 2026 decision deadline remains single-sourced and is not treated as fact here.
CHI does not score a company for building a good ranking system, and this page does not. TikTok's feed is, by measured satisfaction, the most successful consumer product in its category. What is scored is what was attached to that feed and what the customer was never given the ability to decline — a distinction the loudest criticism of TikTok routinely collapses, and which is the reason this assessment lands at 74 rather than in the nineties.
Evidence & methodology
- Company documentationHow TikTok recommends content — TikTok Help CenterTikTok's canonical list of nine recommendation controls, containing no shopping or commerce control of any kind; and the statement that the non-personalised feed is "only available in the EU and Kazahkstan" [sic]. Retrieved 18 August 2026.
- Company documentationFor You — TikTok Help CenterPrimary source for "The For You feed is the first feed you'll see when you open TikTok." Confirms the landing feed directly rather than by inference.
- Company documentationSpark Ads — TikTok Business Help CenterSource of "All of the Spark Ads post engagement will be attributed to the original organic post", the private-to-public conversion, the caption lock and the creator-profile link carried into the ad.
- Company documentationAbout the Commercial Content Disclosure setting — TikTok Business Help (updated April 2026)Documents automatic flagging "when it shows clear marketing intent", the 24-hour response window, and For You feed ineligibility as the sanction. Rollout completeness remains secondary-sourced.
- Company documentationTargeted ads — TikTok Help CenterSource of "You'll always see ads on TikTok based on the content you interact with or other data described in our Privacy Policy." As of this build TikTok's readable Help Center documents no ad-free product.
- Company announcementIntroducing TikTok Shop — TikTok Newsroom, 12 September 2023The US launch: Shop tab in main navigation, shoppable video, LIVE shopping, affiliate programme, in-app checkout. Outbound storefront links were retired the same day.
- Company announcementMore ways to discover new content and creators you love — TikTok Newsroom, 3 June 2025Source of "These settings won't eliminate any topics entirely, but can influence how often they're recommended" — TikTok's own description of the reweight-not-exclude design, and of the 200-keyword filter cap.
- Court recordUtah Attorney General v. TikTok — public complaint, 3 June 2024Reproduces the Virtual Items Policy verbatim ("a rate of conversion to be determined by us from time to time in its absolute and sole discretion"), TikTok's coin pricing logic and 30% mobile processing fee, the Gifter Level exhibits, and the record that the 50% commission appeared only in a single December 2022 article. Allegations; unadjudicated.
- Court recordNew York Attorney General v. TikTok — complaint, 8 October 2024Source of "TikTok does not allow users to disable this feature" (autoplay, ¶70), echoed in Pennsylvania's August 2026 complaint. Filed litigation, not a finding.
- Unsealed internal materialTikTok's own documents, unredacted in Kentucky's lawsuit — NPR, 11 October 2024The 108.5 → 107 minute measurement, "improving public trust in the TikTok platform via media coverage", "Our goal is not to reduce the time spent", and the LIVE observation that "the content that gets the highest engagement may not be the content we want on our platform." TikTok says the quotes are cherry-picked and out of context; that response is carried at every mention.
- RegulatoryEuropean Commission — preliminary finding on TikTok's addictive design, 6 February 2026Names infinite scroll, autoplay, notifications and the recommender; finds wellbeing tools "easy to dismiss". Preliminary and contested — TikTok calls it "categorically false and entirely meritless".
- RegulatoryEuropean Commission — TikTok commits to permanently withdraw TikTok Lite Rewards in the EU, 5 August 2024Carried as counterevidence: a programme that paid users for watch time, withdrawn permanently within weeks. Credited under CHI's Fixed-Under-Pressure principle.
- RegulatoryEU/CPC influencer sweep — 15 February 202497% of checked influencers posted commercial content; 20% disclosed systematically; 334 of 576 accounts were on TikTok. Establishes the scale of the disclosure gap during the pre-enforcement years.
- Peer-reviewedKaplan et al., "When 'For You' Isn't For You" — arXiv, May 2026The independent audit of "Not interested": an 84% reduction while the signal is applied, against 47.5% for skipping — but feeds "dominated by such content again" once accounts stop indicating disinterest. Used in both directions: it establishes both that the control works and that its effect is not durable.
- Independent researchSitra — How to turn off recommendation algorithms on the main social media platforms, 24 April 2024Rated TikTok's non-personalised feed toggle 4/5 — above Instagram (3/5), Facebook (3/5), YouTube (2/5) and Snapchat (1/5). Carried at full weight as counterevidence.
- Independent researchCommon Sense Media — Constant Companion, September 2023The measured notification volumes — a median 2.9/day for TikTok against Snapchat's 19.6 and Instagram's 8.9 — that defeat the Notification Saturation finding outright.
- Satisfaction indexAmerican Customer Satisfaction Index — social mediaTikTok at 78 in 2024 (first-ever category leader) and 76 in 2026, above the industry average every year since 2023. Load-bearing for the CVI, and for the refusal to characterise commercial densification as a measurable experience collapse.
- Original reportingScreenshots suggest TikTok is circumventing App Store commissions — TechCrunch, 30 April 2024Source of TikTok's own in-app prompt — "Try recharging on tiktok.com to avoid in-app service fees" — and the in-app versus web coin price divergence. The ratio is the load-bearing figure; the specific 2026 price ladders in circulation are third-party observations of unstated methodology and are not quoted here as documented prices.
- Original reportingInside TikTok LIVE battles — Rest of World, 22 August 2024The five-minute battle format, single-match spends, and the reconstruction of ~50% of net gift revenue reaching creators. Reporting, not company disclosure.
- Original reportingTikTok won't dial back on TikTok Shop content — eMarketer, April 2024Source of "We're not considering dialing back on Shop content", and of the survey finding that of the 62.5% who noticed more shopping content, only 8.3% used TikTok less. Both halves are carried.
This page is a synthesis of a frozen evidence package: a deep research dossier covering sixteen parallel investigation streams, followed by a dedicated evidence-closure pass that re-tested the ten most load-bearing unresolved items against primary sources. Six were confirmed, two were modified, none were contradicted, and two remain unresolved and are stated as such.
Where the closure pass modified the research, the closure pass controls. Two corrections are load-bearing and are reflected throughout. First, the non-personalised feed toggle is available in the EU and Kazakhstan, not the EEA — Norway, Iceland and Liechtenstein are outside TikTok's stated scope — while the separate ad-interest tool is scoped to the EEA, the UK and Switzerland. Second, the UK Ad-Free tier moved from "disputed exception" to corroborated market-specific consent-or-pay, without disturbing the finding that no core functionality has been paywalled.
Evidentiary status is preserved and never upgraded for readability. State attorney-general complaints are labelled as allegations at every appearance and TikTok's disputes are quoted alongside them. European Commission findings are labelled preliminary every time. The Diamonds economics are stated as not disclosed in consumer-facing policy or help documentation rather than "never published", because litigation records one December 2022 article. The reported 5% → 10% → ~2% feed-allocation figures are single-sourced and are presented as indicative rather than measured. The reported 11 November 2026 UK deadline is single-sourced and is not asserted.
Omitted rather than softened: any claim that feed ad frequency has risen by a measurable amount; any specific 2026 Coin price ladder; TikTok's adult notification defaults, which TikTok does not publish and this research could not observe; the on-screen rendering of Spark Ad and Search Hub labels, both of which require live interface observation that was not available; and the terms of the Super Fan subscription, for which TikTok publishes no help documentation at all. CHI 74/100, CVI 81/100 and CFS +7 are settled values under Methodology v2.0. They are not a range.
Final verdict
CONCERNING — THE ENGINE NEVER CHANGED. THE FREIGHT DID.
TikTok is the one company in this category that cannot be accused of repossession, because it never handed anything over in the first place. It arrived in 2018 as an attention-optimised algorithmic product — full-screen, autoplaying, endless, personalised in real time — and that is what it still is. Customers accepted that bargain in enormous numbers, and the independent satisfaction measures in the record show no corresponding decline: TikTok became the first company ever to lead its ACSI category, and it did so during the years documented on this page. Anyone who tells you TikTok's story is a decline from a better product is describing Instagram, not TikTok.
What TikTok did instead was spend seven years attaching things to an engine that never had to improve. It withdrew the right to refuse behavioural ad targeting everywhere the law allowed. It built an advertising format whose entire purpose is to be indistinguishable from the creator content around it, and left the disclosure rules covering everything else unenforced for four years. It ran a cash economy through a two-step currency whose conversion rate it reserves to its own absolute discretion, wrapped in status ladders, public rankings and countdown battles, and describes the result as "popularity" redeemable for "rewards." It inserted a shopping mall into ordinary discovery — including, by its own documentation, into the feeds customers build by hand — declared it was "not considering dialing back," and has still not shipped a single setting that reduces it. It sold the top of its own search results as a guarantee. And when it built the safeguard the public was asking for, it measured that safeguard's success in media coverage and recorded the result as a minute and a half a day.
None of which makes the counter-case disappear, and CHI is not in the business of pretending it does. TikTok gives away the best short-video creation stack on earth for nothing, has raised creator pay repeatedly and now beats YouTube's subscription split, ships teen protections one to three years ahead of Meta's, has not deployed consent-or-pay in the EU, unlike Meta, whose version of it drew a €200M DMA fine, withdrew a pay-to-watch programme within weeks of being challenged, sends a fraction of the notifications its rivals do, has paywalled no core functionality, and already runs a free ad-free, Shop-free version of itself — in three European markets. CVI 81 is not a courtesy. It is the highest value score in this category, and the reason CFS lands at +7 rather than deep in the negative.
So the finding is narrow and it is specific. Not that TikTok is addictive; that argument is contested and this page does not settle it. Not that algorithms are hostile; CHI does not score ranking. The finding is that a company which never had to take anything away has instead spent seven years adding — advertising, commerce, currency, incentives, guaranteed placement, purchasable interruption — and that at nearly every commercial layer, the corresponding user control was either absent, partial, region-gated, or arrived only under regulatory pressure. The customer's side of the bargain has been frozen since 2018. TikTok's side has been under continuous construction.
Nobody ever changed your feed.
So why is there nothing in it you asked for?