@orbsah  
They were cast by robots. :I

@orbsah  
A huge spike in votes in a very short period of time for one particular character, they wouldn’t know the exact amount that are fake because legit votes may have been cast during that time but they can get a close estimate.

how can you “fake vote” and how did they find out which ones are “fake”?

Took me a minute to remember who Coloratura was

@Seabridge Drive  
I agree.

@orbsah  
I think I get what you mean, although voting against ponies you dislike is what you’re supposed to do. I think you mean people would vote against ponies they do like, but whose popularity threatens ponies they like more.
 
It’s a known flaw of Borda counts. Borda himself said he’d made a voting system specifically for honest men. But runoffs are subject to even worse manipulation.

@Ferrotter
 
using concordat and borda is a horrible way to make a popularity poll. everyone would vote against ponies they dislike.

@cheezedoodle  
It’s not appropriate. Marble (Inky), Limestone (Blinky), and Minuette (Colgate) each had a ****load of fan opinions of them before S5 even started. That puts them in a whole different category than ponies who didn’t even exist prior to S5. And the problem is it not only affects them, but because of the mathematical defects of a runoff, including them when none of them are the winner means votes for or against them actually screw up not just their place in the order, but it screws up the entire order of candidates
 
There are methods to have compensated for that. This poll wasn’t using them. That much is pitifully obvious. Basic proof: In a runoff poll, Night Glider edged out Smooze by 50 votes. In the final poll, Night Glider trailed Smooze by 55 votes. So is Smooze more popular than Night Glider, as shown in the final poll? Or is Night Glider more popular than Smooze? They can’t both be true, even though the polling proves they both are true. But that’s because the methods of the poll are so inappropriate to the measurement being made. This kind of ****up is common in runoffs. But it also proves something else – the final order of popularity was changeable based on the order of the early-round polling(!) – meaning that redoing the poll with different runoff mixes could result in a different winner. Which means the poll didn’t find the winner, it found a winner for that specific order of choices!
 
And lastly, S5 was the first time Lyra ever had a speaking part. Same for Octavia. S5 was the first time that Special Agent Sweetie Drops existed; prior to E100, Bon Bon in her best moments was just a background foil for foreground pony action. She didn’t have a character. She didn’t even have a single voice actress say two different lines for her. And Party Favor, the Bugbear, and Grandpa Gruff weren’t even in the background before this! Plus we have a very different Derpy than we’ve ever (one ****ing time, which got substantially deleted/altered) had before. None of them were in the poll! So assuming you were correct that Colgate and the Pie sisters belonged, at the very least Lyra, Octavia, Party Favor, and the Bugbear belonged and were missing. And probably Bon Bon and Derpy too. (Lyra alone presents a very interesting case. She’d never once spoken, but the fact that nobody even considered her a new character who belonged in the poll only goes to show how little that matters in determining their popularity. She’s basically proof that the ones who existed before S5 are not new to S5 regardless of whether they were better characterized in S5 than they had ever been before.) Add to that the fact that the comments to the poll prove that there were people who treated them differently (didn’t vote for them because they weren’t actually new); combined with the defect of using a runoff, there’s no way (aside from random chance) that the polling order is accurate.
 
So how could they have done better? As I said before, the best way would’ve been a Borda count. I don’t know if they even have polling software that could do a proper Borda count. Another way (not as good, but at least it probably would’ve been accurate at picking the winner if not the order of those below) would’ve been to use a seed system and poll only the top ~5 as chosen by their staff. Let’s face it, anyone paying any attention to the fandom couldn’t possibly think Sassy Saddles, Tree Hugger, Greta, Rutherford, Svengallop, Wind Rider, Spoiled Rich, or the hillbillies had the slightest chance of winning. And it’s ridiculous to think that Limestone could’ve outpolled Marble; impossible for Twinkleshine or Lemonhearts to have outpolled Colgate; and unlikely that Sugar Belle or Party Favor could’ve outpolled Night Glider. Then take out the pre-existing ones so there is absolutely no confusion that it is for new characters only. So if you put only possible winners up, you don’t have that complicated a poll. Rara, Moon Dancer, Starlight Glimmer, Troubleshoes, and Night Glider. One easy 5-way poll. You could do it even by brute force Condorcet methods in only 10 1-on-1 polls. At 3 days a poll that’s less than a month of voting, and it would probably be accurate at giving you the entire popularity ranking for those top five.
 
And they could’ve put up much more specific instructions to vote for which one is the best, not which one you like the most.

@RDash5001  
@Background Pony #5D40  
@Ferrotter
 
They appeared, yes, but only as background fill. They didn’t become actuall characters until season 5, so having them appear on the poll is entirely appropriate.

@RDash5001  
That’s a small part of the detailed explanation I’m skipping.

@RDash5001  
Same with Colgate and Lemon Hearts.

Marble Pie and Limestone Pie aren’t even new characters. XD They’ve been here since what? Season one?

@LinkofEquestria  

 
@Background Pony #5D40  

 
This is a very concise statement of a big problem. It’s a complete methodological fail for its stated purpose of determining “best new pony”. It’s not a poll of “best pony,” it’s a poll of “most popular secondary pony at the end of Season 5 among EqD regulars” very specifically. And for that very narrow purpose, it could’ve been a good poll. (And probably still is after discounting the 600 fraudulent votes.)
 
To do a poll of “best new pony” you’d need to carefully characterize what it is you’re measuring, and design the poll specifically to measure that. This poll is not that. I’ll skip the detailed explanation why.
 
And that’s aside from runoffs being a defective measurement technique even for popularity contests. That’s easier to prove with simple math. Let’s say hypothetically, Night Glider is everyone’s #2, but votes are split only among Moon Dancer, Troubleshoes, and Party Favor. Those who vote for Moon Dancer consider Troubleshoes and Party Favor as the two worst, and likewise for the other two. Objectively in that scenario, Night Glider is the most popular new pony; everyone at least likes her a lot. Yet she drops out in round 1 because she would get no #1 votes.
 
There are three valid ways to do the math. One is you have to poll every character against every other character, one-on-one. The winner is the one who wins the most polls; in the event of two equal numbers, the one who wins their one-on-one contest wins. This is called a Condorcet vote. The second is a Borda count. You have everyone rank all n entries from #1 to #n. Then for each candidate, you add up all the numbers voted for that candidate; the candidate with the lowest total wins. Very closely related to a Borda count is “instant runoff;” there, you also need to do all the ranking, but you only count the #1 votes. It’s just that each time you select a winner with less than a majority, you drop the lowest ranked candidate, renumber everyone’s votes with that candidate removed (so if someone’s #1 was removed, their #2 becomes their #1, etc.), and recalculate the count of #1 votes. Repeat until you have a winner with a solid majority of #1s. All three methods have flaws and strengths. Borda is probably the best way to go with a popularity poll; it tends to get the “average” preference of the group rather than favoring #1 votes.

EQD really has a bias against stallions.

600 fake votes!
 
That is genuinely hilarious.

I don’t really get why people are in a tizzy over the results of a popularity poll. It is a little disappointing not having your favorite character on top, but we shouldn’t let those results embitter us since they ultimately don’t really matter, especially when it comes from an unofficial source like fan sites and what have you. Knowing what the fandom thought was the “best” character is only going to do just that.
 
And may I remind you, Starlight was a hot-topic example. The spotlight was on her literally the last two episodes of the series thus far. A couple weeks or so back, GameFAQs had a similar popularity contest amongst its games, and Undertale of all games came out on top, even past the classics like Super Mario 64, Legend of Zelda: Ocarina of Time, and many other games of their caliber. When it comes to memorability, what is most recent tends to stick out in people more than past examples, even if their memorability is somehow the same. (I don’t think there’s a way of truly quantifying memorability.)
 
Of course, that’s assuming she fairly won, which she didn’t, but still, that definitely has something to do with why both Coloratura and Starlight were both so high up on this poll, since both their spotlight episodes were fairly recent. Even if Starlight had two episodes already dedicated to her, like I said, the last two episodes of the series released thus far were focused on her, so they were more than able to renew interest in her character.

@some of the commenters here
 

Aww… there was a shitstorm and I missed it. That’s what I get for being on vacation with crappy internet.
 
PS: At least Treezie made it to the finals.

@zippysqrl  
Sorry ;_;

@Seabridge Drive  
Better silence than drama
 
Maybe we can’t ground Spider Man, but we can quiet those who post him, if they keep at it.

@Seiya-Meteorite  
better Spiderman than drama.
 
p.s.  

@zippysqrl  

@zippysqrl  
Hey, at least no one’s bickering anymore :P