Gamebot / Survivor analytics
Who votes together?
Follow the voting coalitions of Pearl Islands, the seventh season of the CBS TV show Survivor. See each council, compare two players, and watch their voting alignment change over the season.
Full-season spoilers. Voting together is evidence of a shared choice, not proof of an alliance.
Every council, ballot by ballot
Choose a council to see the votes. Episode seven has two separate councils; the finale has two more.
How often did they choose the same target?
Each cell shows agreement as a percentage, with matching / shared ballots beneath it. Select a cell to follow that pair over time. Players are grouped by their original tribe.
Darker cells mean more agreement. A dash means no shared ballots, not 0% agreement. Small denominators can make a pair look more consistent than the evidence warrants. Agreement in a nearly unanimous council also tells us less about a particular relationship. Open a pair’s history to compare it with agreement across all voter pairs at each council.
A pair’s history, one vote at a time
Compare cumulative agreement with the individual ballots beneath it. A flat line can mean the players agreed at the same rate—or that they did not both vote.
The table also shows how many voter pairs agreed at each council. For a 7–1 vote, 21 of 28 pairs agree (75%); a matching pair is less distinctive in that setting. This is descriptive context, not a chance-adjusted score. This view always shows the full season. Filled points mark a shared ballot. Gaps before the first shared ballot mean there is no evidence yet.
Three patterns to explore
These comparisons describe recorded votes. They do not establish who planned a move or why someone changed sides.
Methods, coverage, and exceptions
The source is Daniel Oehm and contributors’ survivoR dataset, via the local Gamebot warehouse snapshot ingested on November 9, 2025. This extract covers US season 7. It contains 104 player-event records across 16 councils. It does not include the Outcasts’ return vote or the final jury vote.
- Shared opportunity. A pair contributes one observation only when both players cast an eligible ballot in the same council and round. Matching target IDs count as agreement. Absence or a non-voting event contributes neither agreement nor disagreement.
- Quit event. All five records at Osten’s departure are excluded from alignment, including four entries naming him as the target. They remain visible in the timeline, labeled as source records rather than counted ballots.
- Other non-voters. Rupert’s kidnapped event and the two non-voters at the final-three council are excluded. Lillian’s final-three ballot remains in the timeline, but creates no shared opportunity because she is the only voter.
- Returning players. Burton and Lillian retain their player IDs across departure and return. Their time out of the game is not scored as disagreement.
- Ordering. Council sequence and voting round identify an event, rather than episode alone. This extract contains no revotes or nullified ballots. The build checks for duplicate player-round keys and stops if unfamiliar rounds appear.
- Interpretation. Raw agreement is not adjusted for how unanimous a council was. Coordinated split votes can look like disagreement. Percentages are descriptive, and the shared-ballot count is always shown alongside them. No social-alliance labels are inferred.
The analysis uses a small SQL extract and a Python validation step. The three views run in the browser from the same exported records. Read about the project.
An independent fan analysis by Maxwell Grody. Experience · All projects