# Thread: Success in Survivor - A Mathematical Model

1. ## Success in Survivor - A Mathematical Model

On a separate thread in this forum, and perhaps in other Survivor forums in this archive and around the internet, this question is asked: “Who do you think played the best game of Survivor?” Inevitably, certain names always seem to bubble to the tops of these lists, and not just the ones who came out as the victor in their respective season. In fact, people who finish the game as low as 5th or 6th are quite often listed as among the best players.

But what is more likely is that a strong consensus as to who really played the game well cannot be reached. People, inevitably, have castaways they like and those they dislike and those opinions can cloud their judgement as to who really played the game well.

So what I thought I would try to do is design a mathematical model which helps describe a successful player of the game of Survivor. In doing so, I attempted to quantify those qualities which I felt helped to make a good player of this game. I came up with five measurements. They are:

1. Immunity Challenges Won This is obvious. The more IC’s you win, the better you are going to be. I gave a greater weighting to individual immunity challenges than I did to tribal ones. I also gave a greater weighting to each successive challenge. The idea is, the more people you are sitting at TC with, the less conspicuous you are. Having immunity at TC, for example, where there are nine other people is not as important as being immune at a TC with two other people.

2. Reward Challenges Won These quite often have a cascading effect. If it is food, for example, it will make you physically stronger so that you’ll perform better at future challenges. I know, it is not always food that is the reward. Matches, blankets, pillows, letters or videos from home, clothing, massages, a warm comfy bed for an evening; it does not really matter. If rewards don’t make you physically stronger, they sure can do wonders for your spirits, morale and emotional health.

3. Total Tribal Council Votes Against This measures both the popularity of each castaway as well as the “under the radar” factor. The fewer the votes a castaway has had against them, the more points they would get. This measure also takes into account the number of TC’s the castaways attend. In other words, it is possible for people voted out earlier in the game to score better in this category than those you go further.

4. Surviving a TC Vote This attempts to measure how impervious a castaway who did not win immunity is at TC. If a castaway wins individual immunity, he or she would get points based on actually winning the challenge (as per measurement 1). In this category, however, points are awarded to those who make it through TC without being voted out in spite of their vulnerability because they do not have immunity. This helps to give a measure of game aspects like alliances or the flexibility a castaway exhibits by avoiding being voted out in the face of a fallen alliance (think Rob from S6 – he scored highest in this category from all 16 in The Amazon). This factor is also weighted more heavily post-merge than pre-merge and more heavily as the game progresses and tribal numbers dwindle.

5. Casting a “Correct” Tribal Council Vote This factor is calculated based on the frequency with which a castaway votes at TC for the person who is booted that evening. Its intent is to measure a castaway’s awareness of their tribes politics. For example, if a castaway is led to believe the tribe is voting a certain way and does so as well, but the rest of the tribe votes in another way, it’s a sign that this particular person has been intentionally left “out of the loop” so to speak. This factor also helps to measure whether or not the castaway is part of a successful voting alliance.

So who, according to my model, were the best players and worst players of Survivor? I have the results from all six seasons thusfar, and I’ll post them later today. This post is long and boring enough already and, besides, I don’t have the data with me right now!

2. Very well thought out. Nice job.

3. I am breathlessly anticisapating your results... This is fun!

4. Very cool, tmcrae. Can't wait to see the results. Did they surprise you?

You aren't by chance an engineer, are you?

5. Originally Posted by chompstick
Very cool, tmcrae. Can't wait to see the results. Did they surprise you?

You aren't by chance an engineer, are you?
Yes. Some results really were quite surprising. But, given the methodology I used, all the results are explainable.

And, no, I'm not an engineer. Just a lowly computer programmer!

6. *shivers with anticipation*

7. There are other models, look at

http://www.realitynewsonline.com/cgi...724.art&page=1

and

http://www.realitynewsonline.com/cgi...025.art&page=1

Those models are fairly complicated.

8. Tmcrae,
Well done, I fancy myself to be a mathemetician (Hope I spelled it right !) at heart. As a quick rememberance, not actually putting it on paper, I would imagine that according to your computations you may have come up with the following conclusions:

1) Kelly from S1
2) Colby from S2
3) Jenna from S6

9. Originally Posted by astrogirl_2100
There are other models, look at

http://www.realitynewsonline.com/cgi...724.art&page=1

and

http://www.realitynewsonline.com/cgi...025.art&page=1

Those models are fairly complicated.
Thanks for the links, astrogirl! But those models are of a predictive nature. In other words, they use data from previous series' of the game in an attempt to predict who will win the current game. My model, on the other hand, tries to explain who played the game better and why. Kind of like giving credit for being a strong competitor where credit was due.

Having said that, the models look very interesting and, as you put it, very complicated! Certainly well beyond anything I would try to tackle!

10. Originally Posted by tmcrae
Thanks for the links, astrogirl! But those models are of a predictive nature. In other words, they use data from previous series' of the game in an attempt to predict who will win the current game. My model, on the other hand, tries to explain who played the game better and why. Kind of like giving credit for being a strong competitor where credit was due.

Having said that, the models look very interesting and, as you put it, very complicated! Certainly well beyond anything I would try to tackle!
So... What's the answers? Did I get it right