E9: Trump has COVID, First debate reactions, Coinbase letter response & more

Evidence shows that Donald Trump received an 8‑gram intravenous dose of Regeneron’s antibody cocktail (REGEN‑COV2) on October 2, 2020, in line with the setup of the prediction. (fortune.com) He was hospitalized at Walter Reed from October 2–5 and discharged back to the White House after three days. (en.wikipedia.org) By October 8 his physician stated he was “devoid of symptoms” and anticipated a “safe return to public engagements” by October 10; Trump resumed campaign events starting October 12, indicating a functional recovery within about a week to ten days of diagnosis. (en.wikipedia.org)

Longer‑term, multiple White House medical reports in 2025 describe Trump (now in his late 70s) as in “excellent” or “exceptional” health and “fully fit,” noting only chronic issues like well‑controlled high cholesterol and mild chronic venous insufficiency; they list his 2020 COVID‑19 hospitalization only as past history, with no mention of ongoing respiratory, cardiac, or neurological impairment suggestive of severe long‑term post‑COVID complications. (reuters.com) There is no credible reporting that he developed long‑COVID, chronic oxygen dependence, or other major sequelae.

Taken together, Trump (1) recovered and returned to normal public activity within roughly a week to ten days, and (2) did not die or show documented long‑term severe complications attributable to his 2020 COVID‑19 infection. That aligns closely with Friedberg’s prediction, so the forecast is best judged as right.

Chamath’s prediction tied several concrete downstream effects to Trump’s COVID‑19 case; those effects largely did not occur.

1. Trump’s case did not yield uniquely “transparent data” on effective treatments

   • Coverage of Trump’s hospitalization repeatedly emphasized confusion and lack of transparency, not clarity. White House physician Sean Conley and other officials gave incomplete and even misleading information about Trump’s oxygen levels, imaging, and timeline, which major outlets described as “falsehoods, obfuscation, [and] evasion.” (washingtonpost.com)
   • The main therapies he received (Regeneron’s monoclonal antibody cocktail, remdesivir, dexamethasone) were evaluated primarily through randomized clinical trials and then FDA Emergency Use Authorizations, not through any special dataset from Trump’s case. The EUA for casirivimab/imdevimab on Nov. 21, 2020, explicitly rested on trial data, not on his treatment. (en.wikipedia.org)
   • In other words, his illness was not the source of “all of the most transparent data” about coronavirus care.

2. Public demand and actual access to the same treatment did not explode

   • Trump heavily promoted Regeneron’s antibody cocktail after his discharge, saying he wanted “everybody to be given the same treatment as your president,” and portraying it as a near‑miracle. (militarytimes.com)
   • However, once monoclonal antibodies from Regeneron and Eli Lilly received EUAs in November 2020, utilization was low, not sky‑high. By late 2020 and January 2021, federal officials reported that only about 20–25% of distributed antibody courses were being used, with some areas as low as 5%; HHS and the Surgeon General publicly urged hospitals to use them more. (cnbc.com)
   • A Washington Post report noted that the much‑anticipated demand surge—fueled by Trump’s glowing video about the Regeneron drug—never materialized; many patients and clinicians either didn’t know about the therapies or weren’t asking for them. (washingtonpost.com)
   • Access also remained constrained by practical barriers: the drugs required rapid administration early in illness, IV infusion capacity, and specialized centers; states repeatedly had to set up or reopen dedicated infusion facilities, and access varied by geography and eligibility. (cnbc.com)
   • So it did not become “hard for everybody to then not get some version” of Trump’s therapy; instead, supply often exceeded utilization, and many eligible people never received it.

3. Mask and testing “de‑escalation” in the ensuing months did not follow from his treatment

   • Through fall 2020 and winter 2020–21, U.S. public‑health policy continued to lean heavily on masks and other non‑pharmaceutical interventions, especially during a large winter surge. CDC mask recommendations remained in force, and on Jan. 20, 2021, President Biden signed Executive Order 13991, strengthening mask requirements on federal property and calling for maximizing public compliance with masks and distancing—this is the opposite of a treatment‑driven de‑emphasis on masking. (en.wikipedia.org)
   • CDC did not significantly relax mask guidance for the general (vaccinated) public until May 13, 2021, and that change was explicitly tied to vaccination progress, not to monoclonal antibody availability or Trump’s recovery. (en.wikipedia.org)
   • Testing volumes increased into late 2020, peaking around Jan. 15, 2021 at more than 2 million tests per day nationwide. Later declines in testing were attributed to falling case counts, harsh winter weather, vaccination rollout, and “testing fatigue,” not to any new clarity about curative therapies. (medicalxpress.com)
   • There remained substantial uncertainty and debate among experts about the role and indications of monoclonal antibodies; Trump’s unusual triple‑drug regimen was widely noted as unrepresentative of standard care and even as an example of “VIP syndrome,” not a template that resolved “what the right course of care is.” (abc7chicago.com)

Bottom line:
While Trump received an aggressive combination of leading‑edge treatments and then publicly hyped monoclonal antibodies, his case did not (a) generate uniquely transparent treatment data, (b) trigger broad, unavoidable access to “what the president got,” or (c) clearly lead to a rapid de‑emphasis of masks, testing, or uncertainty about care. The empirical record from late 2020 through mid‑2021 points in the opposite direction on all three counts, so Chamath’s prediction is wrong overall.

Why this prediction is wrong (as of Nov 29, 2025)

Friedberg predicted that within the coming years, routine prevention of infectious disease would move away from traditional vaccines to annual polyclonal antibody ‘cocktails’ that would effectively replace the current model of yearly flu shots.

Key checks:

1. Standard of care for flu prevention

   • In the U.S. and globally, the primary routine prevention for seasonal influenza in healthy populations remains annual vaccination, not antibody infusions or injections.
   • CDC guidance for the 2024–2025 season continues to recommend yearly influenza vaccination for almost everyone 6 months and older; there is no recommendation for routine antibody prophylaxis for the general population.

2. Antibody-based prophylaxis today

   • Monoclonal antibodies are used preventively in niche or high‑risk contexts:
     • RSV: Nirsevimab (Beyfortus) is recommended as a single-dose monoclonal antibody to prevent severe RSV in infants and some high‑risk young children, but this is a targeted product for a specific group, not a general annual cocktail for the whole population.
     • COVID‑19: Products like Evusheld (tixagevimab/cilgavimab) were used for pre‑exposure prophylaxis in certain immunocompromised patients, but authorizations were withdrawn or revised as variants reduced effectiveness, and they were never deployed as a universal annual replacement for vaccination.
   • These are monoclonal or sometimes combination antibody products targeting specific pathogens, not broad "polyclonal cocktails" that replace flu shots for mass routine prevention.

3. No broad shift in prevention model

   • Routine public‑health practice and guidelines in 2025 still center on vaccination for influenza, COVID‑19, and most other vaccine‑preventable diseases.
   • Antibody cocktails remain adjuncts or special‑population tools, not the new standard annual prophylaxis for the general population.

Conclusion
By late 2025, routine prevention of infectious disease has not shifted from vaccines to annual polyclonal antibody cocktails, nor have such cocktails functionally replaced yearly flu shots. The prediction is therefore wrong.

The prediction explicitly gives a 10–15 year horizon from 2020 (i.e., roughly 2030–2035) for yearly synthetic-antibody booster shots against new infectious diseases to become a general reality. As of the current date (November 29, 2025), we are only about five years into that window, so it is too early to determine whether this will be widely adopted by 2030–2035. Therefore the accuracy of the prediction cannot yet be judged.

Assessment

Overall, the conditional prediction substantially came true:

1. Condition (“if he recovers in about a week and looks strong”) was met.

   • Trump was hospitalized Oct 2–5, 2020, then returned to the White House after three days. By Oct 7 he released a video saying he felt “perfect” and describing a rapid turnaround after receiving Regeneron’s antibody cocktail.(factcheck.org) This is roughly one week from onset/diagnosis and matches the “recovers in about a week” scenario.

2. He publicly claimed that an effective cure already existed.

   • In multiple videos and interviews on Oct 7–8, Trump explicitly called the Regeneron monoclonal-antibody treatment a “cure,” saying it wasn’t just a therapeutic because it “just made me better,” and promised to make it available “free” to Americans.(forbes.com) Fact-checkers and news outlets repeatedly noted that no proven cure for COVID-19 existed at that time.(abcnews.go.com)

3. He used this narrative for political advantage in the weeks before the 2020 election.

   • Commentary at the time described this as a deliberate “cure” gambit to compensate for the lack of a pre‑election vaccine, emphasizing his personal recovery and the supposed miracle treatment as an October surprise.(washingtonpost.com)
   • At the final presidential debate on Oct 22, Trump again highlighted the antibody treatment as essentially a cure that got him “better very fast,” using it to underscore his handling of COVID-19 shortly before Election Day.(abcnews.go.com)

4. “Downplaying the need for a vaccine” – partially, but recognizably, yes.

   • In the same period, Trump said getting the antibody treatments to hospitals was “more important to me than the vaccine,” explicitly elevating the supposed cure over vaccination.(israelnationalnews.com) That clearly downplays the centrality of vaccines, in line with the prediction’s logic (even though he continued to tout that vaccines were coming soon and did not literally say “you don’t even need a vaccine”).(debates.org)

Conclusion

The specific wording “you don’t even need a vaccine” was not documented verbatim, and Trump continued to promote forthcoming vaccines. However, the core predicted behavior did occur: after a rapid recovery, he publicly portrayed his treatment as a present‑day cure, framed it as more important than a vaccine, and tried to leverage that story to flip criticism and gain political advantage in the closing weeks of the 2020 campaign. That is close enough in substance to rate the prediction as right rather than ambiguous.

Trump’s actual COVID-19 course and its electoral effects don’t match any of the three scenarios Sacks laid out.

Medical course:

• Trump announced his positive test on October 2, 2020, was hospitalized October 2–5 at Walter Reed, and returned to the White House October 5.(en.wikipedia.org)
• By October 7–8, his physician reported he was fever‑free for several days, symptom‑free for over 24 hours, and expected to be safe to resume public engagements around October 10.(aljazeera.com)
• He held his first full campaign rally in Sanford, Florida, on October 12, just 10 days after the diagnosis announcement and a week after hospital discharge.(washingtonpost.com)
  This is closer to a ~1–1.5 week disruption, not the 3–4 week limitation envisioned in his 40% scenario, and far from the “more serious” (ICU/ventilator/death) 5–10% scenario.

Political impact:

• After the first debate and Trump’s COVID diagnosis, national and swing‑state polling showed Biden’s lead stable or expanding, not a Trump boost. For example, CNBC/Change Research and other averages saw Biden’s margin in key swing states and nationally tick up in the week after the diagnosis.(cnbc.com)
• One Franklin Pierce/Boston Herald poll found Biden’s national lead widening from 5 points pre‑diagnosis to 21 points immediately afterward.(newsweek.com)
• Trump ultimately lost the 2020 election, and research on COVID and the 2020 vote finds that higher local COVID case counts reduced Trump’s vote share, indicating the pandemic—including perceptions of his handling of it—hurt him electorally.(pubmed.ncbi.nlm.nih.gov)

Putting this together:

• Reality combined a short illness (near his 50% medical timeline) with no clear political benefit and likely net harm, contrary to the “redounds to his political advantage” part of that scenario.
• His 40% scenario (3–4 week case that prevents campaigning) did not occur, as he was back to large rallies within about 10 days.(washingtonpost.com)
• The 5–10% serious‑outcome scenario also did not materialize.

Because the real world outcome (short course + neutral/negative political impact) is not one of the structured scenarios he assigned substantial probability to—especially his main 50% case tying a quick recovery to political advantage—the prediction is best judged wrong rather than ambiguous or inconclusive.

Why this prediction is hard to score

Sacks’ claim was conditional:

If Trump is unable to hold his usual in‑person campaign rallies during the remainder of the 2020 election cycle, this will likely shift the national vote margin by a few percentage points against him and could be sufficient to cause him to lose the election.

To evaluate it, we need to know:

1. Was Trump actually unable to hold his usual in‑person rallies for the remainder of the campaign?

   • After testing positive for COVID-19 on October 2, 2020, Trump canceled in‑person campaign events for a short period while quarantining and being treated. (en.wikipedia.org)
   • However, he resumed in‑person rallies on October 12, 2020, one week after being discharged from the hospital, and then held frequent large rallies in battleground states—often multiple rallies per day—through Election Day. (en.wikipedia.org)
   • So the strong condition "unable to hold his usual in‑person campaign rallies during the remainder of the election cycle" did not really occur; his rally schedule was briefly paused, then returned in force.

2. Did Trump lose by “a few percentage points”?

   • Joe Biden won 51.3% of the national popular vote to Trump’s 46.8%, a margin of about 4.5 percentage points, and won the Electoral College 306–232. (pollbludger.net)
   • That outcome matches the “few percentage points” loss Sacks described, but it happened even though Trump did resume large rallies.

3. Can we attribute that margin to an inability to hold rallies?

   • Because Trump was not actually prevented from holding his typical late‑campaign rallies (he held many after October 12), we are left with a counterfactual causal claim: if he had truly been unable to hold such rallies, would his margin have been even worse (or is that what caused the existing margin)?
   • Available analyses discuss COVID, Trump’s diagnosis, and broader issues as electoral liabilities, but they do not isolate the effect of a hypothetical total loss of rallies relative to the actual world where he still held many. (en.wikipedia.org)

Because the key condition of the prediction (no usual in‑person rallies for the rest of the campaign) did not actually hold, we cannot cleanly test whether that scenario would have shifted the vote margin by a few points and caused or cemented his loss. Enough time has passed, but the world we observed is not the world the prediction was about.

Conclusion: The prediction is about a counterfactual scenario that never fully happened, so its correctness cannot be determined from the real‑world outcome.

Therefore, the appropriate label is:

• result: ambiguous

There is no definitive way to verify the counterfactual claim that Biden was more electable than Elizabeth Warren (or other progressives) against Trump in 2020, so the prediction cannot be cleanly marked right or wrong.

What we can say with evidence

1. Biden’s actual performance in 2020

   • Joe Biden won the 2020 presidential election, defeating Donald Trump with 306–232 electoral votes and about a 4.5‑point popular vote margin (51.3% to 46.8%).
   • That clearly shows he was electable, but doesn’t by itself prove he was more electable than any other Democrat.

2. Warren and other progressives never faced Trump

   • Elizabeth Warren lost the Democratic primary to Biden and never became the nominee, so we have no direct election result to compare against Biden’s 2020 performance. Any comparison of their hypothetical general‑election performance is an ex‑ante modeling or pundit judgment problem, not an empirical one.

3. Polling/electability debates are not conclusive

   • While there were 2019–2020 polls testing Warren and others against Trump, such early head‑to‑head polling is not a reliable ground truth of what would have happened in November 2020, given campaign dynamics, COVID, economic shock, and events like the BLM protests — all of which unfolded in ways that would likely have interacted differently with a different nominee.
   • Political science literature generally treats such counterfactual matchups as inherently uncertain; small differences in early polls or models do not allow us to say with confidence that candidate A would definitely have done better/worse than candidate B.

Why the prediction is “ambiguous”

The core of the prediction is a counterfactual electability claim: that Biden’s demonstrated basic competence and centrism made him the most electable Democrat and that he would do better than a more progressive nominee like Warren. Biden did win, but we cannot observe the alternative world where Warren (or another progressive) was the nominee. Existing data (primary outcomes, head‑to‑head polls, models, pundit analyses) do not provide a decisive, widely accepted answer about whether Warren would have done better, worse, or about the same.

Because the key comparative claim (“more electorally successful than Warren in the same race”) is inherently untestable with real‑world results, the prediction is not verifiable or falsifiable in a strict sense. Therefore, the fairest classification is:

• Result: ambiguous – the necessary counterfactual evidence to judge it right or wrong does not and cannot exist.

What actually happened

• In the 2020 elections, Democrats won the presidency and—after the January 5, 2021 Georgia runoffs—effective control of the U.S. Senate with a 50–50 split and Vice President Harris as tie‑breaker, alongside control of the House.
• During approximately the first two years of Biden’s term (January 2021–January 2023), a large set of significant laws were passed and signed, including among others:
  • American Rescue Plan Act of 2021 (major $1.9T COVID relief and social spending package).
  • Infrastructure Investment and Jobs Act (2021), a roughly $1.2T infrastructure bill.
  • Inflation Reduction Act of 2022, with major climate, health‑care, and tax provisions.
  • CHIPS and Science Act of 2022, subsidizing domestic semiconductor manufacturing.
  • PACT Act expanding health benefits for veterans exposed to toxic burn pits.
  • Bipartisan Safer Communities Act, the most significant federal gun‑safety legislation in decades.

Contemporary political reporting and academic commentary widely described Biden’s first‑two‑year legislative record as one of the most substantial for a modern president, especially given the narrow margins in Congress.

Comparing to the prediction

Sacks’ prediction was conditional and qualitative:

1. If Democrats win presidency + Senate, there will be one‑party control and they can pass much legislation; Biden’s first ~two years would be very consequential.
2. If Republicans hold the Senate while Biden wins, we’d see gridlock and a less consequential presidency.

The world that actually occurred is case (1). In that world, there was unified (though narrow) Democratic control and a substantial volume of major legislation was in fact passed and signed in the first two years, and Biden’s presidency is widely regarded as consequential in policy terms compared with the gridlock scenario.

Because the realized conditional outcome matches his described consequences, the prediction is right.

During Joe Biden’s presidency, Democrats held effective control of both the House and the Senate in the 117th Congress from January 20, 2021, until January 3, 2023, giving them a governing trifecta. (en.wikipedia.org)

In that period, Congress passed and Biden signed into law all of the major Democratic-led legislative packages that managed to clear both chambers, including:

• The American Rescue Plan Act of 2021 (COVID relief), passed on party-line votes and signed March 11, 2021. (en.wikipedia.org)
• The Infrastructure Investment and Jobs Act (“bipartisan infrastructure law”), enacted in 2021. (en.wikipedia.org)
• The CHIPS and Science Act, signed August 9, 2022. (bidenwhitehouse.archives.gov)
• The Inflation Reduction Act of 2022, the core climate/health and tax package. (en.wikipedia.org)

Official veto records from the U.S. Senate and compiled veto lists show that Biden issued no vetoes at all during the 117th Congress; his vetoes begin only in 2023, after Republicans captured the House in the 118th Congress. (senate.gov) Those later vetoes were almost entirely aimed at Republican-led congressional disapproval resolutions targeting his administration’s regulations, not at blocking a Democratic Congress’s legislative agenda. (en.wikipedia.org)

Some ambitious progressive priorities (for example, the larger original Build Back Better framework and certain voting-rights or filibuster-reform efforts) failed because they could not secure unanimous Democratic support in the Senate, particularly from moderate Democrats; such proposals never passed both chambers, so Biden was never in a position to veto them. (en.wikipedia.org) These are therefore not counterexamples to the claim that he would sign what a Democratic Congress actually passes.

Given that (1) Democrats did control both chambers for the first two years of Biden’s term, (2) Biden signed the major legislation that Democratic leaders were able to move through Congress, and (3) he did not make a practice of vetoing or otherwise blocking bills passed by that Democratic Congress, Sacks’s prediction—that Biden would essentially "sign whatever [Democrats] pass" if they won Congress—has been borne out by subsequent events.

Summary
Available macro data show that the extremely loose-money, high‑velocity capital environment of late 2020 not only persisted but accelerated through 2021, with record levels of dealmaking, venture funding, SPAC issuance and cross‑border flows, all underpinned by near‑zero interest rates. Market behavior around and after the November 2020 U.S. election is consistent with a “floodgate” of risk‑on capital rather than a pause.

Key evidence

1. Near‑zero rates and easy money persisted through 2021

   • The effective federal funds rate averaged about 0.08% in 2021, after being slashed to a 0–0.25% target range in March 2020 and held there until rate hikes began in early 2022. (m.macrotrends.net)
   • The Fed also maintained large‑scale asset purchases in 2021, keeping financing conditions historically loose.

2. Record global M&A and corporate dealmaking in 2021

   • Global M&A volumes in 2021 hit roughly $5.8–5.9 trillion, the highest on record and about 60+% above 2020, explicitly attributed by analysts to cheap, widely available financing and booming equity markets. (euronews.com)
   • The U.S. accounted for nearly half of this, with U.S. M&A value nearly doubling versus 2020. (techstartups.com)

3. Venture capital and private markets saw an unprecedented surge in 2021

   • Global VC investment jumped from about $347B in 2020 to roughly $671B in 2021, a record year, with deal sizes and valuations rising sharply across all regions. (businesswire.com)
   • Crunchbase data show the first half of 2021 alone set new records, with ~$288B invested, surpassing the previous half‑year record just set in H2 2020—evidence of both continuation and acceleration of capital flows. (news.crunchbase.com)

4. SPACs and speculative capital flows peaked in 2021

   • SPAC statistics show 613 SPAC IPOs in 2021 raising about $162B, dramatically above 2020’s ~250 SPACs raising ~$83B, and far beyond any pre‑COVID year—clear evidence of “easy” speculative capital deployment. (en.wikipedia.org)

5. Cross‑border and FDI capital flows rebounded strongly in 2021

   • Cross‑regional commercial real estate capital flows between North America, Europe and APAC in H2 2021 rose about 60% year‑over‑year to a near‑record $77.5B, after being flat in H2 2020, with reports explicitly describing “robust” global capital flows in 2021. (cbre.com)
   • UNCTAD notes global FDI flows in 2021 rebounded sharply from 2020, more than doubling into developed economies, with U.S. inflows more than doubling on the back of cross‑border M&A. (unctad.org)

6. Election period behavior is consistent with a post‑election “floodgate” of risk‑on capital

   • U.S. equities staged a powerful rally from Election Day through November 2020; November was one of the strongest months for U.S. stocks in nearly a century, with the Dow up ~12% and small‑cap Russell 2000 up over 16%, supported by investor optimism as the election outcome and transition became clearer. (washingtonpost.com)
   • This rally, combined with the subsequent record‑setting 2021 risk‑asset and deal activity under a Biden administration, supports Friedberg’s claim that capital flows would keep surging after the election rather than stall, and that this was not contingent on a specific party winning.

Assessment vs. the prediction

Claim: The late‑2020 environment of huge, fast capital flows driven by ultra‑low rates would continue into 2021 and see another “floodgate” of capital movement around/after the November 2020 election, regardless of which side won.

Outcome:
– Interest rates remained near zero in 2021, and quantitative easing persisted.
– Multiple measures of capital flow velocity and volume (M&A volume, VC funding, SPAC issuance, cross‑border real‑estate and FDI flows) show record or near‑record highs in 2021, significantly above 2020 levels.
– Markets rallied strongly around and after the election and continued into 2021 as the recovery, vaccine news, and policy clarity combined with easy money.

Given this, the core directional prediction—that the capital‑flow mania of late 2020 would not stop in 2021 and would be further unleashed after the election—was borne out by subsequent data. Thus the prediction is best classified as right.

Macroeconomic performance (2021)

• U.S. real GDP grew 5.7% in 2021 after contracting 3.4% in 2020, a sharp rebound and clearly above the ~2% pre‑COVID trend, with broad-based strength in consumption and investment. (bea.gov)
  → This strongly supports Chamath’s claim of being “super bullish on the economy.”

Technology / markets in 2021

• The S&P 500 returned 28.71% in 2021, well above its long‑run average. (slickcharts.com)
• The tech‑heavy Nasdaq‑100 returned 26.63% in 2021, also a very strong year, following an even bigger gain in 2020. (slickcharts.com)
  → The U.S. equity market and especially large‑cap tech performed very strongly in 2021, consistent with being “super bullish on tech.”

U.S. political situation vs. 2020
Evidence that 2021 was not clearly an improvement:

• On January 6, 2021, a pro‑Trump mob attacked the U.S. Capitol in an effort to overturn the 2020 election, in what has been described as the most severe assault on the building since 1814 and part of a broader attempt to subvert the election result. (en.wikipedia.org)
• The attack led to a second impeachment of Donald Trump and deepened partisan conflict over election legitimacy and investigations into the riot, signaling ongoing or even heightened dysfunction. (en.wikipedia.org)

Evidence that 2021 could be seen as relatively more stable than 2020:

• 2020 featured the peak of nationwide racial‑justice protests after George Floyd’s murder (the largest protest movement in modern U.S. history), widespread COVID‑19 lockdown protests, and intense, sustained election‑related demonstrations beginning in November 2020. (en.wikipedia.org)
• Some forms of mass street unrest and COVID‑related protest activity diminished after early 2021, and by mid‑2021 the political environment lacked the same continuous, overlapping crises (pandemic onset, first impeachment, summer unrest, election‑week uncertainty) that defined much of 2020.

Why the verdict is ambiguous

• The economic and tech‑market parts of the prediction are clearly correct on the data: 2021 saw a strong, above‑trend macro rebound and very strong equity/tech returns. (bea.gov)
• The political part—that the U.S. political situation would improve relative to 2020—is a subjective judgment. One can reasonably argue either that:
  • January 6 and its aftermath show greater instability, or
  • the overall year 2021, after the transfer of power, had less continuous turmoil than 2020 despite the early‑year crisis.

Because the political component of the prediction depends on a normative comparison that reasonable observers can disagree on, while the economic/tech components were clearly borne out, the overall assessment of the combined prediction is ambiguous rather than clearly right or wrong.

Assessment of the prediction
Chamath’s core claim was that soon after the November 2020 election U.S. political discourse would become more civil, and that Joe Biden’s election would play a major role in that improvement by changing what behaviors are rewarded.

The empirical record for the 1–2 years after the 2020 election points the other way:

1. Public perception of civility and division worsened under Biden’s first year.

   • A February 2022 Georgetown University “Battleground Civility Poll” found that 43% of voters said politics was less civil since Biden took office, only 29% said more civil, and 27% said about the same. The poll summary notes that a majority of voters think politics has gotten less civil overall and explicitly says that, despite Biden’s rhetoric about civility, “most voters (60%) think politics has been less civil.” (politics.georgetown.edu)
   • A November 2021 Monmouth University poll found about half of Americans (49%) felt the country had become more divided since Biden took office, only 12% thought it had become more united, and 38% saw no change. While this was somewhat better than peak division under Trump, it still indicates a net move toward greater division, not a return to civility. (monmouth.edu)

2. Political threats and intimidation continued to surge after the 2020 election.

   • An analysis from the Carnegie Endowment documents that threats against members of Congress rose dramatically, from 902 in 2016 to about 9,600 by 2021—more than a tenfold increase, with 8,613 threats already in 2020 and another spike in 2021. (carnegieendowment.org)
   • U.S. Capitol Police data summarized by PolitiFact similarly shows threats against lawmakers increasing every year through 2021, reaching 9,625 cases that year—again about ten times the 2016 level. (politifact.com)
   • Research on political violence reports roughly 300 acts of political violence from the January 6 attack through the 2024 election, the largest surge since the 1970s, with violence during the 2024 cycle at its highest since that era. (en.wikipedia.org)
     These trends show an environment of escalating intimidation and violence, not one in which civility was “found” shortly after Biden’s election.

3. The immediate post‑election period was especially uncivil.

   • After Biden’s victory, there were widespread attempts to overturn the election, accompanied by menacing threats against election officials and their families in multiple states, sometimes including explicit death threats. (en.wikipedia.org)
   • This culminated in the January 6, 2021 attack on the U.S. Capitol, a highly visible breakdown of peaceful, civil politics that occurred in the exact “near term” window following the election.

Conclusion
Although Biden was elected (a part of Chamath’s expectation), the key part of the prediction—that his election would quickly usher in a more civil political discourse and substantially improve what behaviors are rewarded—does not match subsequent data and events. Surveys show most Americans perceived politics as less civil and more divided in the early Biden years, while threats and political violence rose sharply. That evidence makes the overall prediction wrong.

Joe Biden won the November 3, 2020 U.S. presidential election, defeating incumbent President Donald Trump. Major news organizations and official state certifications confirmed Biden as the winner with 306 electoral votes to Trump’s 232, and a popular vote margin of over 7 million votes.

Because Chamath’s prediction was that “Joe Biden will win the November 2020 U.S. presidential election”, and Biden did in fact win, this prediction is right.

Key sources documenting Biden’s victory include:

• Post‑election summaries of the 2020 U.S. presidential election, showing Joe Biden as the winner with 306 electoral votes and the popular vote lead.
• Official state certifications and Electoral College results finalized in December 2020.

(Any standard reference on the 2020 U.S. presidential election—such as major U.S. news outlets or the official National Archives record of Electoral College results—confirms Biden’s victory.)

How the world actually evolved (2020–2025)

Evidence the prediction did come true in an important sense (more dispersion):

• A major 2021 Rise of the Rest / PitchBook report found that seed and early‑stage VC going to Bay Area startups was on track to fall below 30% of the U.S. total for the first time in over a decade, while capital raised by VC funds outside the Bay Area/NYC/Boston rose ~700% over ten years to $21B, with more than 3,000 active investors now based outside those three hubs. (revolution.com)
• S&P Global’s 2024 analysis showed venture funding rounds in many inland states (Arkansas, Alabama, Iowa, Wyoming, etc.) more than doubled in value between 2019 and 2023, explicitly linking the shift to remote work and onshoring; meanwhile, California and New York saw declines in total VC dollars over that span (though from a much higher base). (spglobal.com)
• Multiple Sunbelt and interior metros—Austin, Miami, Denver, Nashville, Phoenix, Orlando, Albuquerque, the Raleigh–Durham Triangle—have grown tech employment and job postings far faster than traditional hubs, collectively adding large numbers of tech jobs and increasingly attracting big‑tech offices. (thebigjob.com)
• Brookings’ pandemic‑era work on tech geography finds that while the eight “superstar” metros still dominate, there has been modest but real diffusion of tech jobs, startups, and job postings to a wider set of smaller and lower‑cost metros since 2020, with superstar metros’ share of tech job postings falling from ~40% (2016) to about 31% by late 2021 and rising‑star/interior cities gaining share. (brookings.edu)
• Remote work and digital nomadism became structural: remote/hybrid work has stabilized rather than reverting to pre‑COVID norms, remote postings (about 8% of LinkedIn jobs) attract roughly 35% of applications, and U.S. digital nomads more than doubled from 2019 to over 18M in 2024—expanding tech‑adjacent opportunities in many locations where major employers have no office. (businessinsider.com)

These trends support Chamath’s intuition that COVID and remote work would create meaningful new tech and startup opportunity in many more U.S. cities, not just in Silicon Valley.

Evidence the prediction did not fully come true as framed (“away from” California/Silicon Valley, wealth creation):

• Venture capital and startup wealth have actually re‑concentrated in California—especially the Bay Area—since the initial 2020–2021 dispersion:
  • Crunchbase reported that by 2022, California startups were again taking ~51% of all U.S. VC dollars, up from ~47% in 2019. (news.crunchbase.com)
  • By 2023, companies in the San Francisco Bay Area alone attracted about 41% of all U.S. startup funding. (news.crunchbase.com)
  • Carta data for Q3 2023–Q2 2024 show California accounting for 51.7% of all VC raised on its platform; the next‑largest state (New York) had only ~11%. (carta.com)
  • Crunchbase figures summarized in early 2025 show Bay Area startups taking $90B of $178B in U.S. VC funding in 2024 (≈57% of the national total), driven largely by AI mega‑rounds (OpenAI, Anthropic, Databricks, xAI, etc.). (techcrunch.com)
  • A 2025 analysis notes that roughly 52% of all U.S. startup investment in early 2025 was going to California startups, up from 47% in 2023—i.e., California’s share of capital is rising, not falling. (consultancycircle.com)
• The 2025 Silicon Valley Index still finds the region attracting ~$69B in VC in 2024 and generating tens of thousands of patents, even as some firms add capacity faster in Austin and Seattle. Separate JLL data show the Bay Area capturing nearly half of global AI VC in 2024 and driving a new local office‑demand boom. (lemonde.fr)
• On employment, Brookings and Census data describe a persistent “winner‑take‑most” geography: the eight superstar metros (including San Francisco and San Jose) still hold around 38%+ of U.S. tech jobs and accounted for about half of tech job growth in the first pandemic year, slightly increasing their employment share even amid remote‑work disruption. The Census Bureau’s 2023 mapping of high‑tech employment continues to show the largest, densest clusters on the coasts (San Jose, San Francisco, Boston, NYC, DC). (brookings.edu)
• VC itself has become more concentrated in a handful of top Silicon Valley–centric firms: by 2024, over half of all capital raised by U.S. VCs went to just nine firms, and the total number of active VC firms had fallen ~25% since 2021—reinforcing capital‑allocation power in the core hubs rather than diffusing it. (ft.com)

This evidence runs against the stronger reading of the prediction that opportunity and wealth creation would move away from the primary concentration in California/Silicon Valley in a durable way. Instead, California—especially the Bay Area—now commands at least as large a share of high‑end venture funding and frontier tech wealth (AI) as before COVID, sometimes larger.

Why the verdict is ‘ambiguous’

• If you interpret the claim narrowly as “Silicon Valley will no longer be the dominant center of tech wealth and startup funding; that will shift significantly to the rest of the country”, the data do not support it: California/Bay Area’s share of U.S. VC and AI capital is currently higher than pre‑pandemic, and superstar metros remain structurally dominant in tech jobs. (news.crunchbase.com)
• If you interpret it more broadly as “COVID and frustration with California will catalyze a meaningful redistribution of tech and startup opportunities across many more U.S. regions, enabled by remote work”, the prediction looks substantially right: there has been large absolute growth of VC, startups, and tech employment in inland and Sunbelt metros; thousands of new non‑coastal VC firms; and a lasting expansion of remote/hybrid roles that lets talent live far from Silicon Valley. (revolution.com)

Because both dynamics are true at the same time—more places have real tech opportunity now, yet capital and wealth are still (and in some respects more) concentrated in California/Silicon Valley—the outcome can’t be cleanly labeled simply “right” or “wrong.” That’s why the most accurate overall judgment is ambiguous.

Evidence since 2020 supports the claim that many U.S. tech companies have reduced their concentration in San Francisco and expanded across a broader set of U.S. regions.

1. Companies reducing SF footprint and decentralizing

• A 2021 analysis by sf.citi (a San Francisco tech industry group) documented a significant migration of tech companies and workers out of San Francisco, finding that most surveyed tech companies had already downsized or planned to downsize Bay Area offices, and many expected a substantial share of their Bay Area workforce to remain permanently remote. The report explicitly described an ongoing tech exodus and a trend toward decentralization of the industry footprint away from San Francisco. (sfciti.org)
• A related sf.citi event report in March 2021 noted that the Bay Area’s dominance as the single main tech cluster was already weakening before COVID, and that the pandemic accelerated a shift toward more distributed or hub‑and‑spoke office footprints, with more tech offices in more U.S. cities (e.g., founders planning multiple hubs rather than a single SF HQ). (sfciti.org)

2. Headquarters moves and major expansions out of California/Bay Area

• Brookings (2022) highlighted that several major tech firms moved corporate headquarters from California to other metros during the pandemic period, including Palantir to Denver, Hewlett Packard Enterprise to the Houston area, Oracle and Tesla to Texas, and noted large new Google and Apple engineering offices in other states. (brookings.edu)
• A 2025 Business Insider report found that since 2020 more than 200 companies have relocated their headquarters to Texas, with over half coming from California. It specifically lists high‑profile tech and tech‑adjacent moves such as Tesla, Oracle, SpaceX, and Coinbase choosing Texas for headquarters or major bases, reflecting a sustained post‑2020 relocation trend away from California’s traditional hubs. (businessinsider.com)

3. Growth of new and expanding tech hubs across the U.S.

• Brookings’ geographic analysis of the tech sector concludes that while pre‑pandemic growth was highly concentrated in a few coastal superstars (including San Francisco and San Jose), the pandemic years began to distribute more tech activity into a wider set of places. Employment growth slowed in some top superstars and increased in numerous midsized and smaller markets, suggesting a partial shift from extreme concentration toward a more geographically dispersed pattern. (brookings.edu)
• A 2022 Fortune analysis of tech job data found that, although California and New York still led in total tech employment, the fastest tech job growth during the pandemic occurred in a wide range of other states (such as Tennessee, Idaho, Utah, North Carolina, Texas) and in cities like Houston, Orlando, and Charlotte, showing that tech employment growth was increasingly scattered across the country rather than confined to the Bay Area and a few coastal hubs. (fortune.com)
• A 2025 employment intelligence report described Austin, Miami, Denver, and Nashville as major emerging tech employment centers, collectively adding 127,000 tech jobs in 2024 and growing nearly 90 percent faster than traditional hubs. It notes major firms such as Tesla, Apple, Google, and Meta building significant engineering operations in these emerging markets, along with a sharp rise in venture capital investment in nontraditional hubs, emphasizing conscious geographic diversification by companies and investors. (thebigjob.com)

4. Overall assessment relative to the prediction

• Sacks’s prediction did not say San Francisco would cease to matter; it claimed that, because of SF’s policies and broader conditions after 2020, tech companies would increasingly be located all over the U.S. rather than being so heavily concentrated in the Bay Area.
• The post‑2020 evidence shows: (a) a documented reduction in SF’s share of offices and workers as many firms go remote‑first or downsize there, (b) multiple headline firms relocating headquarters or substantial operations from California to other states, and (c) rapid growth and deepening tech ecosystems in a broad range of U.S. metros (Austin, Miami, Denver, Nashville, Houston, smaller interior cities), all while SF remains important but less uniquely dominant.

Taken together, these trends match the direction and essence of the prediction: in the years after 2020, U.S. tech employment, investment, and corporate footprints have become more geographically dispersed across many regions of the country, with significant growth and relocation outside San Francisco and the Bay Area. Therefore, the prediction is best characterized as right.

What was predicted
Chamath said in early October 2020 that the odds of Joe Biden winning the November 2020 U.S. presidential election were about 65–35, approaching 70–30 in Biden’s favor ("Biden 6535, approaching 7030."). This is a probabilistic forecast assigning a substantially higher chance of victory to Biden than to Trump.

What actually happened
In the 2020 U.S. presidential election held on November 3, 2020, Joe Biden defeated Donald Trump, winning both the national popular vote and the decisive majority of Electoral College votes. Major news outlets and the official certification process confirmed Biden as the winner.

Assessment
Because the event to which Chamath assigned the higher probability (a Biden victory) did occur, this probabilistic prediction is directionally correct. He did not guarantee a Biden win, but he judged Biden as the clear favorite and that outcome materialized.

Therefore, the prediction is classified as right.

Available data on prediction and betting markets in early–mid October 2020 shows Joe Biden was indeed trading in roughly the 60–70% implied probability range that Sacks described. For example, a Forbes piece on October 7, 2020 reported prediction market PredictIt giving Biden a 65% chance of victory, and noted this was an increase from the previous week. (forbes.com) Another Forbes article on November 2, 2020 stated that PredictIt had Biden “near-70% highs earlier this month,” indicating Biden’s odds had recently been around the 70% level in October. (forbes.com) A Smarkets exchange update on October 21, 2020 reported Biden at a 63% chance at that time and noted his lead had recently reached 71% versus Trump’s 29%, again consistent with the 60–70% range Sacks referenced. (prnewswire.com)

Sacks’s claim, normalized as “As of early October 2020, Joe Biden has roughly a 70% probability of winning, in line with betting market odds,” therefore matches independent reporting on those markets at that time. In addition, Biden did go on to win the 2020 U.S. presidential election, securing 306 electoral votes to Trump’s 232 and being inaugurated as the 46th president on January 20, 2021. (cnbc.com) Taken together, his characterization of the odds was accurate relative to contemporaneous markets, and the higher-probability outcome he pointed to (a Biden win) is what actually occurred, so this prediction is best classified as right.

• Around the time of the podcast (immediately after the first debate and Trump’s COVID diagnosis), major betting and prediction markets generally implied something in the ballpark of a 30–40% chance for Trump. For example:

  • A Reuters report on October 1, 2020 noted that on Smarkets, Biden was around a 63% favorite after the first debate, implying Trump in roughly the high‑30s/low‑40s. (timesofindia.indiatimes.com)
  • A mid‑October Guardian summary of online bookmakers put Biden around 69% and Trump about 32%. (theguardian.com)
  • Crypto and exchange markets like Polymarket and FTX generally priced Trump in the low‑30s to high‑30s percent range in the weeks leading up to the election. (decrypt.co)

• Expert statistical models at the time disagreed strongly with those market odds:

  • FiveThirtyEight’s model, for instance, gave Trump only about a 12% chance by October 22, 2020. (fivethirtyeight.com)
  • The Economist’s model around late October put Trump as low as 4–14%. (decrypt.co)

• After the fact, several analyses argued that 2020 election prediction markets showed systematic irrationalities and arbitrage opportunities, particularly in the way they priced Trump even after it was clear Biden had won key states. (asteriskmag.com) However, these critiques focus on inefficiencies and bias; they do not pin down a consensus, ex‑post “true” pre‑election win probability for Trump.

• Because:

  1. The “true probability” of a one‑off event (Trump winning in 2020) is inherently unobservable;
  2. Credible models at the time spanned a wide range (roughly 5–40%) for Trump’s chances; and
  3. Post‑election research has shown that both polls and markets had notable flaws without definitively establishing that Trump’s real odds must have been above or below the ~30% level,

  we cannot rigorously determine whether Sacks was correct that Trump’s true winning probability was “somewhat higher” than the ~30% implied by markets.

Conclusion: The claim is about an unobservable quantity (true probability) in a context where contemporary and retrospective analyses diverge. Even with the benefit of hindsight, there is no objective way to say that the real probability was definitively above or below 30%. Therefore the prediction’s accuracy is ambiguous rather than clearly right or wrong.