We Investigated Uber Again. It’s Worse Than Last Time.
Credibility score: 46/100 — Mixed Credibility. Several questionable claims detected. Watch with healthy skepticism.
BSmeter analyzed "We Investigated Uber Again. It’s Worse Than Last Time." and rated it 46/100 for credibility (a BS score of 54/100 — mixed credibility), on 2026-09-10. Its weakest claim — "Uber publicly claims a take rate under 20% while the actual rate is 50% — setting up a direct contradiction." — scored 20/100 and was flagged as false equivalence. 30 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Claims analyzed
Opens with 'everything expensive' frame then narrows to Uber — sets up inflation as cover story — Missing Context (45/100)
Puts Uber fares in the 'everything is expensive' bucket — makes price hikes look inevitable rather than chosen.
Sets up 2024 and 2026 experiments as parallel — classic before/after framing — Missing Context (45/100)
Pairs two separate tests (drivers vs riders) as if they prove the same point — omits that they're different sides of the transaction
Switches from tested sample sizes to 'millions' without defining the source — volume game — Volume Game (35/100)
Jumps from controlled experiments to 'millions' of pooled receipts — the scale sounds damning but the sourcing is never named
Frames variable charges as someone 'getting rich' — emotional button without naming who benefits — Emotional Button (40/100)
Implies a hidden profiteer but never identifies who — steers outrage toward a vague 'somebody' instead of showing the actual accounting
Treats pay variation as inherently suspicious — missing context on how variable pricing actually works — Missing Context (50/100)
Presents $51–$90 swings as proof of randomness — skips the documented factors (time of day, demand, destination, wait time) that normally explain driver pay differences
Frames insurance as the hidden villain — suspenseful repetition — Emotional Button (30/100)
Repeats 'commercial auto insurance' like it's a scandal — priming viewers before any data
Lists worst-case scenarios to justify the cost — emotional framing — Emotional Button (35/100)
Stacks 'crashes, injuries, lawsuits' to make insurance feel ominous instead of routine
Contrasts 45% LA vs 20% national — cherry-picked extreme without context — Cherry-Picked (40/100)
Highlights the 45% LA figure as shocking — doesn't explain if it's typical or an outlier
Driver lost $20k to insurance — frames it as someone else's profit — Missing Context (45/100)
Calls the $20k 'someone getting rich' — no evidence who profits or if it's even excess
Claiming insurance rates are inconsistent — setting up a 'problem' for Uber. — No Frame (75/100)
They're presenting a specific observation from their data analysis. Seems straightforward.
Stating a principle of insurance — framing Uber's practice as abnormal. — No Frame (75/100)
This is a general principle of how insurance is supposed to work. It's a baseline, not a claim about Uber directly.
Asserting only two factors impact Uber's insurance fee — directly contradicting Uber's stated policy. — Confidence Mismatch (45/100)
They're making a very strong, definitive claim about what impacts the fee based on their analysis, but Uber disputes it.
Recalling Uber's 2014 promise of 80% driver pay — setting up a historical comparison. — No Frame (75/100)
This is a historical claim about a specific public statement by Uber in 2014. It's verifiable.
Attributing a 'slash' in driver pay to the new CEO's need for margins — using strong, loaded language. — Loaded Language (45/100)
Using 'slashed' and 'almost overnight' paints a dramatic picture, implying sudden, drastic, and negative intent.
Uber's take rate jumped from 15% to over 50% — presented as a dramatic, sudden increase. — Loaded Language (45/100)
Describing the jump as 'almost overnight' and 'exceeding 50%' uses strong language to amplify the perceived change. — It's a classic emotional button.
Uber publicly claims a take rate under 20% while the actual rate is 50% — setting up a direct contradiction. — False Equivalence (20/100)
They're presenting two numbers (50% vs. 20%) as directly comparable without explaining the different calculation methods. — It's a setup for a 'gotcha' moment.
Uber bundles 'commercial insurance and operating expenses' into a single line item — highlighting a lack of transparency. — Missing Context (45/100)
Pointing out the single line item implies Uber is hiding something, but doesn't explain *why* they might bundle these costs. — It's a subtle dig at transparency.
Aleka Insurance's board is all Uber execs and 95% of premiums stay with Uber — reinforcing the 'department of Uber' framing. — Loaded Language (45/100)
Emphasizing 'entirely current or former Uber executives' and 'stay inside Uber' reinforces the idea that Aleka is just an internal department. — It's a classic 'us vs. them' setup.
Fee tracks price, not risk — loaded framing of intent — Loaded Language (35/100)
Says 'seemingly track price and not risk' — assumes the fee should track risk, not revenue
Frames captive insurance as suspicious by asking 'why does it track price not risk' — Missing Context (45/100)
Captive insurers are legal but they're setting up the question as if tracking price is automatically shady — omits that captives often price based on their single client's loss history.
Suggests profit and loss statement would answer the fee question — Missing Context (45/100)
Skips that captive insurers aren't required to publish P&L like public insurers
Insurance fee tracks price, not risk — loaded question — Loaded Language (45/100)
Frames the question as suspicious before Uber's side is presented — rhetorical setup.
Frames fee as suspicious by pairing 'set by company collecting it' — loaded language — Loaded Language (45/100)
Sets up the fee as inherently shady before showing any data — the 'company collecting it' line does the heavy lifting.
Poses loaded question — assumes fee tracks price, not risk — Loaded Language (45/100)
Asks 'why' something happens before proving it happens — classic loaded setup
Insurance fee tracks price not risk — Missing Context — Missing Context (45/100)
Asks why the fee tracks price — but never says what the actual risk model should look like.
Question frames fee as tracking price, not risk — loaded language — Loaded Language (45/100)
Sets up the 'price not risk' conclusion before showing the analysis
Questions if fee tracks price over risk — frames as suspicious — Loaded Language (45/100)
Word 'seemingly' plants doubt before evidence — classic loaded framing
Questions insurance fee tracks price not risk — rhetorical setup — No Frame (75/100)
Clean framing: poses a question that invites investigation without assuming guilt.
Frames Uber's fee as suspicious by implying it should track risk, not price — loaded language — Loaded Language (45/100)
Sets up 'price not risk' as the only suspicious explanation — ignores that captive insurers often price on volume, not actuarial risk
Aleka doesn't publish data because Uber owns it — captive insurer explanation — No Frame (75/100)
Straightforward context — captive insurers legally exempt from public filings.
See the full analysis with sources and timestamps →