E45: Theranos & VC fraud risks, China bans video games, Texas SB8, Apple app store, CA fires, Rogan

Chamath’s prediction was that MrBeast (Jimmy Donaldson) would become a “multi multi‑billionaire”, i.e., have a net worth in the multiple billions of USD (well above $2B).

What we can see today:

• Multiple outlets now report MrBeast as a billionaire, with net‑worth estimates around $1 billion, based largely on Celebrity Net Worth’s figure.(celebritynetworth.com)
• MrBeast himself has publicly confirmed that he is a billionaire “on paper” in a 2025 appearance on Steven Bartlett’s Diary of a CEO podcast.(hypebeast.com)
• Separate reporting on his business, Beast Industries, indicates that recent funding rounds value the company at about $5 billion, and in a 2024 deposition he stated he owns “a little over half” of it, implying an equity stake above $2.5B on paper.(businessinsider.com)
• Some analyses therefore frame his wealth as roughly $1B under conservative assumptions, but $2.5–3B+ if you fully mark his private equity stake to the latest $5B valuation.(thetrademedia.com)

Because:

1. There is no consensus: widely cited “net worth” sites and mainstream coverage still describe him as roughly a $1B billionaire, while business/valuation–based analyses plausibly put his stake above $2.5B on paper.
2. His company is private, so we cannot independently verify whether those $5B valuations (and the implied $2.5B+ stake) should be fully counted as personal net worth today.
3. The original prediction had no time horizon, so we are not yet in a position to say he failed to become multi‑billionaire either.

Given these conflicting estimates and definitional issues around how to compute his “net worth” (cash vs. illiquid private equity vs. conservative vs. aggressive valuation), it cannot be determined with confidence whether he is already a multi‑billionaire, even though it is clearly plausible on paper. Hence the outcome of the prediction is ambiguous, not cleanly right or wrong at this time.

Episode listings show that E45 ('Theranos & VC fraud risks, China bans video games, Texas SB8, Apple app store, CA fires, Rogan') was released on 4 Sep 2021 and the next episode, E46 ('False Ivermectin narratives, regulatory grift, wartime mentality in solving issues & more | LIVE from TPB Symposium!'), was released on 14 Sep 2021, explicitly marked as recorded live from The Production Board (TPB) Symposium. (pods.ee) Transcripts and YouTube summaries for E46 open with Friedberg welcoming the audience to the Production Board Symposium 2021 and explaining that they are hosting the All-In pod that night for the symposium attendees, confirming it was an in-person taping with a live audience at that event in mid-September 2021. (podscripts.co) The podcast’s history notes that All-In launched during the COVID‑19 pandemic and that early episodes were recorded remotely; episode catalogs up through E45 do not describe any prior episodes as being recorded live from a venue, making E46 the first clearly documented in‑person/live episode with the hosts together. (en.wikipedia.org) Although external sources do not detail the internal schedule of TPB’s poker night versus the main symposium sessions, the verified facts match the substance of Friedberg’s prediction: the besties’ first in‑person All‑In episode was indeed recorded at his TPB symposium in mid‑September 2021, shortly after E45. Therefore the prediction is judged accurate in practice, with only minor unverified scheduling details about the exact 'poker night' timing.

Jason predicted that the All-In hosts would hold an “All-In Summit” with roughly 100–200 attendees in Q1 or Q2 of 2022. In reality, the first All-In Summit did occur in Q2 2022—running May 15–17, 2022 in Miami Beach, Florida—but it was limited to about 700 attendees, not 100–200. (en.wikipedia.org) Because the prediction explicitly included a much smaller attendee count as part of its content, and there is no evidence of a separate 100–200-person All-In Summit in Q1 or Q2 2022, this forecast is considered wrong overall. (en.wikipedia.org)

Evidence shows Miami stayed open for large in‑person events throughout 2022, with no COVID-related shutdowns that would have prevented a conference.

1. Florida policy context. On May 3, 2021, Governor Ron DeSantis signed Executive Orders 21‑101 and 21‑102, suspending and invalidating all remaining local COVID‑19 emergency orders and restrictions, explicitly ending local mandates and lockdown-type measures statewide.(cbsnews.com) Subsequent summaries of his governorship note that from 2021 onward Florida restricted local governments’ ability to impose COVID‑19 mitigations, voided future local restrictions, and positioned the state as staying open (no new lockdowns, no vaccine‑passport requirements, etc.).(en.wikipedia.org) State-level reopening guidance for Florida places all counties in Phase 3, allowing businesses and entertainment venues to operate without capacity limits.(radio.wcmu.org) There is no record of Florida or Miami re‑imposing shutdowns on large events in 2022.

2. Large events actually held in Miami in 2022. The All‑In Summit itself was held in Miami Beach at the New World Symphony from May 15–17, 2022, with roughly 700 attendees, confirming that the hosts were able to schedule and execute their conference in person.(en.wikipedia.org) Other major events also proceeded: Ultra Music Festival returned to Miami’s Bayfront Park on March 25–27, 2022, drawing about 165,000 attendees after being canceled in 2020 and 2021 due to COVID.(en.wikipedia.org) NBA games in Miami during the 2021–22 season were held with fans in attendance; at most, masks were required until late February 2022 but games were not shut down.(en.wikipedia.org) These examples indicate Miami was hosting large in‑person gatherings without government closures.

3. Match to the normalized prediction. The normalized claim is that Miami would remain open for large in‑person events in 2022, without COVID‑related shutdowns that would prevent a conference. In reality, Florida policy removed local authority to re‑impose such shutdowns, and Miami in 2022 hosted multiple large events—including the All‑In Summit itself—without being blocked by COVID restrictions. While some venues or events may have had voluntary mask rules or other minor protocols, these did not constitute shutdowns preventing conferences. Therefore, the prediction is best scored as right.

Chamath’s claim was that, from the 2021 baseline, incidents of startup founders stretching the truth or committing fraud to secure venture funding would get “meaningfully worse” as capital inflows rose and diligence weakened.

Since 2021 there has been a conspicuous wave of high‑profile, venture‑backed founder fraud cases centered on inflated metrics and fabricated customers. A 2024 Association of Certified Fraud Examiners piece explicitly frames a pattern of “fraudulent founders,” grouping together Theranos, Frank (Charlie Javice), HeadSpin, FTX, Celsius, Terraform Labs and others as recent examples where founders misled investors about technology, revenues, or user numbers. (acfe.com) TechCrunch’s 2024 coverage of the IRL social‑app case notes that, by mid‑2024, the SEC had charged a venture‑backed founder with fraud “for the second time this week — and at least the fourth time in the past several months,” underscoring an unusual density of such cases. (techcrunch.com) Additional post‑2021 examples include Joonko’s founder allegedly fabricating customers and bank statements to raise ~$27 million, IRL’s founder allegedly faking user growth to raise ~$170 million, CaaStle’s founder allegedly using falsified financials to obtain over $300 million, and Frank’s founder, Charlie Javice, convicted and sentenced in 2025 for inventing millions of fake student accounts to induce JPMorgan’s $175 million acquisition. (insurancejournal.com) Crypto startups like FTX and Terraform Labs, heavily financed by top VCs, were also found liable for multi‑billion‑dollar frauds in this period. (techcrunch.com)

System‑level enforcement data show overall misconduct cases rising from the 2021 baseline. SEC enforcement actions increased from 697 total in FY 2021 to 760 in FY 2022 and 784 in FY 2023, with stand‑alone actions rising from 434 (2021) to 462 (2022) to 501 (2023), and monetary penalties hitting record or near‑record levels. (sec.gov) SEC Chair Gary Gensler reported more than 35,000 tips and complaints in FY 2022—over double 2016’s level—and tied much of the new misconduct to emerging, tech‑heavy areas such as crypto. (sec.gov) A 2023 Wall Street Journal analysis explicitly described an SEC crackdown on startup “fake it till you make it” culture, highlighting startup cases (e.g., Skael, Zymergen, Medly) and linking them to headline scandals like Theranos and FTX. (wsj.com) These sources treat founder‑led securities fraud in the startup ecosystem as a growing, systemic problem rather than isolated one‑offs.

The causal mechanism Chamath cited—cheap money and weak diligence—also appears in post‑2021 reporting. FTX’s huge 2021 raise, with competing blue‑chip VCs and a record $900 million round, is widely viewed as a product of the ZIRP‑era capital glut; later commentary questioned how much substantive diligence investors actually did, despite Sequoia’s claims. (coindesk.com) In the Frank case, court filings and post‑mortems describe JPMorgan as being in an aggressive 2020–2021 fintech acquisition push and criticize its rushed, shallow diligence compared with peers like Capital One; at sentencing, the judge faulted both Javice’s fraud and the bank’s lack of scrutiny. (fortune.com) These narratives align closely with Chamath’s warning that abundant capital and relaxed vetting would encourage founders to exaggerate or cross into outright fraud.

We do not have a clean time‑series of all “startup founder fraud incidents,” so the exact magnitude of the increase can’t be quantified. But given (a) the clear post‑2021 wave of venture‑backed founder frauds, repeatedly treated by regulators and fraud experts as a broader phenomenon, and (b) rising enforcement and tip volumes from the 2021 baseline, the available evidence supports the judgment that the problem did, in practice, become meaningfully worse in the years following his 2021 prediction. On that qualitative basis, the prediction is best assessed as right.

Elizabeth Holmes’s federal criminal trial concluded on January 3, 2022, with the jury finding her guilty on four counts (three counts of wire fraud and one count of conspiracy to commit wire fraud) and not guilty on four counts, with no verdict on the remaining three counts, which were later dismissed. She was subsequently sentenced to prison. These outcomes are documented in major news coverage and court summaries.

Sacks’s prediction was that there were greater-than-even (>50%) odds that she would be convicted at her Theranos criminal trial, not that she would be convicted on every single charge. Since the actual outcome included multiple felony convictions, his probabilistic forecast that she would be convicted at all was borne out by events.

Therefore, the prediction is right.

Jason’s prediction has two parts:

1. Guilt at trial (80% likelihood)
   Elizabeth Holmes was found guilty in her criminal trial. A federal jury in United States v. Holmes convicted her on four counts (three wire‑fraud counts and one conspiracy to commit wire fraud) on January 3–4, 2022. (en.wikipedia.org)
   Qualitatively, assigning an 80% probability to an outcome that did occur is directionally reasonable.

2. Time actually served: ~32.5 months ("over/under")
   This part is clearly wrong.

   • In November 2022, Holmes was sentenced to 135 months (11 years, 3 months) in federal prison for defrauding Theranos investors. (fda.gov)
   • She reported to Federal Prison Camp Bryan in Texas on May 30, 2023 to begin serving that sentence. (en.wikipedia.org)
   • After routine good‑conduct reductions, the Bureau of Prisons has shortened her effective sentence, but she is still projected to serve about nine years total, with release now scheduled in 2032 (variously reported as August 16, 2032 or December 29, 2032, depending on the update). (theguardian.com)
   • As of the current date (late 2025), she has served roughly 2.5 years (~30 months) and still has many years remaining, so her ultimate time served will be far above the 32.5‑month over/under.

Given that the core quantitative component of the prediction—time actually served—is off by a factor of roughly three even after accounting for sentence reductions, the overall forecast does not match reality, despite correctly favoring a guilty verdict. Therefore the prediction is best graded as wrong.

Available reporting from 2021–2025 shows that China’s 2021 crackdown on minors’ online gaming time (limiting under‑18s to one hour from 8–9 p.m. on Fridays, weekends, and holidays) was implemented and remains in force, with periodic reinforcement and additional draft rules, but it did not trigger notable large-scale protests or social unrest.

• The National Press and Publication Administration’s August 30, 2021 notice formally imposed the “three hours per week” limit for minors starting September 1, 2021, framed as an effort to curb youth gaming addiction and tighten control over the tech sector.(china.org.cn)
• Subsequent coverage and official summaries describe criticism, online complaints, and practical problems (e.g., game server overload during the allowed hour, minors seeking workarounds, and some being scammed while trying to evade the anti‑addiction system). These are documented as controversies and inconveniences, not as organized street demonstrations or broader unrest.(zh.wikipedia.org)
• Official and industry reports in 2022–2024 present the policy as largely successful in reducing minor gaming time, with authorities and state media emphasizing that minors’ game addiction has been “basically solved” and that coverage of minors by anti‑addiction systems exceeds 90%. Again, no mention is made of significant protests connected to these rules.(globaltimes.cn)
• Over the same period, when large‑scale protests did occur in China, they centered on other issues—zero‑COVID policies (the “white paper” protests), the Ürümqi fire, labor disputes at Foxconn, bank deposit freezes in Henan, and healthcare reform affecting pensioners—not on video‑game regulations.(en.wikipedia.org)

Given more than four years since the policy took effect and the absence of any widely reported, regime‑salient protests or unrest attributable to the minors’ gaming limits, the specific outcome Jason was worried about—those limits contributing to significant social unrest or protest in China in the ensuing years—did not materialize. While his statement was explicitly probabilistic (5–10% chance), if we evaluate it on the binary question “did the forecasted type of event occur?”, the answer is no, so the prediction is best scored as wrong for this single realized outcome.

Within 45 days of the early‑September 2021 recording (i.e., by about October 19, 2021), the main nationwide protest against Texas SB8 was the October 2, 2021 “Rally for Abortion Justice” / 2021 Women’s March, organized in direct response to the Texas Heartbeat Act and held in 600+ locations across all 50 states. Contemporary reports and later summaries describe total turnout as “tens of thousands” or “more than 100,000” nationwide, with organizers’ own numbers in the ~120,000–250,000 range—well below anything that could reasonably be called a roughly one‑million‑person march. No credible source reports a million‑person (or near‑million) U.S. march over SB8 in that period, so Friedberg’s prediction did not come true.

Texas Senate Bill 8 (the Texas Heartbeat Act) took effect on September 1, 2021 and bans abortions after detection of embryonic or fetal cardiac activity (around six weeks), enforced exclusively through private civil lawsuits rather than government officials. As of 2025 its status is still listed as “in force.” (en.wikipedia.org)

Litigation never produced the kind of court ruling Sacks predicted. The U.S. Supreme Court declined to block the law, allowing only a narrow challenge to proceed while keeping SB8 in effect. (congress.gov) In March 2022, the Texas Supreme Court held that no state officials are authorized to enforce SB8, meaning providers had no suitable government defendant to sue; following that, the Fifth Circuit ordered dismissal of challenges to SB8’s private-enforcement mechanism. The net effect was to leave the law intact and operational rather than overturn it. (texasattorneygeneral.gov)

After Dobbs (June 2022), Texas’s near‑total abortion ban is enforced through both Chapter 170A (the “trigger” law) and SB8’s civil‑enforcement scheme; contemporary reporting in 2025 still describes the Heartbeat Act as a core part of Texas’s abortion restrictions, not as a law that has been struck down. (houstonchronicle.com) There is no court decision overturning SB8 or removing it from force.

Because SB8 has not been overturned in court and remains in force more than four years after the prediction, the claim that “this is going to get overturned” is best scored as wrong.

California held its gubernatorial recall election on September 14, 2021. Official results from the California Secretary of State and major news outlets show that the “No” vote on recalling Governor Gavin Newsom won by a large margin, approximately 61.9% “No” to 38.1% “Yes”, a spread of nearly 24 percentage points, which is widely characterized as a decisive or comfortable victory.

• The California Secretary of State’s certified results show about 7.94 million “No” votes vs. 4.84 million “Yes” votes on the recall question, confirming Newsom was not removed from office.
• Major coverage at the time (e.g., AP, NYT, LA Times) consistently described this as a solid/comfortable win, not a narrow escape.

Given that Sacks predicted Newsom would “cruise towards defeating the recall,” and Newsom in fact defeated the recall by a large, clear margin, the prediction is right.

Key sources (paraphrased):

• California Secretary of State – certified 2021 gubernatorial recall election results (showing ~61.9% No vs. 38.1% Yes on the recall question).
• Contemporary news summaries from major outlets confirming Newsom’s wide margin of victory and characterizing it as a decisive win.

Chamath’s prediction about how the Supreme Court would handle the Texas SB8 case specifically essentially came true.

1. Declining to rule on the underlying abortion issue in the SB8 litigation.
   When SB8 took effect on September 1, 2021, the Supreme Court refused to block it and explicitly said its order was “not based on any conclusion about the constitutionality of Texas’s law” and did not resolve any substantive claims, allowing the law to go into effect on procedural grounds instead of ruling on abortion rights. (texastribune.org) In Whole Woman’s Health v. Jackson (Dec. 10, 2021), the Court again focused on who could be sued (standing and enforcement issues), permitted only a narrow suit to proceed, and still did not decide whether SB8 was constitutional, while leaving the law in force. (supreme.justia.com) To this day, there is no Supreme Court merits ruling on SB8’s constitutionality; the law was effectively sidelined, not struck down. (en.wikipedia.org)

2. Allowing the Texas law to stand as a Texas-only regime and shifting abortion to the states.
   SB8 functioned as a Texas‑only six‑week ban from September 2021 until it was overtaken by Texas’s broader post‑Roe bans (trigger law and pre‑Roe statutes), and the Supreme Court never enjoined SB8 itself. (congress.gov) In Dobbs v. Jackson Women’s Health Organization (June 24, 2022), the Court then overruled Roe and Casey, held that the Constitution provides no right to abortion, and stated that authority to regulate abortion “lies with the 50 individual states” — a clear states’‑rights framing. (mondaq.com) As a result, Texas’s own bans now govern abortion almost entirely within Texas, consistent with the scenario Chamath described of the Court effectively letting Texas’s restrictive regime stand as a state matter.

While Dobbs did ultimately opine on abortion rights in a different case (Mississippi), in the Texas SB8 litigation itself the Court did exactly what he anticipated: it avoided ruling on the underlying abortion right and allowed Texas’s law to operate within Texas, as part of a broader move to return abortion regulation to the states.

There has been at least one subsequent case in which Texas SB8 was held unconstitutional on grounds focused on its novel standing/enforcement mechanism rather than the underlying abortion right, which matches the prediction’s core claim.

Key state‑court ruling (matches the prediction most closely)

• In Van Stean v. Texas Right to Life (Travis County, 98th Dist. Ct., Dec. 9, 2021), Texas state trial Judge David Peeples issued a declaratory judgment holding SB8’s enforcement mechanism unconstitutional. He found that SB8:
  • authorizes lawsuits by parties who have suffered no injury (a standing problem),
  • unconstitutionally delegates enforcement power to private parties, and
  • imposes a punitive $10,000 civil penalty without due process, under the Fourteenth Amendment. The ruling explicitly targeted SB8’s private “bounty” enforcement scheme and did not invalidate the abortion restriction on substantive abortion-rights grounds.(reason.com) Contemporary coverage described this as a judge ruling the Texas abortion law unconstitutional while noting that the decision did not stop the law’s practical operation statewide.(courthousenews.com)

Supporting federal ruling (also emphasizes the enforcement scheme)

• In United States v. Texas, federal District Judge Robert Pitman (W.D. Tex., Oct. 6, 2021) issued a 113‑page order blocking enforcement of SB8, concluding that the law was facially unconstitutional and describing its private-enforcement design as a contrived attempt to evade judicial review and established abortion precedents.(en.wikipedia.org) The Fifth Circuit quickly stayed his injunction, and the Supreme Court later dismissed the case as improvidently granted, so this did not produce a lasting nationwide invalidation, but it is still a judicial ruling that SB8 was unconstitutional.

Why this is scored as “right,” with caveats

• The prediction (normalized) was: “In a future case, the Texas SB8 abortion law will be ruled unconstitutional, primarily on grounds related to its standing/enforcement mechanism rather than on the underlying abortion right.” That has occurred: Judge Peeples’ decision squarely held SB8’s enforcement mechanism unconstitutional on standing/delegation/due‑process theories and explicitly framed the dispute as about civil procedure and enforcement, not about the substantive right to abortion.(reason.com)
• However, these rulings have had limited practical and precedential effect. SB8 remained formally “in force” and functioned as intended until Texas’s broader post‑Dobbs criminal abortion bans took over, and Texas and other states have continued to copy the private-enforcement model in new anti‑abortion measures.(en.wikipedia.org) No high appellate court (state or federal) has yet issued a definitive, generally controlling decision striking down SB8’s enforcement mechanism across the board.

So, while the ultimate, system‑wide invalidation Sacks likely envisioned has not (yet) happened, the literal prediction that a future case would hold SB8 unconstitutional, on standing/enforcement‑mechanism grounds rather than on abortion rights themselves, has in fact come true.

From 2021 to late 2025, both Apple’s and Google’s mobile app stores have clearly moved in the direction Chamath described—away from a single, closed payment/distribution stack, toward more fragmented and open models—especially in regulated markets.

Apple (focus of the prediction)

• Reader apps: After the Japan Fair Trade Commission probe, Apple globally allowed “reader” apps (e.g., Netflix, Spotify) to include an in‑app link to their own website so users can create/manage accounts and pay outside Apple’s IAP system, a capability rolled out in 2022 via the External Link Account Entitlement. (macrumors.com)
• South Korea alternative billing: Following Korea’s 2021 law banning mandatory in‑app billing, Apple introduced an external purchase entitlement so apps distributed solely in Korea can use third‑party payment providers, albeit with a 26% Apple commission and some disabled App Store features. (macrumors.com)
• EU DMA – alternative stores and payments: To comply with the EU Digital Markets Act, Apple changed iOS 17.4 so EU users can install apps from alternative app marketplaces (third‑party app stores) and developers can use alternative payment processors or link-out payment flows, with reduced App Store commissions. (apple.com) By 2025, multiple independent stores—AltStore PAL, Setapp Mobile, Epic Games Store, Aptoide, Skich, Mobivention—are live for iOS users in the EU, distributing apps outside Apple’s traditional App Store “walls.” (techcrunch.com)
• Steering and external deals: The EU fined Apple €500m for restricting developers from offering apps and deals outside the App Store and then forced Apple to revise EU terms so developers can steer users to cheaper options on the web or in alternative stores. (theguardian.com) Separately, a U.S. judge found Apple violated an earlier injunction by erecting new barriers to developers directing users to non‑Apple payment methods, underscoring that regulators are actively pushing Apple’s model toward more open payment routing. (investopedia.com)

Google / broader app‑store ecosystem

• User Choice Billing and sideloading: Google agreed, in a multi‑state U.S. settlement, to support “user choice billing” (developers can show rival billing options) and to make sideloading from outside Play easier, explicitly to increase competition in app distribution and payments. (investopedia.com)
• Epic v. Google – mandated opening: After Epic’s trial win, a U.S. judge ordered Google to restructure Play: it must allow third‑party app stores inside Google Play, stop requiring Play Billing, and permit developers to promote alternative payment systems and sideloading options; the Ninth Circuit upheld both the verdict and the permanent injunction in 2025. (theverge.com) This is a direct legal deconstruction of Google’s app‑store/payment bundle.

Net effect by late 2025: compared with 2021, iOS and Android app distribution and payments are meaningfully more open and fragmented—with reader‑app links, country‑specific alternative billing regimes, EU‑only third‑party iOS stores, and court‑ordered structural changes to Google Play. While Apple continues to resist and these openings are often geographically limited and encumbered with new fees, the direction of travel is very much toward the “beginning of the beginning” of deconstructing app stores and enabling alternative payment and distribution models. That aligns with Chamath’s directional prediction, so it is best classified as right.

Evidence since 2021 shows both Apple’s iOS and Google’s Android have faced sustained, high‑level antitrust and regulatory scrutiny specifically focused on their platform gatekeeping role, in ways clearly analogous to the Microsoft Windows antitrust era.

Key points:

1. Major U.S. antitrust cases targeting mobile platforms

   • The U.S. Department of Justice and state AGs filed an antitrust suit against Google’s Android app store (Google Play) in 2021, explicitly alleging monopolistic control over Android app distribution and in‑app billing—i.e., gatekeeping over the mobile software ecosystem, similar in structure to the Microsoft Windows/Internet Explorer case.
   • In March 2024, the U.S. DOJ and multiple states sued Apple for monopolizing smartphone markets through iOS restrictions, including limitations on app distribution, browser engines, and payment systems; the complaint repeatedly frames Apple as a platform gatekeeper and seeks remedies to open up iOS to more competition.

2. EU antitrust findings and structural remedies

   • The European Commission had already fined Google in prior years over Android’s tying and default‑setting practices (e.g., search and browser defaults on Android), explicitly comparing the case structure to historical Microsoft tying cases; enforcement and appeals continued into the mid‑2020s, reinforcing scrutiny on Android’s gatekeeping role.
   • The EU Digital Markets Act (DMA), enforced from 2023–2024 onward, designates both Apple and Google as “gatekeepers” and imposes specific obligations: allowing alternative app stores and sideloading on iOS in the EU, loosening default app and browser restrictions, and limiting self‑preferencing in app stores. These are remedial measures directly aimed at their gatekeeper control over mobile ecosystems, strongly echoing the remedies and behavioral constraints imposed on Microsoft Windows.

3. Clear focus on platform gatekeeping behavior

   • Across U.S. and EU actions, the theories of harm and remedies center on the same kind of conduct that defined the Microsoft Windows cases: controlling distribution channels (app stores), defaults and preinstallation, technical and contractual barriers to rivals (e.g., browser engines, payment systems), and leveraging OS dominance into adjacent markets.

Given this record—multiple high‑profile antitrust suits and regulatory regimes explicitly labeling Apple and Google as mobile "gatekeepers" and imposing or seeking remedies to open up their platforms—the prediction that iOS and Android would face Microsoft‑style antitrust scrutiny and gatekeeper‑focused remedies is substantially borne out.

Apple has in fact been forced by regulation to allow alternative app stores on iOS devices (at least in the EU), which matches the substance of Sacks’s prediction.

In January 2024, Apple announced changes to iOS and the App Store in the European Union to comply with the EU’s Digital Markets Act (DMA). These changes, delivered in iOS 17.4 starting March 2024, explicitly include new options for distributing iOS apps from alternative app marketplaces and APIs for creating such marketplaces.(macrumors.com) MacRumors’ coverage confirms that iPhone and iPad users in the EU can download and install apps outside the App Store through alternative app marketplaces.(macrumors.com)

Apple’s own and independent reporting emphasize that these options are limited to the EU and require both an EU-region Apple ID and physical presence in the EU, with access restricted or removed when users travel outside the bloc.(macrumors.com) Nonetheless, this still means that iOS devices now support alternative app stores under regulatory pressure, which is exactly the mechanism (“forced by regulation”) Sacks described. Subsequent reporting on the Epic Games Store operating as an iOS third‑party marketplace under the DMA regime further demonstrates that alternative app stores are now a practical reality on iOS in the EU.(theverge.com)

While the change is geographically constrained and tightly controlled by Apple, the core claim—that Apple would eventually be forced to allow sideloading or alternative app stores on iOS devices due to regulation or market pressure—has come true. Therefore the prediction is right.

Jason’s prediction had two main parts: (1) major media platforms like Netflix and Amazon will bundle video, games, podcasts, and music into one service and (2) Disney+ specifically will have games built into the Disney+ subscription.

What has happened by late 2025:

• Netflix now includes a catalog of games as part of every subscription (mobile and expanding to TV / cloud gaming) at no extra cost. (help.netflix.com) Netflix has also invested in companion podcasts around its shows, and is scheduled to host video podcasts from Spotify/The Ringer starting in 2026, but it still doesn’t offer a general music service. (netflix.com)
• Amazon comes closest to his vision: a Prime membership bundles Prime Video, Amazon Music (songs plus a large catalog of mostly ad‑free podcasts), and Prime Gaming/Luna for games, all under the same paid subscription. (aboutamazon.com)
• Disney+, however, has not added games to its subscription. Disney+ has been expanded by bundling/integrating Hulu and ESPN/ESPN+ content in the same or related apps, but this remains video (shows, movies, and sports), not interactive games. (theverge.com) Disney continues to license and publish Disney‑branded games (e.g., Disney Dreamlight Valley, Disney Speedstorm, Disney Illusion Island), but these are sold separately on consoles, PC, and mobile—not included as a perk of a Disney+ subscription. (en.wikipedia.org) Bob Iger earlier stated that Disney+ was intended as a pure TV-and-movie service with no games, and there’s still no evidence of a policy reversal by 2025. (gamespot.com)

So while he correctly anticipated some convergence (especially around Amazon, and partially Netflix), the specific claim that Disney+ would have games “built into Disney Plus” as part of the subscription has not materialized, and the broader idea that all these media companies would unify video, games, podcasts, and music into one offering is not generally true yet. Hence the prediction is best classified as wrong overall.

The prediction explicitly covered “over the next decade” from 2021, so the full forecast window runs to roughly 2031, while we are only about four years in as of late 2025. Empirically, although climate‑related disaster losses and federal aid have risen, they are still far below the specific scenario of four to five Hurricane Katrina–scale federal bailouts per year. Hurricane Katrina caused about $125 billion in damage and led to roughly $110.6 billion in federal spending on relief, recovery, and rebuilding.(en.wikipedia.org) NOAA’s billion‑dollar disaster data show that from 2020–2024 the U.S. averaged about $149 billion per year in total disaster damages—less than the damage from two Katrinas per year, let alone four or five—and that 2024’s disasters cost about $182.7 billion, with only a handful of individual events approaching Katrina‑scale losses.(climate.gov) Federal appropriations for disaster assistance across 2015–2024 totaled at least $448 billion, i.e., roughly $45 billion per year, again far below repeating Katrina‑level federal outlays multiple times annually.(files.gao.gov) On the other hand, there is clear evidence of growing economic strain and repricing in climate‑exposed real estate markets: major insurers have curtailed or stopped writing new homeowners policies in wildfire‑ and hurricane‑prone areas (e.g., State Farm and others in California), pushing many owners into state FAIR plans and highlighting a shift in how climate risk is being capitalized into property values.(kiplinger.com) Overall, the most concrete quantitative part of Friedberg’s prediction (4–5 Katrina‑scale federal bailouts per year) has not materialized so far, while the more qualitative part about a shift in economic value away from climate‑exposed real estate shows partial early evidence. Because the decade‑long horizon has not yet elapsed and future disaster losses and bailouts could still change the picture, the forecast is best classified as inconclusive (too early) rather than definitively right or wrong.

Chamath predicted that as climate change worsened, suburban and far-flung communities would be disproportionately damaged and their viability undermined because insurers would increasingly refuse to underwrite climate risks. Since 2021, research has shown that climate-amplified wildfire danger is now acutely threatening cities and suburbs once thought safe; about 115 million Americans, many in urban and suburban areas, are now in zones that could face catastrophic fires, driven by hotter, drier conditions and expanding development into the wildland–urban interface (WUI). (theguardian.com) Insurers are in fact retreating from these high‑risk areas. State Farm, the largest US home insurer, stopped issuing new homeowners policies in California in 2023, explicitly citing wildfire catastrophe exposure and rising costs, and other major carriers have similarly paused or severely limited new business while raising premiums. (en.wikipedia.org) A US Senate Budget Committee analysis found hundreds of thousands of homeowners in California lost coverage between 2018 and 2023, with the highest nonrenewal rates concentrated in rural and small‑town counties in the Sierra foothills and other fire‑prone regions – exactly the kind of exurban and WUI communities Chamath described. (sfchronicle.com) Where private insurers pull back, many households are pushed into bare‑bones state backstops or go uninsured: enrollment in California’s Fair Plan (the insurance of last resort for those unable to secure private coverage) has nearly doubled since 2023 to more than 600,000 policies by mid‑2025, and insurance problems caused roughly 13 percent of California home sales to fall through in 2024, signaling direct damage to local housing markets and community viability. (sfgate.com) In Florida and other hurricane‑exposed regions, sustained climate‑driven losses and insurer exits have contributed to such high premiums that about one in five homeowners now forgo property insurance altogether, again indicating that the availability and affordability of coverage is becoming a binding constraint on living in vulnerable coastal and exurban communities. (axios.com) Legal and policy scholarship now explicitly describes a rapid insurance retreat, as many major insurers exit high‑risk states or sharply curtail coverage where climate‑exacerbated disasters make some locations effectively too vulnerable to insure, leaving state programs and taxpayers to absorb growing risks. (scholarlycommons.law.emory.edu) While not every suburb or far‑flung community has yet become uninsurable, the core mechanism Chamath described — climate‑driven damage falling heavily on such areas as private insurers increasingly refuse or sharply limit underwriting of those risks — is clearly visible and already reshaping housing, insurance, and local fiscal realities, so the prediction is directionally right rather than premature or disproven.

Evidence since 2021 shows that as climate‑driven disasters make insurance in high‑risk fire and hurricane zones more expensive, unavailable, or inadequate, U.S. homeowners are increasingly relying on federal money to cover losses.

• Rising uninsurability/underinsurance in high‑risk zones. A 2024 Joint Economic Committee report finds that insurers are pulling out of wildfire and hurricane‑exposed states like California, Florida, Arizona and North Carolina, leaving some areas “uninsurable” and many homeowners underinsured. It notes that FAIR (last‑resort) plans and gaps in coverage are growing as climate‑fueled disasters worsen. (jec.senate.gov)
• Greater federal fiscal role and reliance on federal assistance. GAO reports that from FY 2015–2024, appropriations for federal disaster assistance totaled at least $448 billion, with an additional ~$110 billion in supplemental disaster aid already appropriated for FY 2025 alone, and that disaster declarations and expectations of federal support have increased. GAO explicitly identifies the rising number of natural disasters and increasing reliance on federal assistance as key drivers of federal fiscal exposure. (files.gao.gov)
• Federal programs explicitly filling uninsured/underinsured gaps. FEMA’s Individuals and Households Program provides grants for home repair/replacement and other uninsured or under‑insured disaster‑caused expenses and serious needs, as highlighted in guidance to Maui wildfire survivors. (cca.hawaii.gov) In 2024 FEMA changed its rules so that, for disasters declared on or after March 22, 2024, housing assistance can be paid even when insurance payouts already exceed the previous FEMA maximum, specifically to address uncovered losses; FEMA expects Individual Assistance awards to rise as a result. (files.gao.gov)
• Concrete climate disasters where federal cash and loans covered property losses. After Hurricane Ian (2022), federal support to Florida totaled $1.74 billion, including $684 million in FEMA Individual Assistance and hundreds of millions in SBA disaster loans and NFIP payments to households and the state. (presidency.ucsb.edu) The JEC report underscores that when homeowners cannot rely on insurance payments to rebuild, they turn to FEMA and the SBA, and notes how quickly FEMA’s Disaster Relief Fund was drawn down by Hurricanes Helene and Milton, necessitating large supplemental appropriations. (jec.senate.gov)

Taken together, these developments match Jason’s prediction: as climate change renders more properties in high‑risk zones uninsurable or underinsured, the federal government is indeed "stepping in" more often—through FEMA grants, SBA loans, HUD disaster housing funds, and NFIP payouts—to compensate owners for disaster losses.

Evidence since 2021 shows U.S. governments have begun to restrict or effectively prohibit new home construction in some of the highest‑risk fire and flood zones, explicitly to avoid unsustainable public disaster and insurance costs, though these moves are still limited and contested.

Key points:

• State‑level prohibition trend in high‑risk flood (hurricane) zones – Vermont:

  • Vermont’s 2024 Flood Safety Act (Act 121) creates the state’s first statewide system to regulate new development in river corridors—the areas where the most destructive flooding and erosion occur. The law gives the Department of Environmental Conservation authority over all development in mapped river corridors and is explicitly aimed at stopping new buildings in the most hazardous areas, rather than repeatedly paying to rebuild them.(rivernetwork.org)
  • Coverage of the bill notes that it is “time to stop building new structures” in these corridors and that the policy is intended to reduce flood damages and the huge costs of disaster recovery for taxpayers.(vermontpublic.org) This directly matches the rationale in the prediction (government cannot keep underwriting repeated losses).
  • Vermont Public’s legislative wrap‑ups emphasize that the Act will limit or restrict new construction in river corridors statewide starting in the next few years, representing a structural shift away from allowing new homes in some of the most at‑risk flood (and hurricane‑exposed) areas.(vermontpublic.org)

• Local and federal‑linked rules already effectively banning new housing in core floodplains, now being tightened:

  • Even before 2021, many communities participating in FEMA’s National Flood Insurance Program already had regulations where new development is effectively banned in the 100‑year floodplain to remain eligible for federal insurance; local officials in Royalton, Vermont describe such pre‑existing rules as already “effectively” banning new development in those mapped areas and were pushing to extend those prohibitions to a wider 500‑year floodplain and river corridors to reduce future damage and secure higher disaster‑aid reimbursements.(vermontpublic.org) That is precisely a government response to the cost of repeated losses.

• Fire zones: emerging prohibitions on added housing in extreme‑risk burn scars:

  • After catastrophic Los Angeles wildfires, California Gov. Gavin Newsom issued an executive order (July 2025) granting local governments in LA County authority to suspend or restrict SB 9 housing development (lot splits and duplexes) in very high fire hazard burn‑scar zones such as the Palisades, Malibu, and parts of Altadena. The order includes a pause on SB 9 projects and lets localities designate where such housing “is or isn’t allowed” to ensure fire‑resilient recovery.(gov.ca.gov)
  • Using that authority, Los Angeles Mayor Karen Bass issued an emergency order that prohibits the city from accepting or processing any new SB 9 applications within the Palisades Fire area’s Very High Fire Hazard Severity Zone, explicitly citing evacuation gridlock and safety concerns in this extreme‑risk zone.(cbsnews.com) While single replacement homes are still allowed, this is a concrete example of U.S. governments beginning to say "you can’t add more homes here" in the highest‑risk wildfire areas.

• Motivation is explicitly about public and taxpayer cost, not just safety:

  • Discussion around the Flood Safety Act repeatedly frames unchecked development in river corridors as imposing “mountain costs associated with flood recovery” on taxpayers and emphasizes that preventing building in these areas is more cost‑effective than continually rebuilding.(vermontpublic.org) This is directly aligned with the podcast’s logic about the public sector no longer being able to underwrite these risks indefinitely.

• Counter‑evidence: bans are still narrow, and risky building continues elsewhere:

  • At the same time, investigations in Texas and elsewhere show thousands of new homes continue to be built in floodplains even after major disasters like Hurricane Harvey, highlighting that many jurisdictions still permit extensive construction in high‑risk zones.(texasstandard.org) In wildfire areas, most policy response is still focused on hardening codes and defensible space requirements rather than blanket no‑build zones.

Taken together, there is now clear, documentable evidence that some U.S. state and local governments have begun to prohibit or sharply restrict new residential construction in specific, clearly defined high‑risk fire and flood zones, motivated in part by the unsustainable public cost of repeatedly rebuilding them. That matches the core of Jason’s prediction, even though the shift is geographically limited and far from universal.

Evidence shows New York City and New York State have moved primarily toward legalizing and regulating basement apartments, not prohibiting them, and flood‑resilient building rules are largely limited to mapped flood zones rather than all new structures.

1. Policy on basement apartments:

   • State legislation backed by NYC officials (Kavanagh/Epstein bill S2276/A1075) is explicitly intended to let the city create a program to legalize and bring existing basement and cellar apartments up to safety standards, not to ban them. (nysenate.gov)
   • The 2024 state budget created a pilot program to legalize basement and cellar apartments in selected NYC neighborhoods, giving owners a pathway to convert them into lawful housing that meets health and safety standards. (thecity.nyc)
   • A 2025 report by the New York State Bar Association recommends legalizing safe basement apartments as a tool to reduce homelessness, reinforcing that the policy direction is toward regulated legalization, not blanket prohibition. (nysba.org)
   • NYC’s 2024 “City of Yes for Housing Opportunity” zoning reforms and related implementation work establish accessory dwelling units (including some basement units) in city codes, while targeting restrictions only to coastal and inland flood‑prone areas, rather than banning all basement residences. (thenyhc.org)

   Overall, tens of thousands of basement units are being treated as part of the housing stock to be legalized and made safer, not declared universally “not livable” or illegal to occupy. This is the opposite direction of the prediction’s claim that NYC would “basically have to say … you can’t live in a basement anymore.”

2. Building practices / Miami‑style ground floors:

   • NYC’s flood‑resistant construction rules (Appendix G of the Construction Codes) require, within the 1% annual‑chance floodplain, that the lowest occupied floor be above the design flood elevation and that any enclosed space below that level be used only for parking, storage, or building access; basements and cellars are not allowed in new residential structures in these zones. (nychazardmitigation.com)
   • Zoning for Coastal Flood Resiliency (adopted May 2021, before the Sept 2021 podcast) and its subsequent implementation give owners in flood zones flexibility to elevate buildings and adjust ground‑floor uses to meet those code standards; they do not newly mandate Miami‑style open first floors citywide after Hurricane Ida. (nyc.gov)

   These measures mean that in designated floodplains, some new buildings already have parking or limited uses at ground level so floodwaters can pass through, but that framework predates the prediction and is geographically constrained. There has been no general shift to designing all new NYC structures with Miami‑style open first levels.

Because (a) policy has moved toward legalizing and regulating basement apartments rather than broadly prohibiting people from living in them, and (b) flood‑resilient ground‑floor design is confined to mapped flood zones and generally pre‑dated the statement rather than emerging as a new, citywide standard, the prediction that NYC would move in the described direction is best characterized as wrong.

It is too early to definitively judge a 20–30 year climate trend prediction made in 2021.

Why it’s too early

• The prediction is explicitly about “the next 20 or 30 years” from 2021, i.e., roughly 2021–2041/2051. As of late 2025, only ~4 years (about 13–20% of the forecast window) have elapsed, so the long‑term trajectory cannot yet be fully validated or falsified.

What has happened so far (2021–2025) Evidence since 2021 is broadly consistent with the mechanisms he describes:

• The IPCC’s 2021 Sixth Assessment Report (AR6 WG1) projected that continued warming will make hot extremes hotter and more frequent, and heavy precipitation events more intense and frequent, with dry regions seeing more severe droughts in many areas.
• Recent WMO and IPCC‑cited assessments show:
  • Record global temperatures: 2023 was assessed as the hottest year on record, with 2024 also among the hottest.
  • Increasing heatwaves: Multi‑continent extreme heat events in Europe, North America, and Asia have intensified, consistent with projections of hotter hot extremes.
  • Severe droughts and megadrought conditions affecting regions like the western United States and parts of the Mediterranean and Horn of Africa, in line with drying trends for some subtropical and mid‑latitude regions.
  • More intense rainfall and floods in many places, e.g., catastrophic flooding events in Western Europe (2021), Pakistan (2022), and parts of China and Libya (2023), consistent with the expectation of heavier downpours in already wet or monsoon‑affected regions.

These observations align with mainstream climate science that was already well‑established by 2021, not just with this podcast prediction. However, the claim concerns a multi‑decadal escalation through the 2040s–2050s. Current data only show the initial slice of that period and cannot yet prove whether the long‑term rate and pattern of escalation will match his statement.

Because the full 20–30 year period has not yet passed, and climate variability over shorter spans can be large, the prediction cannot be declared fully right or wrong at this time, even though early evidence is consistent with it.

Friedberg’s prediction was about the YouTube version of All‑In episode E45 (video ID LNLcL5E95bs) getting a 6–10% downvote rate. The episode’s YouTube mirror is confirmed by third‑party summary sites like Glasp, which show the video’s title, date (September 4, 2021), and view count, but do not expose like or dislike statistics for that video.【7search1】

Between November and December 2021, YouTube permanently hid public dislike counts and removed the dislike field from the official YouTube Data API, so there is no longer an official public source for per‑video dislike totals.【12search0】【12search3】 Independent reporting confirms that only creators can see the dislike counts privately in YouTube Studio.【12search2】

There is a community project, Return YouTube Dislike, which reconstructs dislike numbers using archived data from before the API shutdown plus estimates based on extension‑user behavior and view/like ratios—its own documentation explicitly describes the counts as coming from a mix of scraped archives and extrapolated estimates, not authoritative YouTube data.【11search0】【12search9】 Although the project offers an API that can, in principle, return an estimated like/dislike breakdown for a specific video ID,【11search3】 this environment cannot safely query that per‑video endpoint directly, and even if it could, those figures would still be modeled estimates rather than ground truth.

A web search also did not surface any archived screenshots, blog posts, or forum discussions that recorded this particular video’s like/dislike ratio at any point in time.【7search1】【8search1】 Without access to the private analytics for the All‑In YouTube channel or a reliable historical snapshot of the exact like and dislike counts, the actual downvote percentage for this episode cannot be determined.

Because enough time has passed for the outcome to be knowable in principle, but the necessary data are not publicly accessible or reliably reconstructed, the correctness of the 6–10% downvote‑rate prediction cannot be established either way. Hence the result is ambiguous rather than simply “too early to tell.”