Suggested Google deliberately favored Kamala Harris by displaying voting information for searches involving her name but not Donald Trump’s.
Evidence chain
From statement to verdict.
- 01
Original statement
Suggested Google deliberately favored Kamala Harris by displaying voting information for searches involving her name but not Donald Trump’s.
Nov 5, 2024 · Factual Assertion
- 02
Measurable target
Present-tense or retrospective claim
Not applicable
- 03
Observed outcome
The result was triggered because Harris and Vance are also county names, not because Google favored a candidate. Musk later acknowledged Google’s explanation.
Evaluated Jul 26, 2026
- 04
Verdict and score
Misleading
50 of 100 points under the v5 rubric. Some facts are accurate, but omitted context or framing materially changes the reasonable takeaway.
Record details
How this row was evaluated.
- Subject category
- Media & Society
- Topic category
- Media & Information
- Organization or context
- Public discourse
- Relationship to organization
- Public sphere
- Assertion mode
- Direct social-media posts
- Deadline result
- Not applicable
- Eventual outcome
- Not applicable
- Factual accuracy
- Misleading
- Canonical classification
- Misleading
- Score points
- 50 / 100
- Included in headline score
- Yes
- Research confidence
- Medium-Low
- Credible-source contestation
- Yes
- Contestation resolution
- The observed behavior was real, but the evidence supported a search-trigger quirk rather than political favoritism.
- Correction status
- Acknowledged the explanation
- Correction date
- Nov 5, 2024
- Deleted after challenge
- Yes
- Repeated after correction
- No
- Evidence-documented repetition count
- 3
- Sensitive-topic metadata
- Media · Information integrity
Topic tags are descriptive metadata and do not independently determine the verdict or score.
Intent stays separate
Was intentional deception established?
No
- Detailed assessment
- Not established
- Intent evidence level
- 0 / 3
The cited evidence establishes material missing context, framing, delay, or partial delivery, but it does not directly establish Musk’s contemporaneous knowledge or purpose. Misleading outcome and intentional deception are evaluated separately.
“No” does not mean the statement was true. It means the cited evidence did not meet the higher standard needed to establish the speaker's state of mind.
Why this distinction mattersEvidence strength
A disclosed measure of evidence quality—not a probability.
Component scores measure source quality, corroboration, and directness. Verdict confidence begins with that evidence strength and applies the deductions documented in the methodology.
Quality and directness of evidence that Musk made the statement.
Quality of evidence establishing the outcome.
Independent corroboration across sources and domains.
How directly the evidence resolves the proposition.
- Evidence sources
- 1
- Independent source domains
- 1
- Claim audit
- Complete
- Source audit
- Pass
Material contestation is documented and resolved or left Unresolved in the row.
Citations
Read the underlying sources.
Field-level evaluation audit
Every calculated output carries its own evidence basis.
Open any of the 12 audited outputs to see its calculated value, written basis, rule, and cited sources.
01Deadline resultNot applicable
Evidence basis
Compared the original target/timeframe (“Present-tense or retrospective claim”) with the documented outcome. Result: Not applicable. Evidence: Google’s ‘Where to Vote’ Search Result Reflects Quirk of Candidate Surname, Not Bias. URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
Calculation rule
Compare original target with documented timing.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
02Eventual outcomeNot applicable
Evidence basis
Classified the later real-world result independently from the headline verdict. Result: Not applicable. Documented outcome: The result was triggered because Harris and Vance are also county names, not because Google favored a candidate. Musk later acknowledged Google’s explanation. Evidence: Google’s ‘Where to Vote’ Search Result Reflects Quirk of Candidate Surname, Not Bias. URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
Calculation rule
Classify the later real-world outcome separately from the headline verdict.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
03Factual accuracyMisleading
Evidence basis
Compared the normalized proposition with the cited records. Result: Misleading. Documented outcome: The result was triggered because Harris and Vance are also county names, not because Google favored a candidate. Musk later acknowledged Google’s explanation. Evidence: Google’s ‘Where to Vote’ Search Result Reflects Quirk of Candidate Surname, Not Bias. URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
Calculation rule
Compare the normalized factual proposition with cited evidence.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
04Canonical verdictMisleading
Evidence basis
Factual-claim rule: apply the published True, Mostly True, Misleading, Unsupported, False, or Unresolved definition. Applied result: Misleading (Misleading). Evidence finding: The result was triggered because Harris and Vance are also county names, not because Google favored a candidate. Musk later acknowledged Google’s explanation. Evidence: Google’s ‘Where to Vote’ Search Result Reflects Quirk of Candidate Surname, Not Bias. URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
Calculation rule
Apply strict promise rule or published factual-claim classification.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
05Score points50
Evidence basis
Score assigned mechanically from verdict and claim type under strict-promise-v1.0: Misleading = 50 points.
Calculation rule
Map verdict mechanically to points under the applicable rule.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
06Inclusion in the Trust ScoreYes
Evidence basis
Included because the claim has a resolved evidence-based verdict.
Calculation rule
Include only resolved evidence-based verdicts.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
07Research confidenceMedium-Low
Evidence basis
Evidence strength 67/100 = statement 24/30 + outcome 31/40 + corroboration 0/15 + directness 12/15. Published deductions yield verdict confidence 56/100 (Medium-Low).
Calculation rule
Calculate from evidence components and published deductions.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
08Credible-source contestationYes
Evidence basis
Credible-source contestation: Yes. Resolution: The observed behavior was real, but the evidence supported a search-trigger quirk rather than political favoritism. Evidence URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
Calculation rule
Record material credible-source disagreement separately from verdict.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
09Correction statusAcknowledged the explanation
Evidence basis
Correction status is bounded to the cited record: Acknowledged the explanation. Evidence URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
Calculation rule
Bound correction finding to cited evidence.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
10Repetition after correctionNo
Evidence basis
Repeated after correction: No. Minimum documented repetition count: 3. Evidence URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
Calculation rule
Bound repetition finding to cited evidence.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
11Intentional-deception answerNo
Evidence basis
The cited evidence establishes material missing context, framing, delay, or partial delivery, but it does not directly establish Musk’s contemporaneous knowledge or purpose. Misleading outcome and intentional deception are evaluated separately.
Calculation rule
Require direct admission, adjudicated intent finding, or comparably conclusive state-of-mind evidence.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
12Strict promise resultNot applicable
Evidence basis
Factual-claim rule: apply the published True, Mostly True, Misleading, Unsupported, False, or Unresolved definition. Applied result: Misleading (Misleading). Evidence finding: The result was triggered because Harris and Vance are also county names, not because Google favored a candidate. Musk later acknowledged Google’s explanation. Evidence: Google’s ‘Where to Vote’ Search Result Reflects Quirk of Candidate Surname, Not Bias. URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
Calculation rule
Pass only if every material promise or forecast term was met; otherwise Fail, Pending, Unresolved, or Not applicable.
- Evidence strength
- 67 / 100
- Verdict confidence
- 56 / 100 · Medium-Low
- Audit status
- Complete
Audit trail
Selection and scoring notes.
- v5 classification rationale
- Factual-claim rule: apply the published True, Mostly True, Misleading, Unsupported, False, or Unresolved definition. Applied result: Misleading (Misleading). Evidence finding: The result was triggered because Harris and Vance are also county names, not because Google favored a candidate. Musk later acknowledged Google’s explanation. Evidence: Google’s ‘Where to Vote’ Search Result Reflects Quirk of Candidate Surname, Not Bias. URLs: https://www.factcheck.org/2024/11/googles-where-to-vote-search-result-reflects-quirk-of-candidate-surname-not-bias/
- v4 → v5 source-package migration audit
- Audited, no headline change. Evidence and derived statistics re-audited.
- Selection basis
- Material v3 expansion claim selected under published inclusion rules
- Source quality
- Original statement or strong reporting paired with a major wire, official record, court filing, or established fact-checker V4 evidence metrics: statement 24/30; outcome 31/40; corroboration 0/15; directness 12/15.
- Methodology note
- V5 uses strict-promise-v1.0. Later delivery does not earn partial credit for a missed dated promise. Scored under the v3 Misleading category at 50 points. This rating measures demonstrated reliability and does not by itself establish deceptive intent.