Visualization section · Dataset v5
See the pattern, not just the headline.
Annual trends, raw counts, verdict composition, subject categories, and claim types show what changed—and how much the sample size shapes what appears to change.
Direction over time
More honest, more dishonest, or somewhere between?
The annual lines normalize each year to percentages. Point details expose the underlying number of tracked and scored claims.
The 2021–2025 window can be compared with 2016–2020 in this corpus, but annual samples range from 1 to 26 records and the topics selected change over time. This is a view of the tracked dataset, not a random sample of everything Musk said in each year. A False verdict also does not by itself establish a deliberate lie. 2 genuinely undated records are excluded from the yearly charts rather than assigned an invented date.
Hover, tap, or focus a year for its exact score, False count, and sample size.
View the annual data table
| Year | Tracked | Scored | Trust Score | False | False share |
|---|---|---|---|---|---|
| 2002 | 1 | 1 | 50.0% | 0 | 0.0% |
| 2006 | 3 | 3 | 100.0% | 0 | 0.0% |
| 2008 | 1 | 1 | 0.0% | 1 | 100.0% |
| 2011 | 2 | 1 | 0.0% | 1 | 50.0% |
| 2012 | 1 | 0 | Not scored | 0 | 0.0% |
| 2013 | 1 | 1 | 0.0% | 1 | 100.0% |
| 2015 | 1 | 1 | 0.0% | 1 | 100.0% |
| 2016 | 8 | 7 | 42.9% | 4 | 50.0% |
| 2017 | 11 | 11 | 18.2% | 9 | 81.8% |
| 2018 | 15 | 14 | 7.1% | 11 | 73.3% |
| 2019 | 10 | 10 | 0.0% | 10 | 100.0% |
| 2020 | 4 | 3 | 33.3% | 2 | 50.0% |
| 2021 | 5 | 5 | 35.0% | 3 | 60.0% |
| 2022 | 20 | 17 | 26.5% | 12 | 60.0% |
| 2023 | 13 | 12 | 14.6% | 9 | 69.2% |
| 2024 | 26 | 21 | 25.0% | 10 | 38.5% |
| 2025 | 25 | 23 | 20.7% | 13 | 52.0% |
| 2026 | 6 | 5 | 15.0% | 4 | 66.7% |
Raw annual volume
False records per year, with the denominator visible.
Bar length shows how many records the corpus tracks in that year. Red shows the zero-point subset classified False.
Three more ways to read the corpus
Verdict mix, subject category, and claim type.
The verdict mix shows the whole corpus. Subject and statement-form views compare independent facets without assuming every claim belongs to one of Musk's organizations.
Part to whole
Verdict mix.
- True
- 19
- Mostly True
- 3
- Misleading
- 14
- Unsupported
- 11
- False
- 91
- Pending
- 12
- Unresolved
- 5
Shared-scale comparison
Score by subject category.
Statement form
Score by claim type.
Claim type describes how the statement was framed, independently of what it was about or which entity it concerned.
Read charts with the rules beside them