Yuku free message boards

Wow Phoenix that was a monster of a post to get through!

Green Coffee...
i like your analysis very much, i love the idea of classification for the winner edits... but i do have some questions regarding how you are determining Type A vs Type B... I'm finding that i disagree wildly with some of your assessments.

I realize that you have been using your own ratings, which is fair, Type A / Type B is your ballpark after all... however... the data we have to analyze is the "official data" so that is what I'm using... and this actually makes up a large part of why i am in disagreement with some of your assessments.

Yeah I assume many of my ratings will lead to different conclusions than the norm would. I fully understand if they come to some different conclusions.

Since we are ONLY at episode 4.... i think that to drop someone from Type A because they are too P or too MOR is inconsistent with past data. For comparison... I think it would be helpful if you could clarify what you think some of our past winners were...

This may be true. Too P and Too MOR are sort of subjective things. The way I'm thinking on tone, I base it on the likelihood that the player's tone will "even out" by the time a pick would reasonably need to be made. So far, I just have JT and Coach in that boat. Both of their edits have been 1 tone all the way. The overall impression (which is what I think tone is all about, really) is at the extremes.

Usually it seems that every season has characters at these extremes. We have superheroes and we have villains. At this point, I see it as unlikely that either of these roles will be sufficiently muted for these characters. Could JT get several Ns in a row, leading to a sort of "Mixed" tone by the merge? He COULD but it seems unlikely at this point. I may have jumped the gun on this one, of course, but in my estimate, he is overly positive.

As for "Too MOR," I'm assuming this refers to Spencer, and to some degree Stephen. With this, I think this is where the two different MOR types thing becomes an issue. I see narration and mostly flat screentime as the "bad MOR" in the long term if that's all they get." If the edit is the MIX kind (some strategy, but some UTR, etc.) that seems to be a little different for winner chances. (Differentiating the two could actually resolve the issue of finding your Type Cs that Craig brought up).

Essentially, Spencer is almost all narration with little personal insight and development. To me, that suggests he lacks meaningful airtime. Stephen has more meaningful airtime (which is why I still like him as a winner), but a lot of his time has been sort of narrator-y.

For comparison... I think it would be helpful if you could clarify what you think some of our past winners were...

Well, knowing that Edgic is slow to warm up to shifts in theory, I originally started by keeping it simple - two types. The catalyst here was all of the defeatist commentary about how "Edgic just can't predict Bob's because they don't fit the guidelines." The same thing happened with Danni, who was an "Exception to the guidelines."

Therefore, my only real goal was to create a set of guidelines based on Danni/Bob, as they were the best examples of "WTF HOW DID THAT HAPPEN!?" The real dividing points for me are two things: 1) Winner does not fit traditional Edgic (What I called Type A), and 2) Constraints on the winner's story are such that Type A is not possible (Bob being unstrategic and Danni hiding it). So the way I saw it was that if Edgic got it right, the winner was Type A (barring spoiler nonsense) - the guidelines worked. If Edgic COULD HAVE gotten it right but missed for whatever reason, the winner was Type A. If the winner violated the guidelines such that Edgic could not get it right, I saw that as Type B. This left me with just Bob and Danni, with outside lines of Amber, Vecepia, and Chris.

This is not the best way to do this. It's based on results and not data. However, it was a start. What you have proposed here is a great way to go with it.

For example...
I think we are all in agreement that Bob & Danni are definitely Type B winners (and based on editing patterns Amber fits here too), they all skew UTR in their early edits:

Bob OTTP3 • UTR1 • MOR2 • UTRP1
Danni UTR2 • MOR2 • UTR2 • UTR2
Amber UTRP1 • UTR1 • OTTP2 • MOR3

And I would also say that everyone would be in agreement that Richard, Yul, Earl & Todd are all true Type A winners... heavy CP & higher visibility in the early episodes - maybe some tone, maybe some cool down:

Rich CP5 • CPP4 • CP4 • CPN3
Yul CP3 • CPP4 • CP4 • UTR1
Earl CP4 • CPP4 • UTR2 • CP3
Todd CP4 • CP3 • CP4 • MOR3

But... here's the rub... we have 10 other winners (more than half) who don't fit these 2 extreme edits of Type A (very complex) and Type B (very UTR).

Just as a clarification, I did not intend for A and B to be at opposite extremes, although it does work out in an interesting way if you think about it. The 3 above were the least noticeable winners. The 4 below were the most noticeable winners.

What do we do with winners Ethan, Chris and Parvati?
are they Type A or Type B?

Ethan CPP4 • MOR2 • UTR2 • MORP3
Chris CPN4 • UTR2 • MOR2 • UTR2
Parvati CP4 • UTR2 • MOR3 • MOR2

Using my classification (which is based on results and whatnot):
Ethan: ??? (No Edgic at that time, although his edit was probably used to craft the original guidelines)

Chris: A. Some people think it's B but I think he was given a lot more strategic talk during the premerge than the chart indicates. His early episodes were pretty muted, but nothing like Bob/Danni. He also had quite a bit of N to his character, which Type Bs probably should not have.

Parvati: A. Edgic got it right using the traditional guidelines.

Now, is this the best way to do it? Again, I think not.

These 3 have very similar early edits. They all have a very strong episode 1, followed by 3 lower-key edits. I don't know what I would classify these 3 as.... are they Type B? they started strong but then all took a back seat to the rest of the action (like Bob). Could they be Type A's because of their strong start?? I don't actually know... I think this is a completely different edit.... but identifiable because of the strong Episode 1. I can't see anyone saying that Chris was a Type B, and even Parvati would be a stretch to call a Type B. I think this is a hybrid edit. Enough of Type A in episode 1 to identify, but with a lot of early cool-down, like the Type B.

This makes sense, too, in that all three of these COULD be identified by traditional guidelines (and since they have A components, they were in one case).

That leaves us with the REST OF THEM, the largest group of winners... they all start out so.... MOR-ish

Do you consider people like Aras (S12) and Tom (S10) to be Type B winners? How about Brian or Jenna or Sandra, even Tina? Unlike Bob & Danni... their early edits do not lean UTR. And Tom is actually MORE POSITIVE than Bob, if that's at all possible. And Brian, Jenna and Aras all have negativity to their edits by Episode 4.

Tina INV • MOR3 • MOR3 • CPP4
Vee UTR2 • MOR4 • CPP4 • MOR3
Brian MOR2 • UTRP3 • CPP4 • UTRN2
Jenna MOR4 • MORN3 • UTRN2 • UTR2
Sandra MORP4 • MOR4 • OTTN2 • UTR2
Tom MORP3 • MORP3 • OTTP4 • MORP3
Aras MOR2 • MORN3 • UTR2 • MORP3

I considered putting Tina & Vee with the Type B / UTR winners, but other than the E1 INV/UTR, they both became became very MORish, so I put them here.

I think most people would want to classify this group as Type A.... there is too much going on for a UTR / Type B edit. However... these edits are not complex in the beginning... these winners are present, they are around commenting on the action... they never get really huge, but they never disappeared either, they skew Middle of the Road... I don;t think that they fit Type A better than Type B... but they certainly don't fit Type B better than Type A!! I want to give them their own classification becuase when you look at them as a group... it sort of stands out, doesn't it?

Again, this makes sense, as well. As an aside, I noticed Tom's tone problem during Palau and I didn't want to pick him for that reason.

So... I guess this brings me to a few questions for Green Coffee, as well as my own analysis...

Firstly GC, my questions are of course, how are you classify the past winners into Type A and B? Who do you see where, and why? And how do you resolve the disparity of the majority of the winners not fitting nicely into the Type A definition. How are you defining Type B? If it's truly UTR, then someone like JT doesn't fit, does he?

Looks like I jumped the gun on commenting here heheh.

All winners except Danni/Bob are A's so far (and Vecepia/Amber MAYBE).

The actual rankings JT is receiving is much more in line with Aras and Tom, rather than Amber, Bob or Danni. And really, most people would never classify Tom, Aras, Brian, Sandra... or even Jenna as a Type B! These winners are much more developed than Amber, Danni & Bob in the early episodes and they have more in common with the Type A's than the Type B's.

And lastly, specific questions pertaining to GC Type A break down... you have listed as probable Type A winners as Sierra, Stephen & Erinn. And then Tyson, Brendan, JT, Spencer are listed as the "middling/hanging on" Type A's:

Using the official data, so that we are comparing apples to apples, that looks like this. The asterisk (*) indicates unofficial vote:

Sierra CPP4 CP4 MOR2 UTR2*
Stephen MOR2 CPP3 MOR2 MOR3*
Erinn UTR2 UTR2 CPM4 CP3/MOR3*

Tyson OTTM2 MOR3 CPN4 OTTM3*
Brendan UTR2 CP5 CP4 MOR2*
JT MOR2 MORP2 MORP3 CPP3/MORP3*
Spencer MOR2 MOR2 UTR1 UTR2*

Sierra fits GC's Type A very well, and Stephen, can still fit the Type A style but is on the MORish side.

But why do you have Erinn listed as a PROBABLE TIER 1 Type A winner?
Erinn does not fit the pattern of a Type A at all... her edit fits the pattern of the Type B winner way better.

Tyson doesn't really fit any of the winner patterns but with all the problems concerning his E1 close calls for both rating and tone, i can't eliminate him either. Does he fit Type A... maybe... but definitely NOT Type B.

Brendan does not fit any of the winner's patterns. Actually... whoever called that Brendan looks like Cirie from S16 is right on the money (i think it was Blake)... the tone is different, but exactly the same ratings and visibility as Cirie UTR2 / CPP5 / CPM5 / MORN3. Correct call to remove Taj UTR2 / CPP4 / CP3 / CPP5* from this list, she is also like Cirie S16... but then why not Brendan? Officially... he has the same rankings as she does.

I think that JT has much more in common with winners like Aras and Tom, than Danni or Bob, and as a result, he should be in your top tier for A's.

Spencer, having 2 back to back cool downs, is a bit much, but could still keep him in contention (like Jenna or Sandra), even though i think he's a slightly better fit for Type B.

And here is where the differences in our strategies become clear. In making my rankings, I am just applying guidelines. Do they look like they are fitting/breaking the guidelines or not? Those who aren't or are meeting guidelines look good. Those who don't go along look bad.

Your analysis comes more from a comparison of actual data patterns. Do the ratings of episodes 1-4 look the same as the ratings for winners' 1-4? Thus, while JT is meeting guidelines for Type B, you are right in that he isn't quite fitting a pattern in terms of the data. While Erinn is fitting all of the guidelines well, her pattern is way off.

I think both ways of looking at it have merit and both are important. I should probably look at whether or not the patterns match more.

Phoenix's Analysis:
So, i found that trying to organize my thoughts about the Type A / Type B analysis to ask Green Coffee some questions, I have realized something that has been bugging me for a while.... there are not 2 types of winners... there are actually 4.

The 4 Winning Edits Model

Type A = Complex Edit
complex individual in early episodes. Leaning Strategic & Logical. Mostly CPs, but by E4 will have one cool down episode (A MOR or UTR). Higher Visibility, averaging between 3 & 4. Mostly Neutral tone, may have one or two episodes with tone, but will not end up overly P or N. Type A's win because we have seen from the start how they played a good game.

Type AB = Complex UTR
This edit combines Type A & Type B. Episode 1 is Huge and Complex. We learn about them and their strategy. They may be toned in E1 (P or N), and they will have a higher Vis (around 4) and they are definitely shown to be complex & having their head in the game. The following early episodes (E2-E5), the AB winner takes a back seat. They may go UTR for a couple of episodes, but will likely stay MORish, and their visibility drops (around 2ish). Their tone will likely remain neutral. This winner wins mostly because of game and partly due to the mistakes of their F2/F3 partners.

Type B = UTR Edit
True Under the Radar, early episodes almost void of character development and thought processes. We meet them, we know a little about them, but we do not see the action through their eyes. Early episodes are mostly UTR with low visibility, but never truly invisible. Will have at least 1, maybe 2 early episodes that tells us something about them (MORish or OTTish, not CP). Tone will be Neutral to positive, but not negative. The Type B winner is the "alternate" winner or the "It's Not XXX" winner.

Type C = The MOR Winner
This contains the majority of the winners. This winner starts off as a MOR character. They are presented in E1* as somewhat likeable, they are not overly high in visibility. By Episode 4, every MOR winner has had tone (either Positive or Negative, or both). The MOR winner's early episodes may range in visibility from 2 to 4, but averages in near 3. The MOR winner has several MOR edits in the early part, but will likely have at least 1 episode that is not MOR (it could be UTR, OTT or CP AND it will likely have tone). The MOR winner edit takes off around the merge. This winner wins equally because of game and due to the mistakes/unlikeable of their F2/F3 partners. (*Tina of course is the exception, but as Mark Burnett has already stated that was intentional)

4 constructs!? Good luck getting that one to fly! ;)

Love it. It does seem to be that there are 4 ways in which winners emerge. I think one way to sort of lend support for this would be to do some sort of "Factor Analysis" on the winner data to see if 4 patterns do emerge. I'm not sure I know, however, how one would enter the data into a statistical program to do such a task (I'll take psychometrics sooner or later, I guess). In the meantime, "eyeballing" it probably is sufficient to show that there are definitely lots of types.

Traditional Edgic works like a sledgehammer - you hit on the big targets that identify a winner throughout most all types (A, AB, C). This is a full tool kit. You see which tool will break which targets with the least amount of mess, and you can also hit smaller targets (Type B), so to speak.

The new task, then, is to figure out if you can make up guidelines for identifying each type, beyond just data patterns. No winner will ever perfectly match a previous winner's chart so I think it's still important to have theories that can be tested in terms of IDing winners from a more qualitative standpoint. Further, you have to differentiate those who look like MOR winners from people who are actually MOR fluff, Type A masterminds from those who get blindsided, etc. This is the tricky part, I think.

I want to take a look at everyone using the expanded definitions that i did above. Type A, AB, B and C. Mostly because i am having difficulty resolving the largest group... the MOR winners.... this group generally start off the season so MORish... and I think that Green Coffee is RIGHT in that by only looking for the Type A or AB edit... we are missing a lot of information and we overlook certain people. Based on all of the above, here is my analysis of the remaining players using the 4 Winners Edits Model...

*Cut for space*

Winners List
1 Stephen/JT (Both solid Type C. Type C contains the most winners)
2 JT/Stephen (Both solid Type C. Type C contains the most winners)
3 Sierra (Possible Type AB, some small issues that prevent her from a True A, but could be AB like Parvati)
4 Erinn (This is so odd to me, I never had her here before... Good fit Type B, Decent fit Type C)
5 Tyson (still needs some more clarification, possible Type A or Type C, but still betrer than the following)
6 Brendan (Possible Type C, but early pattern looks like Cirie from S16)
7 Taj (Possible Type C, but early pattern looks like Cirie from S16)
8 Spencer (Possible Type B, but only because I am being generous)

Eliminated
Coach (Too N, 4 in a row!)
Debra (INV / too UTR)
Joe (Too UTR)
Sydney (INV / too UTR)

I agree with your assessments, here, for the most part.

One thing I'm curious about (and another thing to think about for the 4 type analysis) is how the editors determine which one takes priority. You have JT/Stephen at the top because most winners are Type Cs. However, is that because Type C is the dominant Type (editors prefer it) or because the editors were unable to craft a Type A/TypeAB in many seasons? If Type C is dominant, Type C winners should be preferred. However, if another Type is dominant, but just rare, those winners should still be preferred, even if they don't happen often.

This would be exceedingly difficult to determine, because we do not know what characters in what seasons had potential to be what edits. Does Eliza have enough pull with the editors to ask them about this one at the next reunion? Haha.

Type B, I think both of us would agree, is the weakest type (editors would much rather give the winner a big story that shows them as strategically sound or at least competent). Is Type C the preferred type? Not so sure. It has the most winners, but I think it's one of the harder ones to differentiate from your average MOR character at this point (another reason for Edgic to think about how to classify MOR, I guess). Type A's are good TV, after all, and I always feel super satisfied and pleased when they win. Type AB and Type C you understand why they won at least, but it may be based on the mistakes of others.

Thus, I think it's also possible that Type A is the dominant one, while Types AB and C only come in when they can't show the strategic mastermind take the game and must show the other details. No proof here - just speculation. So how do AB and C stack up?

Looking forward to this line of conversation, as I really think you're on to something.

flf:
Same ratings. Different content.

Earlier this season, some posters were asking how two characters could get the same ratings when they were so different in content.

This episode seems a good example of the limits of the stats.

Sierra is at the bottom of the visibility stat this ep. She, Debbie, Joe, and Spencer all were classified as UTR2.

Yet, even with the lowest visibility score, her edit had a confessional about the long game and her role in it, as well as SPV from Brendan, Taj, Sydney and Spencer.

This illustrates my argument against bumping such situations up to MOR2

Some may argue that she should be MOR because of her complexity, but the UTR/MOR cut-off this week seems to be at Vis 2.0 by consensus.

I would agree that she had complex comments and obvious edit-tending with SPV from the other tribe, during its pre-TC scenes (If that doesn't make you sit up, nothing will.)

During the time when the edit is normally trying to generate tension and interest around the TC vote, the eds stick in comments about Sierra. Go figure.

But, with such a low Vis count and relatively high complexity, the edit rational is best expressed as intended to fly Under the Radar.

The problem is that Sierra fits the name "Under the Radar" but the others do not. Sierra has all of this game related content, but is not an overpowering character (hidden and under the radar of the viewers). Joe, Debbie, and Spencer aren't even in a plane this season, so they aren't flying under any radars. For them, UTR means "virtually no character development."

This means 2 very different edits with different qualitative content (strategy vs nothing) and different winner implications (good vs. bad) are given the same Rating label based solely on quantitative content (low vis). This quantitative measurement is already represented in her visibility number.

I have no issue with people ending up with the same rating with different edits (it will happen). I just think sometimes it becomes very misleading on the chart and there are many situations where it doesn't need to be.

.