(Warning: this post contains spoilers, so if you haven’t been watching Celebrity Traitors, you might wish not to read on…)
As well as being the TV event of the year so far, Celebrity Traitors, which concludes tonight, is a near-perfect natural experiment. My interest was piqued by a conversation between contestants historian David Olusoga and comedian Lucy Beaumont in the Scottish castle where the show is filmed. Olusoga had only just escaped banishment in the previous round table by drawing lots after tying for the most votes. He bemoaned his situation and thought he was likely to be voted off next. Beaumont, though, responded by saying that as a group, “We tend to move on, to be honest”.
She tapped into a sense I had felt while watching the show that the voting patterns for banishment were different from those in the other “regular” series. And this posed the question of whether celebrities play the game differently from ordinary contestants? If they do, it hasn’t worked to their advantage. As an article in the Guardian noted, “the sheer ineptitude has been staggering”, with the celebrities only managing to catch their first traitor at the end of programme 8 of a 9-episode run.
When the contestants gather at the roundtable, there is a stronger sense of chaos and a lack of direction. Specifically, my instinct is twofold:
- Individually, the celebrities’ voting is less consistent than normal contestants – as per Beaumont’s comment, they were less likely to vote for the same person two rounds in a row.
- Collectively, the celebrities were voting in a more fragmented way than players in other iterations of the series.
But are these impressions correct? With a bit of help from screen captures of the voting records for four series of the Traitors thus far in the UK (three regular series plus the Celebrity version) and Claude AI to do the maths, we can actually test these ideas (I’ve uploaded the prompts and outputs from Claude here).
First, in terms of vote consistency, the Celebrity edition is undoubtedly lower than the last two series, Series 2 and 3. Celebrity contestants voted for the same person that they had in the previous round in the round immediately afterwards on 7.8 per cent of occasions. This isn’t an all-time record, though: in the first season, the equivalent figure was just 6 per cent. So, here, the figure is low but not unprecedented.
| Series | Repeat voting percentage |
|---|---|
| Celebrity | 7.8 |
| Season 1 | 6 |
| Season 2 | 17 |
| Season 3 | 12 |
The fragmentation of voting tells a slightly different story, though. To measure this, we can use a statistical tool used by political scientists to compare the fragmentation of a party system after an election: the Effective Number of Political Parties (ENPP). Essentially, the higher the number the calculation generates, the more fragmented a party system is. To illustrate this, you can use it to compare the vote share in the US Presidential and UK General Elections of 2024. The US vote gives you an ENPP of 2.08, very close to a perfect two-party system. This reflects the very high vote share achieved by the Republican-Democrat duopoly. In contrast, the UK produces a figure of 4.75, a system approaching 5-party politics, reflecting the success of the Liberal Democrats, Reform, the Greens and the SNP.
The table below shows the ENPP for the Traitors roundtable. In addition, I have calculated it as a percentage of the total number of options available to voters (e.g., if there were 10 players and the ENP came out as 3, that would be shown as 30 per cent). This is because the number of players changes during the game, and the various iterations of the Traitors have different player counts at different points due to rule tweaks. The table is also colour-coded, with lower figures in green and higher numbers in red. Here, the celebrities stand out far more, with consistently more fragmented voting patterns than in any iteration of the regular show.
Here, we can clearly see the difference between the celebrity and regular versions of the series, with celebrity voting patterns much more fragmented. Indeed, while the average for all three iterations of the regular version of the series is around 20 per cent, this leaps to 30 per cent for the celebrities, indicating a much greater level of fragmentation in their voting patterns.
Now, many of you might be asking at this point, what is the wider significance of this? We can think about this by asking why the celebrity version has different characteristics. Of course, it could be that there are meta-explanations for these patterns, outside the scope of the game. The celebrity contestants might know each other already, or be conscious that a particular figure in the Castle has the potential to advance their career in the real world. They may also have what the podcast The Rest is Entertainment refers to as “producer brain,” where they are using their own knowledge of show business to try to second-guess how the game is designed or to make themselves stand out in the game. And of course, it could also simply be luck, something that happens when you throw a random group of people together. After all, we only have a single celebrity series as a sample. But there may also be deeper explanations. Everyone in the Castle is highly successful in their own field. The median participant in the series is likely to be very driven and quite individualistic. With these kinds of personalities, is it really surprising that voting patterns are fragmented and that there is less convergence around a collective strategy?
In a 1967 interview with Time magazine, Andy Warhol predicted that society was entering a phase “when everyone will be famous for 15 minutes.” As well as a comment on the transience of fame, Warhol is arguing that the traits of celebrity are becoming more widespread. The characteristics that seem to be shaping celebrity voting patterns in Traitors and making them distinct from the regular iteration of the show are precisely those associated with growing fragmentation in broader society, including in terms of party politics, where socially grounded collective action is becoming increasingly hard in fragmented societies. The celebrities’ attempts to catch the Traitors may appear chaotic and lacking an overall strategy, but in reality, they are demonstrating hyper-concentrated versions of the traits that are having a massive impact across contemporary society. It is therefore not the case that the celebrities are fundamentally different to regular players, but instead they are harbingers of particular ways of acting and engaging, the promulgation of which we are already feeling far beyond the Scottish highlands.