A Question of Predictions

A QUESTION OF PREDICTIONS

HOME

Oh, I know we have all the spoiler sites in the world but that has never really been my cup of tea. It's partly because I couldn't appreciate the "spoilers of mass destruction" - SMDs. All they did was ruin the show for me and I disliked the condescending nature of these web-sites that pretended they were so brilliant when, lately, all that was really happening was someone slipping an envelope under their doors. I think the biggest reason I didn't get into the whole spoiling aspect of the game though, was that I just plainly wasn't that good at it.

I never took much notice of me ranking the play of players as any kind of indicator of who might win until, seemingly out of nowhere, Vecepia won in Survivor-Marquesas and I realized I had her, along with Sean, towards the top of my vote point standings from very early on in the show. Looking back, it was pretty apparent that the eventual winner was always towards the top, more often right at the top, in points.

This is all fine and dandy in hindsight but the questions became, when was a good time to look at the points accumulated to act as a predictor of the eventual winner? In my mind, the first major transition in the game occurs when we are down to nine players. I consider this the transition from the Opening-Game to the Middle-Game (see my Survivor Strategy Guide for the meaning of these terms). By this time the tribes are usually merged and the players have found themselves into one of the following rolls, dominant pair, third or fourth, weaker alliance or weaker tribe (see A Question of Trends for definitions of the terms). Besides points, there is another indicator that I look at, that is position. I'll explain them both, giving examples from each of the Survivors.

POINTS

The following table shows the points accumulated by each player at this stage of the game in the first seven Survivors. The winners are bolded.

Vote Points When Nine Players Left
Borneo		Australia		Africa		Marquesas		Thailand		Amazon		Pearl Islands
Rudy	5	Colby	5	Brandon	3	Sean	6	Brian	3	Rob	5	Christa	6
Sue	5	Tina	5	Ethan	3	Vecepia	6	Clay	3	Matt	4	Rupert	6
Colleen	4	Keith	4	Kelly	3	Neleh	3	Ted	3	Jenna	3	Sandra	6
Richard	4	Amber	3	Kim J	3	Paschel	3	Jan	2	Alex	2	Lillian	4
Kelly	3	Jerri	3	Kim P	3	John	2	Helen	1	Deena	2	Jon	2
Jenna	1	Alicia	2	Lex	3	Kathy	2	Jake	-1	Butch	1	Tijuana	1
Gervase	-1	Elizabeth	2	Frank	0	Robert	2	Ken	-1	Heidi	1	Burton	0
Sean	-2	Nick	2	Teresa	0	Tammy	2	Erin	-2	Christy	-1	Darrah	0
Greg	-3	Roger	2	Tom	0	Zoe	2	Penny	-2	Dave	-1	Ryan	-2

As you'll notice, five winners (Tina, Ethan, Vecepia, Brian & Sandra) were in or tied for first while two winners (Richard & Jenna) were in or tied for third. Not bad by my count, but not entirely satisfactory either, especially considering the number of ties - Ethan was tied with five other players for first. In fact, we had ties for first in each season but Amazon and there the top person didn't even win. Not much of an indicator, so I began looking at others.

POINT PERCENTAGE

It's been brought up that points alone isn't quit fair as players often have gone to council a different number of times giving some players (especially the ones from tribes that lose a lot of challenges) more opportunities to earn points. Interestingly enough, it rarely makes a difference as the following table shows.

Vote Percentage When There Are Nine Players Left
Borneo		Australia		Africa		Marquesas		Thailand		Amazon		Pearl Islands
Rudy	125%	Colby	125%	Brandon	75%	Sean	120%	Brian	100%	Rob	125%	Christa	120%
Sue	125%	Tina	125%	Ethan	75%	Vecepia	120%	Clay	100%	Matt	100%	Rupert	120%
Colleen	100%	Keith	100%	Kelly	75%	John	100%	Ted	100%	Jenna	75%	Sandra	120%
Richard	100%	Amber	75%	Kim J	75%	Neleh	100%	Jan	67%	Alex	50%	Lillian	100%
Kelly	75%	Jerri	75%	Kim P	75%	Paschel	100%	Helen	33%	Deena	50%	Jon	40%
Jenna	25%	Alicia	67%	Lex	75%	Robert	100%	Jake	-25%	Butch	25%	Tijuana	25%
Gervase	-25%	Elizabeth	67%	Frank	0%	Tammy	100%	Ken	-25%	Heidi	25%	Burton	0%
Sean	-50%	Nick	67%	Teresa	0%	Zoe	100%	Erin	-50%	Christy	-25%	Darrah	0%
Greg	-75%	Roger	67%	Tom	0%	Kathy	67%	Penny	-50%	Dave	-25%	Ryan	-50%

Not a heck of a lot to take note of, except that Jenna was the only winner not to have at least 100%, and she still had a respectable percentage of 75%. All and all there isn't much information here, so I'll continue to use just the points table as my indicator of strategic ability.

POSITION

As much as I would like to pretend otherwise, this game is not all about strategy and social standing within the game is huge. You can see my arguments in A Question Of Leadership on how I divide these players into their different roles, but one thing seems pretty obvious to me and that is that there is a social hierarchy in Survivor and players tend to vote for the people in the higher standing in the final two. They just don't appreciate up-starts. This definitely puts the players in the higher positions at the advantage.

What I do at this stage is try and guess who each player supports the most, who is that player discuss strategy with and they would tend to take with them, if it came to that, into the final two. If you visit my Leadership page you will see all kinds of diagrams at the bottom that show these support arrows. The first players to take note of are the ones in mutual supporting pairs in which each member of the pair supports the other. These pairs are very powerful as these players will pretty much always vote together if they have a choice. In fact, of the twenty-one pairs that have existed at this stage of the game, all but one of them (Kelly and Sue from Survivor-Borneo) was broken up only when one of the members of the pair went out of the game in a situation that was out of the control of the other.

Every winner has come from one of the paired relationships, so the first thing to do is look for the pairs. Second is to place the players according to the following rankings: player with the most support, that player's second, third and fourth, weaker alliance, weaker tribe. The table below ranks the players according to the power of their position. Again, winners are bolded.

Position of Players When Nine Left
		Borneo	Australia	Africa	Marquesas	Thailand	Amazon	Pearl Islands
Paired	Most Support	Richard	Tina	Ethan	John	Brian	Jenna	Christa
	Second	Rudy	Colby	Lex	Tammy	Clay	Heidi	Sandra
	Thirds & Fourths	Kelly Susan
	Weaker Alliance		Amber Jerri		Paschel			Burton
					Neleh			Jon
	Weaker Tribe	Colleen Gervase Greg Jenna	Alicia Elizabeth Nick Roger	Brandon Frank Kim P Teresa	Sean Vecepia	Penny	Butch Dave	Tijuana
						Jake		Darrah


Unpaired	Thirds & Fourths		Keith	Kim J	Robert Zoe	Helen Jan	Alex	Rupert
				Tom			Rob
	Weaker Alliance	Sean		Kelly	Kathy	Ted	Deena	Lillian
							Christy Matt

	Weaker Tribe					Erin Ken		Ryan O

The higher up, the better the position of the player or, if you like, the higher their social standing. Players in the same box are in equivalent positions. What immediately stands out is five of our winners (Richard, Tina, Ethan, Brian & Jenna) were the player with the most support. Another thing that is worth taking note of is that four players have been first in both points and position (Tina, Ethan, Brian & Christa). Three of these four went on to win. Not bad odds there, eh? Certainly it's at least on par with some of the super-spoilers that have come down the pipe, but we're not finished yet.

PUTTING IT TOGETHER

I think the best way to sum this all up is to add their ordinates (1st, 2nd, etc.) from the point list and the position list and then rank the players according to the sum. Well, here it is.

Ordinate Sum
Borneo		Australia		Africa		Marquesas		Thailand		Amazon		Pearl Islands
Rudy	3	Tina	2	Ethan	2	John Paschel Sean Vecepia	6	Brian	2	Jenna	4	Christa	2
Richard Susan	4	Colby	3	Lex	3			Clay	3	Rob	6	Sandra	3
		Amber Jerri	7	Brandon Kim P	4			Ted	8	Heidi	8	Rupert	8
Colleen Kelly	8							Jan	9	Butch	9	Jon	9
		Alicia Elizabeth Nick Roger	11	Kim J	8	Neleh Tammy	7	Helen Jake	10	Alex Matt	10	Burton	10
Jenna	11			Frank Kelly Teresa	10							Tijuana	11
Gervase	12					Robert Zoe	12	Penny	11	Dave Denna	11	Lillian	12
Greg	14							Ken	14			Darrah	13
Sean	17	Keith	12	Tom	15	Kathy	14	Erin	16	Christy	16	Ryan	18

What immediately stands out is the position of the winners. In four of the seven seasons, the eventual winner was alone in first. Sandra was alone in second, Richard was tied for second while Vecepia was tied for first, though admittedly with three other players. Perhaps more significant to note is that the eventual winner was never more that one point out of first place.

SUMMING UP

So what do we have? All told, these methods would have had us picking the winner in our first pick in four of the seven Surivors (Tina, Ethan, Brian and Jenna). These are certainly pretty good odds considering that we are looking at 63 players spread across seven games. To put it into perspective, the odds of this happening randomly is only about 0.4%. The one thing that is clear, this game is anything but random.

Even with our less than successful picks, things didn't go all that bad. In Survivor-Marquesas we would have been having difficulty picking between four players, but one of those players was the eventual winner. In the Borneo and Pearl Islands games, we picked the wrong person, but both Richard and Sandra were just one point out of first. That means that if we limit ourselves to players in first or one point behind, a winner is in that group every time.

A final thing that can be looked at is editing. The producers of the show definitely like to give the players a bit of character and that means giving them a sort of story arc. Often this gives people a sense of how players are going to do. This is very subjective, but to give an example, in Survivor-Marquesas one could certainly sense John's eminent demise at this stage of the game and most were not too shocked when he went in the very next tribal council, meaning few would have picked him in their top three at this stage. As well, as the show progresses, sometimes some of your picks will fall by the wayside. In Survivor-Pearl Islands, Christa was gone four episodes after you would have made the call and you would have gone to your second pick, Sandra. I can guarantee you that you would have been a damn sight ahead of most everyone else. In fact, the spoiling community was, for the most part, calling Darrah as the eventual winner, but we would have recognized that as hog-wash long ago.

I write this while Survivor-All Star is in it's run. Currently we are down to ten players, so after the next episode I will be making my call as to the potential winners. It will be interesting to see how I do.

H.C.

May 12/04:
All-star didn't turn out to well for Rob, but it did for my system. Anyway, on to the numbers. If you check out the my predictions I made at the time, you'll find the positions slightly different. At the time I had Kathy more with Shii Ann and I didn't believe Rob planned on honoring his final four deal with Rupert. Both assumptions were wrong, but in the end inconsequential.

First the voting points when nine players left.

Amber	2
Jenna	2
Alicia	1
Rob	1
Tom	1
Rupert	0
Kathy	-2
Lex	-4
Shii Ann	-4

Next, position.

Player Position
Paired	Most Support	Rob
	Second	Amber
	Weaker Alliance
	Weaker Tribe	Kathy
	Weaker Tribe	Lex
Unpaired	Thirds & Fourths	Jenna
	Thirds & Fourths	Rupert
	Weaker Alliance	Alicia Tom
	Weaker Tribe	Shii Ann

Now, adding their rank in both

	Points Rank	Position Rank	Total
Amber	1	2	3
Rob	3	1	4
Jenna	1	5	6
Tom	3	7	10
Alicia	3	7	10
Kathy	7	3	10
Lex	8	4	12
Rupert	6	6	12
Shii Ann	8	9	17

Jenna's unpaired status (something she created herself by not reciprocating Rupert's support for her) makes her indeed a long shot despite her high point standing, making the only good picks Amber and Rob with a slight edge to Amber. And guess how it turned out. :-)

Unfortunately, if you read the above link you'll see that I ignored my own numbers and picked Rob at the time. You would think I would learn by now not to question the data.