Series Info...Trials, Triumphs & Trivialities #135:

Anatomy of a Game: Survivor Twists

by Shannon Appelcline

First, apologies for the mysteriously missing column two weeks ago. I've been up and down with mysterious ailments which have on occasion mucked with my ability to write. With that said, this week I want to return to ground that I've very recently covered: reality TV shows. No, I'm not going to turn this into a regular reality TV column (honest), but I do continue to believe that some reality TV shows offer good examples of game design by showing what we designers do every day--and the show that seems to do that best is Survivor. This week I want to specifically talk about Survivor "twists"--how the producer Mark Burnett has changed the game from season to season in order to surprise and upset players, and how he may be trying to shape gameplay by doing so. It all centers around one of my most basic tenets of game design, The Game is what the Game is, which is to say you entirely determine how a game works by the positive (and negative) stimuli you offer within the game.

Survivor Play & Problems: An Overview

Survivor plays a lot like Big Brother, that other reality TV show that I discussed just a couple of weeks ago. Here's the main points:
  1. 16 people gather together in a rugged terrain.
  2. They're divided into two teams of 8.
  3. The two teams compete and each week the losing team is forced to go to tribal council and vote off one of its members.
  4. In all councils each members has equal power: one vote per player.
  5. When the total number of players decreases to 10, the tribes combine into one.
  6. At this point, members compete each week for "individual immunity"--that person can't be voted off that week.
  7. After each immunity contest, there's another tribal council, and another person is voted off.
  8. When there are only 2 people left, the 7 previous people voted off vote on a winner.
The gameplay is fairly robust but has created some issues in gameplay. Among them:
  • Voting Blocs. Voting blocs form which can cause boredom during the game, as a certain set of people are voted off one after another. For example in each initial tribe of 8, a bloc of 5 can vote off the other 3 one after another. In the merged tribe, a block of 6 can vote off the other 4 one after another. This is made worse by the fact that players are likely to form blocs based on their initial tribes, and thus the 3-5 weeks after the merge are likely to be very obvious once one tribe takes dominance.
  • Static Players. Because there's so much power in blocs, once they're formed players are not likely to play dynamically (and stasis doesn't make for good TV).
  • "Bad" Player Retention. Because of the power of individual immunity, it becomes important to vote off potential individual immunity winners--and this tends to slop over to the end of the tribe-oriented play. If you're voting the 6th player out of the game, and expect there to be a merge before the next tribal council, it makes sense for you to vote out a strong player before they ever have an opportunity to start winning individual immunities--and thus you have the potential to end up with weak players as winners and strong players as losers in the game.

Enter The Twists

Beginning with Survivor III (Africa), Survivor has introduced twists into every game that shake it up in various ways. Some of these twists have most likely been introduced solely for the purpose of creating better television. For example:
  • No Food. In Survivor IV (Marquesas), the show stopped providing the castaways with food. Potentially this has some gameplay effect, as it allows for a food-provider strategy, but in actuality the rations offered have always been short enough for that to be a possibility, as winner Rich Hatch proved in Survivor I.
  • Sexed Tribes. In Survivor VI (Amazon), initial tribes were based upon sex. Again, this has some gameplay effect by changing strategies. (A few women & at least one man complained that they couldn't win by flaunting their bodies to the opposite sex.) However, it was largely a publicity stunt for ratings.
  • No Luxuries. Survivor VII (Pearl Islands) took an interesting tact by throwing the players into the ocean with nothing but the clothes on their backs--and then gave them the opportunity to purchase supplies in nearby village. Again, no big changes in gameplay, but great TV (particularly as one of the contestants admits, "I'm not wearing any underwear").
More interesting, however, have been a number of different twists that seem specifically planned to offset some of the flaws in the initial game design, and I'll be discussing those twists further in a moment. Before I get there, however, I do want to mention one thing that every single twist did, whether it had a more purposeful game effect or not. Every single twist threw the players off their game and forced them to quickly rethink strategies. Doing so can, in general, force players to think more dynamically about the game and make them more willing to shake up their initial plans. If one of your goals in twisting around a game is to shake up "static players", that's a great start.

The Switches

Since Mark Burnett started twisting the game in Survivor III, the biggest twist has been the switch, where the two tribes exchange members well prior to the merge. It's been done almost every season:
  • Blind Selection. In Survivor III (Africa) three members from each tribe were self-selected to join the other tribe at the point when the tribes were six each. (The self-selection occurred without the players knowing what was going on, and thus was essentially random, though presumably selected toward groups of friends heading off together.)
  • Random Selection. In Survivor IV (Marquesas) a similar switch was totally random, based on selecting team colors blindly.
  • Optional Selection. In Survivor V (Thailand) players were given the option to join the other tribe if they wanted. No one did.
  • Self Selection. In Survivor VI (Amazon) two players, one from each tribe, sat down and choose new team members one at a time.
  • Minimal Selection. In Survivor VII (Pearl Islands) there was no switch per se, though one member of one tribe stayed with the other tribe for a couple of days, and helped them in a competition.
When these switches have worked (particularly in III and IV), they have had a remarkable affect on the way Survivor is played. In Survivor III (Africa) a cocky bloc in one tribe was utterly destroyed prior to merge, while in Survivor IV (Marquesas), the more powerful tribe coming into the merge ended up losing because the underdog members of their bloc teamed up with the other tribe to form a larger voting bloc. Overall, these switches seem to encourage the following types of gameplay:
  • "Nicer" Players. Tribes in the early game are forced to be nicer (or more deceitful) to underdog members outside their bloc, lest those underdogs turn against them following a switch. This of course increases doubt in what's going on, and thus makes for better TV.
  • Fuzzier Tribal Lines. Because tribes become more familiar with each other, players are more likely to break their original tribal alliances and make the later stages of the game more chaotic. Even the very limited switch in Survivor VII (Pearl Islands) could potentially have this result because one player is now in with the other tribe.
Given the initial problem of voting blocs, this method of making interpersonal connections fuzzier both at the tribal and the game level isn't a bad way at all to try and encourage a preferred type of gameplay--more dynamic and changing voting blocs.

Delayed Merge

Survivor V (Thailand) did something very different and perhaps a little cruel. It moved the two tribes into one location when they reached 10 members, but didn't actually merge them--though players were allowed to assume that that's what was going on. As a result one player totally shot herself in the foot by going over to the other tribe before discovering that her own tribe still had the ability to vote her off (and they did). However the delayed merge tactic did shape several types of gameplay:
  • Fuzzier Tribal Lines. Everyone staying together at one camp was yet another way to make those tribal lines fuzzier.
  • "Nicer" Players. In addition, it was another way to get players to play their game closer-to-the-vest. With both tribes watching it was harder to scheme, but made better TV when people managed it.
  • "Good" Player Retention. The show had actually been threatening delayed merges as far back as Survivor III (Africa) and this had an interesting meta-effect on the game. It encouraged tribes not to vote strong players off prior to the merge, because they could never know when they'd have to win a tribal immunity, where their strong players were assets, as opposed to an individual immunity, where their strong players were deficits. This particular point might not have affected Thailand, but in more recent games players have been somewhat more reluctant to vote off strong players early on because of this threat.

Returned Players

Last week Survivor VII (Pearl Islands) presented what they'd been calling the greatest twist ever. When the tribes were down to 5 members each, the 6 players who had been voted out first were brought back to compete as a third tribe. Everyone was told that each tribe these "Outcasts" beat would be forced to vote out a member immediately, and then the Outcasts would get to vote that number of members back into the game. The Outcasts beat both other tribes and thus one player was voted out of each of the extant tribes (and tonight the Outcasts will vote two of their members in). It was a great TV moment, when those 6 prior players emerged from the jungle, but just like the late merge two seasons before, it'll have a meta-effect on future shows:
  • "Good" Player Retention. Because ousted players might come back to compete, players will be even more reluctant to vote out strong players prior to the tribal merge.


At a minimum I think this liturgy of Survivor twists will make my core point: once you've found any core flaws in your game design you can adjust gameplay by adjusting that design and encouraging different types of behavior. However, I think it also points out that gameplay can adjusted in two different ways: within a game and without a game. In the first case it's immediately obvious to players, in advance, that new gameplay will have benefits. In the second case players don't realize that their undesirable gameplay is bad for them until the game shifts under them ... but then future players might have that much more incentive to change their ways. And if you've found this all interesting, you might now want to go back and read my original article on Survivor written a year and a half ago. Feel free to grade me on consistency.

[ <— #134: A Brief History of Roleplaying, Part Three | #136: Social Gaming Interactions, Part One: A History of Form —> ]

Miniseries ...