E150: Israel/Gaza escalating or not? EU censorship regime, Penn donors revolt, GLP-1 hype cycle

1. Antecedent was satisfied (Israel did invade Gaza by land after 2023‑10‑20). A Reuters key‑events timeline records that Israel launched its ground offensive into Gaza on 27 October 2023, a week after the podcast release, so the condition “if Israel proceeds with the ground invasion” was met. (reuters.com)

2. Mass protests across the Middle East and globally during the invasion period. A global dataset on Gaza‑war protests finds nearly 48,000 protest events worldwide between October 7, 2023 and late 2025, with about 22,000 (46%) in the Middle East alone. (en.wikipedia.org) In the U.S., the Crowd Counting Consortium reports 1,869 pro‑Palestine events from October 7–November 26, 2023 (the first month of the ground invasion), noting that this movement was producing “scores of events” with thousands of participants almost every day across the country. (countingcrowds.org) UK reporting similarly describes huge London marches and solidarity actions in scores of other cities on six consecutive weekends by November 25, 2023. (english.ahram.org.eg) A Times of Israel/AP roundup on October 29, 2023—specifically tying that weekend’s demonstrations to Israel expanding its ground operation—describes tens of thousands marching in Britain, France, Switzerland, Pakistan and elsewhere. (timesofisrael.com) This evidence shows sustained, near‑daily protest activity, heavily concentrated in the Middle East but also spread across Europe, North America, and Asia during the invasion.

3. Riots and serious unrest did occur during this period. An NPR/OPB report on October 30, 2023, written as Israel’s intensified ground operations “pushed into a fourth day,” describes tens of thousands joining pro‑Palestinian protests that weekend in New York, London, Madrid, Casablanca, Istanbul, Islamabad and other cities and notes that protesters in Dagestan (Russia) stormed an airport as a Tel Aviv flight landed, forcing its closure. (opb.org) The global‑protests dataset also documents that roughly 1% of the nearly 48,000 Gaza‑war demonstrations turned violent, which still amounts to hundreds of riot‑like or disorderly events worldwide. (en.wikipedia.org) Together, these support Sacks’s expectation of recurring protests with intermittent riots once a ground invasion began.

4. Major, ongoing social‑media “eruptions” over Gaza. Meta announced on 19 October 2023 that, in the days after October 7, it had already removed over 700,000 pieces of content related to the Israel–Hamas conflict and set up a special operations center to handle the surge, indicating a massive volume of online posting and controversy. (dig.watch) Human‑rights monitoring citing Human Rights Watch reports that between October 7 and November 14, 2023, Israel’s Cyber Unit sent platforms about 9,500 takedown requests—60% to Meta—with an estimated 94% compliance rate, reflecting constant content disputes about Gaza on major platforms. (graduateinstitute.ch) HRW‑based analyses further document at least 1,050 cases of censorship of Palestine‑related content on Facebook and Instagram from October–November 2023 across more than 60 countries, nearly all targeting peaceful pro‑Palestinian speech. (decodingaffairs.com) This level of posting, moderation and counter‑moderation is consistent with sustained, large‑scale social‑media “eruptions” throughout the early ground‑war period.

5. Assessment. Sacks did not make a precise quantitative claim, but he predicted that if Israel launched a ground invasion, we would see protests/riots and major social‑media storms over Gaza on an almost daily basis around the Middle East and globally. Israel did invade by land after the podcast date, and the ensuing weeks featured extremely frequent protests (often dozens per day in the U.S. alone, with tens of thousands of events worldwide and an especially dense concentration in the Middle East), sporadic riots, and continuous, large‑scale conflict on social media. Given the documented frequency and breadth of these phenomena, the prediction is best characterized as right in substance, even if the phrase “almost every day” cannot be verified literally for every single day.

Evidence since the DSA came into force is mixed and does not clearly show a simple “one global DSA-style rulebook” outcome, but it also does not clearly refute the possibility of DSA-driven global convergence.

On the segmentation / non‑convergence side:

• The DSA’s legal scope is explicitly limited to services offered to users in the EU; it does not require platforms to apply its standards globally, only to EU users.(commission.europa.eu)
• Platforms have rolled out clearly EU‑only features to meet DSA obligations. TikTok, for example, allows users in the EU/EEA to turn off personalised “For You” recommendations and use a non‑personalised search feed, and to restrict personalised ads for teens, specifically to comply with the DSA, while explicitly not offering the same option in non‑EU markets like the UK.(geekculture.co) This shows they are willing to maintain different moderation/recommendation regimes by region rather than standardise globally.
• TikTok’s withdrawal of the TikTok Lite rewards programme applies only in the EU in response to a DSA investigation, not worldwide.(reuters.com)
• Meta and Google’s decisions to halt or limit political ads are being taken specifically for the EU in response to the separate EU political‑ads transparency regulation, rather than exported to the US market, again indicating distinct regional rule sets.(ft.com)

On the convergence / Brussels‑effect side:

• Some policy and think‑tank analyses argue that, because it is operationally simpler, platforms may choose to apply DSA‑driven standards more broadly than required, and warn that this risks imposing EU speech rules outside Europe.(itif.org) This supports Sacks’s logic (the risk that global companies will standardise on the strictest regime), but these pieces mainly describe incentives and possibilities, not clear proof that US services now fully run on a DSA‑style model.
• A U.S. House Judiciary Committee Republican report, citing non‑public materials, claims the DSA is “forcing companies to change their global content moderation policies” and that EU regulators expect platforms to modify worldwide terms and conditions to satisfy DSA obligations.(judiciary.house.gov) However, this is an adversarial interpretation based on limited evidence; it is not independently corroborated by platform announcements that they have simply adopted EU rules as their default everywhere.
• Major platforms have introduced some globally uniform safety features that are clearly responsive to regulatory and political pressure, including from the EU. Meta’s “Teen Accounts,” with stricter defaults and controls for minors, are being rolled out worldwide across Instagram, Facebook and Messenger, not just in the EU.(about.fb.com) Likewise, Facebook’s global “Feeds” tab offering a chronological, less‑personalised feed makes it easier to satisfy the DSA’s requirement that EU users can opt out of algorithmic personalisation, while also being available to users in other regions.(in.investing.com) These moves are consistent with a partial Brussels effect, but they are also driven by lawsuits, youth‑safety scrutiny, and political pressure in the US and elsewhere, so causality is not clean.

Netting this out: by late 2025, platforms do sometimes export DSA‑aligned features globally, but they also clearly maintain EU‑specific restrictions and US‑specific practices, and in some cases roll out more invasive data uses outside the EU (for example, using Meta AI chat data for ad personalisation globally except in the EU, UK and South Korea).(reddit.com) That pattern does not cleanly match a world where companies simply adopt a single DSA‑style moderation/compliance model everywhere.

Because there is credible evidence in both directions—regional differentiation in some areas, partial globalisation of DSA‑aligned protections in others—and no clear, widely acknowledged shift to a unified DSA‑style regime for U.S. users, the prediction that platforms would standardise on DSA rules globally is neither clearly confirmed nor clearly falsified. The outcome is therefore best classified as ambiguous.

Evidence since late 2023 shows that the large tech and social media firms largely chose to comply with the EU’s Digital Services Act (DSA) rather than exit or mount existential resistance, matching the core of Sacks’s prediction.

• VLOPs implementing DSA obligations rather than leaving the EU. The European Commission designated very large online platforms and search engines (VLOPs/VLOSEs) such as Facebook, Instagram, TikTok, YouTube, Amazon, etc., and required them to file regular transparency reports and adopt extensive content‑moderation and risk‑management systems. These companies have continued to operate in the EU and have been submitting the mandated reports rather than withdrawing from the market.(digital-strategy.ec.europa.eu)
• TikTok’s behavior is typical: rapid compliance infrastructure and faster responses to government orders. TikTok is now on its fifth DSA transparency report, describing how it removed tens of millions of policy‑violating posts and highlighting that it cut median response time to government orders from six hours to three, even as such orders increased. That is precisely the kind of fast, systematic takedown and transparency behavior the DSA demands.(newsroom.tiktok.com)
• Platforms are inside the DSA enforcement regime, not defying it. The Commission has opened formal proceedings and issued preliminary findings that X, Meta, TikTok, AliExpress, and others breached various DSA obligations (e.g., data access for researchers, dark patterns, inadequate reporting tools). But these are enforcement disputes inside a framework the companies have accepted—they remain in the EU, are subject to investigations and fines, and continue adjusting their systems instead of simply refusing the rules or pulling out.(digital-strategy.ec.europa.eu)
• X as the most defiant case still commits to DSA compliance. Even X (Twitter), which withdrew from the voluntary disinformation code and has been the most openly hostile to EU content‑moderation politics, has not exited the EU. It created a DSA portal for illegal‑content reports and has stated it “remains committed to complying with the Digital Services Act,” while fighting specific findings via the regulatory process.(cybersecforum.eu)
• Some narrow legal pushback, but no broad refusal. Companies like Zalando and Amazon have taken the Commission to court to contest their designation as “very large online platforms” and related supervisory fees, yet the EU General Court has upheld these designations and the firms remain subject to the stricter DSA regime. These targeted legal challenges show normal regulatory friction, not a strategy of rejecting or fleeing the DSA framework.(euronews.com)
• Voluntary codes vs. binding law. Around the DSA‑linked disinformation code of practice, most major players (Google, Meta, TikTok, Microsoft, etc.) signed on or treated it as a benchmark for compliance; X is a notable hold‑out. Even where some firms later scaled back on voluntary code commitments, they still must—and do—show equivalent measures under the binding DSA rules.(disa.org)

Taken together, major platforms have overwhelmingly opted to stay in the EU and build substantial compliance machinery that implements the EU’s content and data‑access demands, including rapid takedown and transparency workflows. While they do litigate details and occasionally resist on the margins (especially X), the broad pattern is accommodation rather than principled refusal or exit. That makes Sacks’s main empirical claim—that tech companies would “fold” to EU regulators instead of standing up to or leaving them—substantively right, even if his language about doing “whatever” commissioners want overstates the lack of pushback at the edges.

Evidence from 2023–2025 shows a mixed picture, not the across‑the‑board, non‑resistant compliance Sacks predicted.

After the Digital Services Act (DSA) began applying to large platforms in 2023, the European Commission opened the first formal DSA infringement proceedings against X (formerly Twitter) in December 2023 over suspected failures in risk management, content moderation, transparency, and alleged “deceptive design.” This was triggered partly by X’s handling of content around the Hamas attacks and the Israel–Hamas war, indicating regulators saw continued non‑compliance rather than a platform simply doing whatever they asked. X is still under in‑depth investigation, not in clear, full compliance. (europeansting.com)

In 2024–2025, the Commission also opened multiple formal DSA proceedings against Meta (Facebook and Instagram) over deceptive ads, disinformation, election‑related monitoring tools, and protection of minors. (cyprus.representation.ec.europa.eu) By 2025 it had preliminarily found Meta and TikTok in breach of transparency obligations (including researcher data access) and found Meta in breach of obligations to provide simple mechanisms for reporting illegal content, criticizing its use of dark patterns. (digital-strategy.ec.europa.eu) Meta and TikTok have publicly disputed aspects of these findings and raised concerns (e.g., TikTok citing conflicts with GDPR), rather than passively accepting all demands. (apnews.com)

Separately, Ireland’s media regulator has launched its own DSA‑based investigation into X over inadequate appeal and complaints mechanisms, again signaling that regulators view compliance as insufficient and contested. (reuters.com)

There are instances where companies arguably “folded,” such as TikTok’s rapid, “voluntary” suspension of the TikTok Lite rewards feature in the EU just days after the Commission opened a DSA investigation and threatened interim measures; that move closely tracks what EU enforcers wanted. (reuters.com) But alongside such concessions, the record is full of probes, preliminary breach findings, appeals, and public disagreement.

Overall, major platforms have partially adapted and sometimes yielded to EU pressure, but they have not uniformly or fully complied with every takedown or policy demand without resistance. The ongoing investigations and alleged breaches directly contradict the prediction that companies would simply be “craven,” “fold,” and “do whatever these EU commissioners want.”

Evidence so far points in the direction Chamath predicted, but it’s still too early to say definitively.

1. Current double‑agonist benchmark (tirzepatide/Mounjaro–Zepbound). In the SURMOUNT‑1 phase 3 trial, the dual GIP/GLP‑1 agonist tirzepatide produced up to about 22.5% mean weight loss at 72 weeks in people with obesity, and ~23% sustained loss at 3 years in longer‑term follow‑up. (investor.lilly.com)

2. Triple‑agonist early results look numerically better. The leading triple agonist retatrutide (GLP‑1/GIP/glucagon) achieved up to 24.2% mean weight loss at 48 weeks in a phase 2 obesity trial, with broad improvements in cardiometabolic markers (blood pressure, lipids, HbA1c, liver fat). Multiple reviews describe this as the greatest weight loss reported in an obesity pharmacotherapy trial so far, and note that its average 17–24% loss in early studies exceeds that of currently approved drugs. (investor.lilly.com) Popular summaries similarly highlight that retatrutide’s ~24% loss appears higher than ~20–22% for tirzepatide and ~15% for semaglutide, while stressing the comparisons are across different trials. (menshealth.com)

3. Why the prediction can’t be called fully right (yet).

   • No head‑to‑head outcome trials of a triple agonist vs tirzepatide have been completed; all comparisons are indirect across different study designs and durations.
   • Triple agonists remain investigational (phase 3 ongoing, no approvals as of late 2025), so long‑term safety, real‑world effectiveness, and cardiovascular/renal outcomes vs existing double agonists are unknown. (pubmed.ncbi.nlm.nih.gov)

Because the key triple‑agonist drugs are not yet approved and have not been directly compared to current double agonists in definitive phase 3 or real‑world settings, we cannot conclusively say they are more effective, even though early data are promising and directionally support Chamath’s view.

As of November 30, 2025, it is too early to determine whether GLP‑1 drugs will ultimately be regarded, in retrospect over “years/decades,” as a disease‑modifying wonder‑drug class comparable in significance and ubiquity to statins.

What we can say now:

• GLP‑1 and related incretin drugs (e.g., semaglutide/Wegovy/Ozempic, tirzepatide/Mounjaro/Zepbound) have shown large, clinically meaningful weight‑loss effects and strong improvements in glycemic control in type 2 diabetes, and are being studied for cardiovascular and other outcomes.
• Major trials have already shown cardiovascular benefit in high‑risk patients with diabetes (e.g., LEADER for liraglutide, SUSTAIN‑6 for semaglutide) and, more recently, in some obesity populations without diabetes, leading to label expansions and intense clinical and commercial interest. These data support the idea that they are important and potentially disease‑modifying, at least for specific cardiometabolic risks, but they do not yet establish a decades‑long, statin‑like track record across broad primary‑prevention populations.
• Utilization has increased rapidly but is constrained by high cost, supply limitations, insurance coverage variability, and unanswered long‑term safety and effectiveness questions in general‑population obesity and other indications. Whether they will reach the ubiquity of statins (prescribed to tens of millions worldwide for long‑term primary and secondary cardiovascular prevention) remains unknown.

Because Chamath’s claim is explicitly about how these drugs will be viewed “when we look back on it” over a multi‑year/decade horizon, and we are only about two years past the prediction date, there has not been enough time to observe their long‑term safety profile, real‑world persistence, cost evolution, guideline penetration, and ultimate population‑wide impact. Those are exactly the factors that made statins the canonical, widely used disease‑modifying class.

So, while early evidence trends in the direction of GLP‑1 drugs being highly significant and possibly transformative, the core of the prediction is a long‑run, retrospective judgment about decades‑scale impact and ubiquity. As of late 2025, that judgment simply cannot yet be made with confidence, making the prediction inconclusive (too early) rather than clearly right or wrong.

As of late 2025, there is not enough long‑term, population‑level evidence to say whether Chamath’s prediction has come true.

What we do see so far

• GLP‑1 drugs (Ozempic, Wegovy, Mounjaro/Zepbound, etc.) have rapidly become more common. Estimates suggest on the order of 6–12% of U.S. adults have used anti‑obesity medicines, with GLP‑1s driving much of that increase.(columbiapsychiatry.org)
• Obesity itself has not been “replaced” as a main public‑health problem: U.S. adult obesity remains around 40%, with severe obesity still rising.(livescience.com)
• Significant issues around GLP‑1 use are emerging: common gastrointestinal side effects; concerns about muscle loss, gallbladder and pancreatic problems; high discontinuation rates (roughly half of users stop within a year) with substantial weight regain when the drugs are stopped; and cost and access inequities.(columbiapsychiatry.org) These are real and being discussed in medical and policy circles, but they are not yet on the same scale as the obesity epidemic itself.
• There is some early behavioral and societal signal (e.g., food‑waste changes, ethical debates, stigma, off‑label and cosmetic use), but current research does not show a clear, large‑scale pattern of people increasing caloric intake because they treat GLP‑1s as a “get out of jail free card.” If anything, the drugs’ main effect remains appetite reduction, and clinicians repeatedly stress that they are tools that must be combined with lifestyle change.(bu.edu)

Why this is inconclusive rather than right or wrong

Chamath’s claim is explicitly long‑term and conditional ("over time," if history is a guide), and it predicts a replacement public‑health problem driven by adaptation and behavior. GLP‑1 use only became widespread in roughly 2022–2024, so we have at most a few years of real‑world data. That is too short to judge whether:

• chronic, decades‑long use will produce new large‑scale physiological problems, and/or
• widespread moral‑hazard behavior (treating the drug as a permanent license to overeat) will offset benefits and create a new epidemic that functionally replaces today’s obesity burden.

Current evidence shows important side effects, equity issues, and high rates of weight regain on discontinuation, but these are still being characterized and managed and have not yet clearly become a new dominant public‑health crisis in place of obesity. Given the short time horizon and ongoing data collection, the fair assessment today is that the prediction’s truth value is still unknown, not yet testable.

Key pieces of evidence point in different directions and the outcome depends heavily on how the “long disrupted / short GLP‑1 winners” trade is defined and weighted.

1. There was extreme outperformance into late 2023. Novo Nordisk (NVO) and Eli Lilly (LLY) were among the best‑performing mega‑caps from mid‑2023 to October 2023, while GLP‑1‑“disrupted” medtech names like Insulet (PODD) were down ~45% in Q3 2023 on GLP‑1 fears. (statmuse.com) This matches the premise of an exaggerated spread between GLP‑1 beneficiaries and those seen as losers.

2. Some disrupted healthcare names then strongly mean‑reverted. Insulet’s monthly adjusted price went from about $132.57 in October 2023 to the low‑$300s by late 2024 and mid‑$300s by November 2025, more than doubling from its GLP‑1‑panic lows. (digrin.com) ResMed, cited as hit by GLP‑1 concerns, traded near $132 in October 2023 and had rebounded to the $230–$270 range by 2024–2025, roughly an ~80–100% move off those lows. (investors.com) A long basket focused on these names from late 2023 would have performed very well.

3. Not all ‘disrupted’ names recovered. Dexcom, another diabetes‑device player often mentioned in the GLP‑1 overhang narrative, closed at $85.97 on Oct 20, 2023 and around the high‑$60s in October 2025, a loss of roughly 20%, and as of late 2025 it is one of the S&P 500’s weaker performers. (statmuse.com) Including such names materially dilutes the long‑disrupted basket’s gains.

4. GLP‑1 winners diverged: Novo normalized sharply, Lilly did not. Novo Nordisk’s ADR closed October 2023 near $92.95 and ended 2023 just under $100, but by late 2025 had fallen to around the low‑$50s, with commentary noting the shares are down on the order of ~44–50% year‑to‑date 2025 amid guidance cuts, trial disappointments, and competitive pressure. (statmuse.com) Shorting NVO from late 2023 would have been profitable by 2025.

   In contrast, Eli Lilly closed $576.33 on Oct 20, 2023 and has since traded above $1,000 per share in 2025, remaining one of the top large‑cap performers on the back of explosive tirzepatide (Mounjaro/Zepbound) sales. (statmuse.com) A short in LLY initiated in late 2023 would have been very painful and still underwater by late 2025.

5. Street consensus recognized GLP‑1 ‘fear’ in medtech as overdone, but there is no single, standard trade. By 2024–25, major sell‑side research (e.g., Morgan Stanley) was explicitly calling GLP‑1‑driven medtech derating “overdone,” arguing that the long‑term impact on many sub‑industries would be limited, and turning more constructive on stocks like Insulet after large drawdowns. (tipranks.com) That supports Chamath’s directional view that the extreme dispersion between GLP‑1 winners and perceived losers would not persist indefinitely.

Putting this together:

• On the qualitative call that the GLP‑1 winner vs. loser spread was too extreme and would partially mean‑revert, the data support him: key “disrupted” medtech names staged very large rebounds, and at least one flagship GLP‑1 winner (Novo Nordisk) subsequently suffered a major de‑rating.
• On the specific trade claim (“it’s probably okay to be on the other side” – long disrupted / short GLP‑1 winners from late 2023 onward), results hinge on implementation. A basket overweight Novo and heavily tilted to big rebounders like Insulet and ResMed could have done quite well; a basket that also aggressively shorted Eli Lilly or included weaker disrupted names like Dexcom could easily have lagged or even lost money over this horizon.

Because there is no canonical, widely‑tracked “GLP‑1 winners” and “disrupted healthcare” index, and reasonable basket choices lead to materially different P&Ls, the prediction cannot be cleanly scored as right or wrong across the board. Hence the outcome is best characterized as ambiguous rather than definitively correct or incorrect.