This is incorrect thinking Soncy.

I recently wrote an article for the upcoming Poker Pro Magazine about poker bots where I interviewed some of the best AI specialists and game theory people. Some of what I learned there applies here.

All games have EV and variance as long as they have an element of luck, math based or not. This is how economists have developed game-theoretical Nash-equalibria for many games not purely math based like checkers, chess, Go etc.

So the fact that Survivor has only been played 20 times does not eliminate the variance it just makes it VERY hard to overcome.

Granted Survivor is FAR more complex than any of the above listed games but this does not free it from the same game-theoretical conditions which apply to simpler games. I know for example that AI specialists are working on bots (which are basically EV optimizing programs) for games as complex as Starcraft and World of Warcraft even though these are very complex.

Thank you. I ammend my protests along the lines of Survivor is not = to poker. And I suspect it is too complex and subjective to create a reliable model with a sample size of 20 seasons to work with. And also, that even with this proposed passive to aggressive scale ranking system, you can't even get people to agree what the rankings should be so how reliable and meaningful can the results be?

If nothing else, it would definitely be interesting to see. As long as someone else does all that work!

I agree the results would not be highly significant but could still be interesting. Obviously the more people that ranked them the more I could average and get a fairer result.

If people dont want to do it though that is totally cool. I just thought it would be interesting to try to REALLY see which strategy was better.

And all those games can be played by thousands upon thousands in various places and with similar conditions. This game is finite, a player may go 4 times(3 has been done now) before physically/mentally they are coconuts.

The beauty of Survivor is the individuality of each set of contestants and how they interact in a social setting under stress.

You can go through the motions of assigning values to variables in an attempt to solve, but like LOST, each value/variable will require another, and so on................but even the most disciplined person will never achieve consistency. The game may use logic, but is not a logical game.

The same game-theoretical optimal plays govern tic-tac-toe and Survivor. It is simply a difference in complexity. It is obviously far harder to model the later, but not impossible.

I would most likely want to boot anyone i couldn't predict or i thought was willing to take huge chances or make illogical decisions. People like Shambo, Dreamz, Sugar, Russell, and Tyson would be the ones i'd most likely blackball or want gone early. If i can't trust you even if you suck, i am going to have trouble wanting to keep you around. If i can trust you to make logical or predictable decisions, even if i think you are very intelligent, i am more likely to keep you around, at least at the start. I think Yul and Cesternino are among the best ever but i would not necessarily want either gone early on if i was out there with them.

Mmmmm, good answer.

Kirby just got mentioned on the 2+2 pokercast. No one is hotter. Go Kirb.

*tear*

But I'm moving to Santa Monica ASAP! lol

If you watch her interview in the beginning of the first episode of HvV, the way she talks about how much the game took out of her in Micronesia then she pauses and then says "but it was worth it right? I got the million"... anyway, <3 Parv

From the Parv zap2it interview... <3 Parvati


You aligned with Russell, is he not such a bad guy? Or is he just as bad, or worse, as we saw on TV?

He is intense. I don't know what America is smoking but he is a lunatic. He's out of his mind. He is a very aggressive game player and very controlling and very competitive. He has a very large ego, so whenever any kind of move went down, he wanted to take the credit for it. Which is why he's so hard to live with because he's constantly watching everyone and wants to know what's going on. I just tried to calm him down every time I talked to him. I'm like, "Listen dude, you're outplaying yourself right now. You're irritating everybody, just chill out, go do some downward facing dog for a second and then we can talk strategy because you're driving me nuts. Go take a time out and we'll talk about this later." That's what you have to do with him, you have to let him know what's going on and hope he backs off.

Russell does have some redeeming qualities. He's a very loyal player. He also is willing to take huge risks -- he's very bold in his game-play so he was a good guy for me to align with. I don't think anybody else would have given me an Idol. He was ready to go home, he thought his name was being written down too that night and he gave me an Idol.

Gotta love Billy

can someone explain to me why everyone thinks rich is one of the greatest of all time? I just watched season 1 and it seems like he's just a solid level 1 thinker who took advantage of probably the weakest competition ever.

everyone who goes on survivor gets some prize money. There is no realistic way to do this EV calculation.

Third gets like 75k, 4th gets like 50k and so on.

I love how when a maniac slashes and burns alliances from halfway through the game is called loyal by the only person he stayed loyal to after aligning with everyone.

There's no way an edit could be tricking us here. Russell is by far one of the most disloyal people to ever play. Parv is just blinded by the fact he only had 1 or 2 shots to boot her and he actually came to the right decision of not.

The biggest problem with doing this is that there are too many variables. Looking at something like aggressiveness on it's own would not paint a true picture. Let's just throw in a second variable, call it likability, which basically represents how well the person is received by their tribemates (and relates fairly closely to how many votes they will get at FTC). If you were to model the four extremes and the proposed winning % it would probably look something like this:

Aggressiveness: 1 Likability: 1 Winning %: ~0%
Aggressiveness: 1 Likability: 10 Winning %: ~5%
Aggressiveness: 10 Likability: 1 Winning %: ~0%
Aggressiveness: 10 Likability: 10 Winning %: ~100%


Obviously the winning % are very rough estimates, but I think it can be agreed that a person with an extremely low likability will basically never win, whereas a person with an extremely high likability will almost always win if they can make it to the end.

Now here comes the biggest problem: how do you rank a person's likability? It is nearly impossible to do so accurately. Not only is it extremely subjective, but it is also completely dependant on who that person is on a tribe with. An individual could be perceived as extremely likable in one group of people while being seen as completely unlikable in another group. It is dynamics such as this that make Survivor such an interesting game, but also one that is damn near impossible to model objectively.

I find Tom Westman's win in Palau to be an excellent example of this. He had what was IMO one of the most straightforward roads to victory ever. He was a good leader, a straight shooter (for the most part), and he got put on a tribe that admired and respected those qualities. Had his tribe resented his leadership or disliked his personality he could have just as easily found himself in the same position as Hunter was in Marqueses.

My rankings based on my often faulty memory. I skipped the ones I don't remember although I've watched all seasons.

1 - very passive
10- very aggro

Hatch - 8
Tina - 6
Ethan - 3
Vecepia - 5
Heidik - 8
Jenna - 4
Sandra - 3
Amber - 2
Chris - 7
Tom - 8
Danni -
Aras -
Yul - 7
Earl -
Todd -
Parvati - 6
Bob -
J.T. - 7
Natalie - 3

Kelly Wiglesworth - 3
Colby Donaldson - tough one. Composite score 4. First 2 seasons 7. This yr 2
Kim Johnson -
Neleh Dennis -
Clay Jordan - 6
Matthew Von Ertfelda -
Lil -
Boston Rob - 9
Twila -
Katie -
Steph -
Double D -
Ozzy - 4
Cassandra -
Dreamz -
Courtney -
Amanda -
Amanda -
Susie -
Sugar -
Stephen - 8
Russell - 10+
Parvati - 6

That is the nature of a model. It is a simplified version of a complex system. Sure using only active/passive does not represent the whole game but it would serve as a good baseline for the game and might provide initial information. It's impossible to totally model any game but that does not make the results of a simple model useless.

wilt chamberlain sucks imo, played vs noobs

Wilt probably fathered half the cast, who's gonna vote off their own dad?

The Stilt would own Survivor.

You want to apply objective analysis using completely subjective data. Data is considered hard, it doesn't carry an emotional component. it can be interpreted in many ways but is always there in the forefront.

The problem with your direction for a model is the old "garbage in, garbage out" system. This is not a Russell bash or Sandra love in now, so don't go there. His play never really changed from s19 to s20, but your perception and respect for those tactics changed. I can't believe YOUR personal leaning towards a base strategy changed overnight, so your perception changed. It was difficult to tell whether your Sandra post was a level or acceptance, but if you could now see what was invisible before, again your perception was in play.

So how can you model perceptions that will always be tainted depending on their holder?

Congrats Sandra even though Parvati owned the entire game.

10000 time for this thread to be closed!

Disagree with pearl isalnds being one of the five toughest fields.