Jason Calacanis
C60/ 100
3 claims reviewed
Gets the revenue and debt arithmetic right; recommends cancer screening too broadly.
Evidence balance 5/10Earnings, debt and cancer treatment
The 30-second take
NVIDIA’s revenue figures and the explanation of bond yields hold up. The screening advice exceeds the clinical evidence, and the Social Security discussion overlooks continuing tax receipts.
Episode: Nvidia's Historic Quarter, SaaS Comeback, Bessent vs Druck, America's Debt Crisis, Cancer Vaccine
Speaker assessments
David Sacks
B75/ 100
4 claims reviewed
Clear institutional and business reasoning.
Evidence balance 8/10David Friedberg
D50/ 100
4 claims reviewed
Explains the cancer therapy but overstates treatment equivalence and Social Security’s shortfall.
Evidence balance 3/10Chamath Palihapitiya
C65/ 100
3 claims reviewed
Sound mechanisms, with one broad software claim.
Evidence balance 6/10Scores, sample sizes & reasoning
| Speaker | Overall | Accuracy 30% | Coherence 20% | Evidence support 20% | Calibration 15% | Evidence balance 15% | Claims |
|---|---|---|---|---|---|---|---|
| Jason Calacanis | 60 / C62 before rounding | 7 | 6 | 7 | 5 | 5 | 3 |
| David Sacks | 75 / B76.5 before rounding | 8 | 8 | 7 | 7 | 8 | 4 |
| David Friedberg | 50 / D50.5 before rounding | 6 | 6 | 5 | 4 | 3 | 4 |
| Chamath Palihapitiya | 65 / C65 before rounding | 7 | 7 | 6 | 6 | 6 | 3 |
Jason Calacanis
Credit. Checks NVIDIA revenue and the debt run-rate calculation.
Deduction. Turns promising cancer developments into a broad testing recommendation.
Evidence balance 5/10
Correct financial comparisons coexist with a one-sided screening recommendation.
Credit. Uses the reported period and year-over-year revenue comparison. NVIDIA revenue
Deduction. NCI identifies unproven mortality benefit and risks including false positives and overdiagnosis. Those tradeoffs can change the case for broad early-and-often multi-cancer testing. Cancer testing
David Sacks
Credit. Distinguishes software business models and identifies a real veto constraint.
Deduction. Does not measure the relative strength of the proposed incentive mechanisms.
Evidence balance 8/10
Distinguishes business models and institutional constraints without a concrete cherry-pick identified in this sample.
Credit. Rejects a single outcome for all software companies. Case-by-case software analysis
Credit. Separates the president’s actual legal powers from a preferred policy outcome. Presidential line-item veto
No concrete selective presentation was found in the reviewed sample.
David Friedberg
Credit. Explains a supported therapeutic mechanism.
Deduction. Overstates treatment equivalence and conflates trust-fund depletion with zero income.
Evidence balance 3/10
Omits continuing Social Security income and the limits of the cancer-treatment evidence.
Credit. Explains the particular tumor-targeting mechanism behind the trial. Personalized cancer vaccines
Deduction. The trustees project continuing OASI income sufficient for 78% of scheduled benefits after reserve depletion. Leaving out ongoing receipts turns a large financing shortfall into an apparent end to payments. Social Security after depletion
Deduction. The clinical evidence concerns a specific therapy, combination and melanoma population. It does not establish comparable benefits for different clinic peptide products. Clinic peptide alternatives
Chamath Palihapitiya
Credit. Explains enterprise context and the inverse price-yield relationship.
Deduction. Draws too clean a boundary between vertical workflows and systems of record.
Evidence balance 6/10
The enterprise-data argument is useful, but the vertical-software contrast excludes a clear counterexample.
Credit. Identifies the concrete role of enterprise context in agent workflows. Enterprise context
Deduction. Veeva’s industry-specific CRM holds customer records as well as workflows. That counterexample breaks the clean horizontal-records versus vertical-workflows distinction. Vertical software
| Speaker | Facts | Opinions / causal arguments | Forecasts |
|---|---|---|---|
| Jason Calacanis | 2 | 1 | 0 |
| David Sacks | 1 | 3 | 0 |
| David Friedberg | 2 | 1 | 1 |
| Chamath Palihapitiya | 1 | 2 | 0 |
Claim reviews
14 claims · 4 speakers
01NVIDIA revenueSupported
NVIDIA reported the stated revenue and growth rate.
The claim paraphrased
NVIDIA reported $96.2 billion quarterly revenue, up 106% year over year.
Evidence checked
NVIDIA’s Q2 FY2027 release reports those revenue figures.
Why this judgment
The stated revenue and growth rate match NVIDIA’s release for the quarter. Profit, valuation and future growth require separate checks.
Confidence in assessment~95%
Both figures match the company’s published quarterly results.
What would change this judgment?
A corrected filing or a mismatch in the period being compared.
02Cancer testingToo broad
The recommendation goes beyond the evidence for multi-cancer screening.
The claim paraphrased
People should get tested early and often for cancer, in a discussion of multi-cancer blood tests.
Evidence checked
NCI says it remains unknown whether multi-cancer detection screening reduces overall cancer mortality. False positives, unnecessary procedures and overdiagnosis matter.
Why this judgment
A screening recommendation needs evidence that testing improves outcomes in the population offered the test. Treatment advances alone do not establish that benefit, especially when false positives can trigger unnecessary procedures.
Confidence in assessment~85%
NCI explicitly identifies the unproven mortality benefit and potential harms of these tests.
Evidence-balance issue
NCI identifies unproven mortality benefit and risks including false positives and overdiagnosis. Those tradeoffs can change the case for broad early-and-often multi-cancer testing.
What would change this judgment?
Randomized mortality evidence and a recommendation tied to a defined population and test.
03Personalized cancer vaccinesSupported with scope limits
The melanoma trial supports the proposed immune response.
The claim paraphrased
Tumor-specific mRNA therapy can train immunity against an individual patient’s cancer.
Evidence checked
Merck and Moderna announced positive recurrence-free and distant-metastasis-free survival endpoints in a phase 3 melanoma trial using an individualized therapy with pembrolizumab after surgery.
Why this judgment
The announced melanoma results support this mechanism in a defined treatment setting. Extending the finding to other cancers requires further trials. Full results would allow scrutiny beyond the sponsor’s announcement.
Confidence in assessment~95%
The sponsor’s phase 3 announcement supports the described mechanism; broader efficacy and safety require the full results.
What would change this judgment?
Full peer-reviewed results, absolute effects and safety data could refine the assessment.
04Fragmented budget incentivesCoherent explanation
Legislators have incentives to protect programs whose costs are shared nationally.
The claim paraphrased
Many legislators protecting their programs make federal spending restraint difficult.
Evidence in the episode
Sacks describes many elected participants, each with reasons to defend particular programs. He does not quantify how much this explains the deficit.
Why this judgment
Concentrated program benefits and broadly shared financing costs can produce a collective-action problem. The explanation leaves out voters’ preferences, revenue choices and party bargaining, so it is one mechanism rather than a complete account.
Confidence in assessment~85%
High confidence in the mechanism; its contribution relative to other causes is unmeasured.
What would change this judgment?
Evidence comparing budget outcomes under different institutions and incentives.
05Social Security after depletionMisleading
Payroll taxes would still fund most scheduled benefits after reserves run out.
The claim paraphrased
Social Security will have no money to pay benefits around 2030–2032.
Evidence checked
The 2026 trustees project OASI reserve depletion in 2032, with ongoing income covering 78% of scheduled OASI benefits. The DI fund is projected to remain solvent through the report’s horizon.
Why this judgment
The financing problem is serious, but exhaustion of accumulated reserves does not erase payroll-tax receipts. Treating depletion as an end to all payments overstates the projected outcome and conflates different trust funds.
Confidence in assessment~95%
Very high confidence in the distinction between reserves and continuing program income.
Evidence-balance issue
The trustees project continuing OASI income sufficient for 78% of scheduled benefits after reserve depletion. Leaving out ongoing receipts turns a large financing shortfall into an apparent end to payments.
What would change this judgment?
Legislation or updated actuarial projections changing the financing path.
06Bond prices and yieldsSupported mechanism
A higher bond price lowers its yield; a buyback’s total fiscal effect needs a separate calculation.
The claim paraphrased
Buying bonds raises their price and lowers their implied yield.
Evidence in the episode
For a fixed stream of payments, a higher purchase price means a lower yield. This follows from discounting those payments, holding their amount and timing constant.
Why this judgment
Purchases can influence prices, but the durable effect on government financing depends on scale, maturity, market expectations and how purchases are funded. The identity does not establish that a particular buyback program lowers total borrowing costs.
Confidence in assessment~95%
Very high confidence in the bond identity; the program’s net effect is not estimated.
What would change this judgment?
A counterfactual estimate of yields and financing costs with and without the program.
07Enterprise contextPlausible mechanism
AI agents still need reliable business records.
The claim paraphrased
Systems of record retain value because AI agents need trusted business context.
Evidence checked
Products such as Veeva’s CRM organize customer records and workflows.
Why this judgment
Customer records give agents business context they cannot obtain from a general model alone. The software owner’s profits still depend on access terms, pricing and competitors.
Confidence in assessment~85%
Veeva’s product documents the business data involved. Its future pricing power is not established by that documentation.
What would change this judgment?
Evidence of durable retention and pricing power as agent adoption grows.
08Vertical softwareOvergeneralized
Veeva shows that vertical software can hold authoritative records.
The claim paraphrased
Vertical SaaS mostly supplies workflows rather than systems of record.
Evidence checked
Veeva offers industry-specific CRM and unified customer data for life sciences.
Why this judgment
Veeva is industry-specific software that holds customer records. That contradicts the proposed division between vertical workflows and horizontal record systems. The broader claim needs a defined set of companies.
Confidence in assessment~85%
Veeva’s documented CRM functions supply a direct counterexample to the categorical distinction.
Evidence-balance issue
Veeva’s industry-specific CRM holds customer records as well as workflows. That counterexample breaks the clean horizontal-records versus vertical-workflows distinction.
What would change this judgment?
A defined company universe and evidence showing which products hold authoritative records.
09Switching costsPlausible mechanism
Migration and compliance costs can slow a switch to new software.
The claim paraphrased
Core enterprise records and compliance needs can protect software incumbents.
Evidence in the episode
Migrating a critical record system carries operational costs, and regulated workflows add constraints.
Why this judgment
That is a coherent reason disruption may be slower than a software-generation demo suggests. The episode supplies no estimate of how much protection these costs provide.
Confidence in assessment~85%
Migration and compliance create recognizable costs, but the episode provides no customer data measuring their effect.
What would change this judgment?
Customer migration, retention and pricing evidence across comparable businesses.
10Case-by-case software analysisWell calibrated
AI’s effect depends on the software company’s product and customers.
The claim paraphrased
AI’s effect on software companies depends on their particular products and moats.
Evidence in the episode
Sacks distinguishes companies by their data, distribution and replaceable features.
Why this judgment
A product-level test can distinguish durable data or distribution advantages from replaceable features. The framework remains qualitative until applied to measurable revenue and retention outcomes.
Confidence in assessment~85%
Sacks identifies relevant company differences without claiming a measured forecast for the sector.
What would change this judgment?
A predictive company-level model and subsequent results could test the framework.
11Clinic peptide alternativesUnestablished
The clinic products need their own clinical evidence.
The claim paraphrased
Cheaper bespoke peptide treatments offered by clinics can deliver similar cancer-treatment benefits.
Evidence checked
The cited clinical evidence concerns a particular manufactured therapy, combination and trial population.
Why this judgment
The melanoma trial does not test the clinic products Friedberg describes. Those products need their own comparative outcomes; a lower price and similar proposed mechanism are insufficient.
Confidence in assessment~85%
The cited trial tests a specific therapy and combination, with no comparative evidence for the clinic alternatives.
Evidence-balance issue
The clinical evidence concerns a specific therapy, combination and melanoma population. It does not establish comparable benefits for different clinic peptide products.
What would change this judgment?
Controlled comparative evidence for the actual product and treatment protocol.
12Debt accumulation arithmeticArithmetic supported
The extrapolation is approximately right; the premise must persist.
The claim paraphrased
Adding $1 trillion every five months implies roughly $10 trillion over four years.
Evidence in the episode
Using the episode’s stated pace: $1T × 12/5 is $2.4T per year and $9.6T across four years. The rounded $10T is consistent with that arithmetic.
Why this judgment
The calculation holds only if debt keeps growing at the stated pace. Receipts, spending, interest rates and growth can all change over four years.
Confidence in assessment~95%
Very high confidence in the arithmetic; the underlying pace is treated as an assumption.
What would change this judgment?
A dated fiscal projection establishing the pace and its sensitivity to policy and growth.
13Presidential line-item vetoSupported
The president cannot cancel individual items in an enacted spending law.
The claim paraphrased
The president lacks a general line-item veto over enacted federal spending.
Evidence checked
In Clinton v. City of New York, the Supreme Court invalidated the statutory cancellation authority in the Line Item Veto Act.
Why this judgment
The president cannot simply rewrite an enacted spending law one item at a time. That constraint does not remove presidential influence through proposals, negotiations, whole-bill vetoes and legally authorized implementation choices.
Confidence in assessment~95%
Very high confidence in the cited holding and the narrow legal claim.
What would change this judgment?
A later controlling decision or constitutional change granting the asserted cancellation power.
14Interest-rate sensitivityNeeds scope
The interest increase would depend on how quickly existing debt is refinanced.
The claim paraphrased
A one-point rise in borrowing costs adds about 1.25% of GDP to annual interest.
Evidence in the episode
On the episode’s $40T debt assumption, 1% equals $400B. That is 1.25% of GDP only with a $32T GDP denominator. Existing fixed-rate debt does not all reprice immediately.
Why this judgment
The arithmetic can describe an eventual gross-debt sensitivity. Near-term federal interest costs depend on maturities, new borrowing, securities held by government accounts and the relevant rate path. Mixing these concepts exaggerates immediate exposure.
Confidence in assessment~95%
Very high confidence in the arithmetic and repricing distinction; no independent validation of the input estimates is implied.
What would change this judgment?
A maturity-weighted calculation separating gross debt from marketable public debt and specifying the GDP baseline.
No claims match these filters.
Source desk9 references
- All-In / LibsynOfficial episode
Original episode, published 2026-08-28.
- All-In / YouTubeWatch the episode
Original recording; timestamps mark discussion chapters.
- ArcmiraAutomated transcript
Automated transcript. Attribution is provisional; links mark discussion chapters.
Company-reported revenue, margins and cash flow for the quarter.
- National Cancer InstituteNational Cancer Institute: multi-cancer detection tests
Current guidance on multi-cancer tests and unresolved screening benefits; retrospective context.
- Merck / ModernaMerck / Moderna: phase 3 melanoma trial announcement
Trial sponsor’s report of the phase 3 melanoma results.
Current product documentation for industry-specific CRM; retrospective context.
- U.S. Supreme Court / Cornell LIIClinton v. City of New York
1998 decision invalidating the Line Item Veto Act.
- Social Security Administration2026 Social Security trustees report
OASI reserves projected to be depleted in 2032; continuing income covers 78% of scheduled OASI benefits.
How the scores work
Five dimensions, one weighted score
Each dimension is scored from 0–10. The weighted total is multiplied by ten and rounded to the nearest five, with halves rounded up.
Grades use that rounded score: A ≥80, B ≥70, C ≥60, D ≥50, F <50.
Accuracy tests factual correctness; coherence tests whether the conclusion follows; evidence support tests the strength of the sources. Calibration asks whether the speaker’s certainty matches the evidence.
Accuracy30%
Coherence20%
Evidence support20%
Calibration15%
Evidence balance15%
Evidence balance
Balance assesses cherry-picking, omitted counterevidence, inconsistent standards and distorted alternatives. Higher scores mean fairer treatment of evidence. Each deduction names the affected claim, the missing evidence and why it changes the conclusion.
Political disagreement and presumed intent do not count. A false claim or weak source alone does not establish cherry-picking.
- 9–10Actively tests strong contrary evidence and represents it fairly.
- 7–8Generally fair selection and relevant qualifications; no material distortion identified.
- 5–6Mixed: fair treatment in some claims, material omissions in others.
- 3–4Materially selective samples, comparisons or treatment of contrary evidence.
- 0–2Repeated, severe distortion or dismissal of directly relevant counterevidence.
What these scores can tell you
Scores assess selected arguments. They do not measure a speaker’s overall reliability or percentage of correct statements. Confidence ratings are subjective; no statistical intervals are calculated. Automated-transcript attribution remains provisional. Small differences are not statistically meaningful, and no independent second rater has checked the scores.
New reviews select up to 20 significant claims, without speaker quotas. The sample is editorial, not random or exhaustive. All-In #286–292 retain their original 14-claim samples. An absent or unassessable speaker has no score and leaves a gap in the chart.
Primary sources support the assessments. Current product and methodology pages are labeled when used retrospectively. Unresolved forecasts are assessed for their support. All-In video links use chapter starts; Pivot links use transcript timestamps, which can differ from ad-supported audio.
This review: 14 selected claims. Published Aug 28, 2026; reviewed Oct 5, 2026.
Rubric v2.0. Earlier scores and correction records remain in the review data.