American Idol Season 10 commentary

AT&T Favoritism

Telescope Inc., which is the company that does the show's vote counting, has revealed (here) that “AT&T customers can receive a text reminder after each performance show with all the contestants' names and numbers.”  This gives a very big advantage to AT&T wireless customers.

When you consider that during the final few weeks, the regular (non-wireless) phone lines are bottlenecks that limit each contestant to almost the same number of phone votes, text messaging gives AT&T wireless customers exclusive "back-door" voting access.  Unlike online voting, text messaging is not limited as to how many votes each customer can cast.  In fact, last season, AT&T employees taught American Idol fans how to do "power" text messaging.  And unlike phone voting, text messaging has virtually no capacity limits for how many messages it can handle (because text messages are tiny and travel through a different path compared to voice phone calls).

Of course this arrangement is unfair, and each season it becomes clearer that the results are now controlled by text-message votes.  (In earlier seasons there were far fewer text-message votes.)  This arrangement of giving out contestant numbers to people who don't watch the show is also unfair to the other sponsors like Coca-Cola and Ford, because those sponsors have fewer people watching their ads.

Apparently Telescope Inc. is still failing to do much statistical analysis on the vote counts, and instead mostly uses the voting counts that AT&T provides.  This is why they failed to identify the major phone-system failure in the California Bay Area that heavily contributed to the early elimination of James Durbin (as described below).  (Vote splitting between James and Haley Reinhart also significantly contributed to his early elimination.)  If they were doing their job, the vote counters at Telescope would have known about the major phone-system failure, and where it happened, just based on the data they get (or should be getting) from AT&T.  They wouldn't even have to wait to read accounts of the phone-system failure in the online forums.  The fact that one of the producers denied the phone-system failure reveals that he was not properly informed by Telescope Inc., and suggests that AT&T did not admit the failure to Telescope.

Each season refinements are needed to keep American Idol voting fair.  Let's hope that the show's producers wake up to what the show's viewers know, which is that credibility for the fairness of the elimination order is declining.  This is unfortunate because the show offers an opportunity to show how voting should be done, so that someday we can get fair results in government elections where the outcomes affect the lives, the health, the livelihoods, and the prosperity of all Americans.

Top 2 results

Of course Scotty McCreery won, as most of us expected.  Although vote splitting causes unexpectedly early eliminations, someone who is the clear front-runner throughout the season still wins, in spite of vote splitting among the other contestants.

Voter turnout was higher than predicted here, partly because Lauren Alaina gave great performances, and Scotty McCreery backed off on his performances, which made his winning seem less certain.

See new section below, titled “Votes for James Durbin got lost?  Were misdirected?”
 

Top 2 — finale — commentary

As predicted weeks ago in Votes for Scotty McCreery from Country-music fans below, Scotty McCreery is certain to win.  The total number of votes cast after the finale might be embarrassingly low because many people won't bother to vote.  This brings up the point that voter turnout increases when voters feel that their vote will make a difference, and voter turnout is low when voters feel that their vote(s) will not make any difference in the outcome.  Next week there won't be much dedication for non-Country-music fans to vote, which is a contrast to the top-3 week when non-Country-music fans desperately, but unsuccessfully, tried to outvote the Country-music fans.

Top 3 results

Country-music fans have taken over the official voting.  This happens each time there is a good Country-music singer on American Idol.  This year, as long as they are voting for one Country singer, namely Scotty McCreery, and there are so many Country-music fans voting, and the real competition, namely James Durbin (and Pia Toscano) have been (unfairly) eliminated, Country-music fans can also vote for the other Country-music singer, Lauren Alaina.

A significant fairness issue is that many of the Country-music fans who vote are not necessarily watching American Idol on TV.  It's not necessary if their Country-music friends tell them, via email or via a phone tree (or learn through a Country-music radio station), which phone numbers to call, or text-message, and when to start voting.  The effectiveness of this approach highlights the opposite of vote splitting, namely vote concentration.  Vote concentration occurs when a group of voters agree, in advance, to concentrate their (first-choice-only) votes on a single contestant — or politician.  In contrast, this week, the (non-Country-music) folks who normally watch the show split their votes among all three singers, which left Haley Reinhart without enough (first-choice) votes to counterbalance the Country-music voters.

Top 3 commentary

This week's poll results — after removing ballot-stuffing attempts, especially from fans of Scotty McCreery — shows that Scotty McCreery and Haley Reinhart are very close in popularity, with Haley Reinhart slightly (but not significantly) ahead. 

The strong popularity of Scotty McCreery, plus vote splitting between the two females — Haley Reinhart and Lauren Alaina — will eliminate one of the females.  Which one?  There are several factors involved:

• Haley Reinhart is more likely (than Lauren Alaina) to get the votes of James Durbin fans.  (This preference is conveyed in the Top-4 VoteFair poll results, where James is the most popular and Haley is ranked higher than Lauren.)
• The loss of James Durbin might demoralize non-Country-music fans, resulting in less efforts to network and support Haley Reinhart.
• As two VoteFair visitors have pointed out (thanks!), Country-music fans will be splitting their votes between Scotty McCreery and Lauren Alaina.  However, the voting for Scotty McCreery is so strong (as measured by DialIdol.com) that lots of Country-music fans can still vote for Lauren Alaina instead of Scotty and possibly put both Country-music singers through to the finale.  If this seems far-fetched, consider that Country-music fans were splitting their votes last week, and yet apparently each of their favorites got more votes than front-runner (in first or second place) James Durbin!
• Apparently the dedication of Country-music fans, combined with their effective networking in the form of telling others when to vote and which phone numbers to dial — because many Country-music fans who vote are not watching the show — has kicked into high gear, in advance of the finale.  That dedication, if it is not matched by voters who prefer Haley Reinhart, can easily lead to an all-Country finale, with both Scotty McCreery and Lauren Alaina.
• First-choice-only voting always has the chance of creating surprises.  The fact that Lauren Alaina might reach the finale is a big surprise.  In the week 11 commentary is a quote that "She is not popular enough to win", which is still true, but who would have expected her to outlast James Durbin and Pia Toscano and have a strong chance of reaching the finale?  (For balance, back when Haley Reinhart was making lots of singing mistakes, she too was unlikely to get this close to the finale.)

What can be done to address the unfairness of non-viewers voting?  Besides once again recognizing that the idea of only collecting first preferences yields flawed results — and that this has big implications politically — it would be fairest not to cast any votes for Lauren Alaina unless she is really your favorite, above Scotty McCreery and Haley Reinhart.  Based on last week's results, she will be getting plenty of votes from non-viewers who are Country-music fans, and any extra votes (including sympathy votes) could put her into the finale.  Expressed another way, in this contest between American Idol viewers and normally-non-viewing Country-music fans, it's important to concentrate votes on either Haley Reinhart or Scotty McCreery.  (Concentrating votes is the opposite of splitting votes.)

Votes for James Durbin got lost?  Were misdirected?

Voting irregularities for last Wednesday night's official voting are beginning to be recognized.  (Thanks to the VoteFair-site visitor who reported this!)  One such forum is at AutismKey.com.  There are numerous claims that attempts to vote for James Durbin were unsuccessful, or were misdirected to Lauren Alaina.  Such mistakes more easily account for why top-ranked James Durbin was eliminated prematurely.

The reports are confusing because three different mistakes were going on.

• Many of the show's viewers dial phone numbers based on the order in which the singers perform, which caused many fans of James Durbin to dial number eight because James Durbin sang in the number-eight spot, yet he was assigned numbers one and five (because he also sang in the number-one spot).  In contrast to previous years, phone voting no longer gives confirmation as to who you just voted for (and instead just confirms the contestant number), so many voters probably cast hundreds of votes (each) for Lauren Alaina instead of James Durbin without knowing they were voting for the wrong contestant.  As one fan wrote: "…but for people who used the phone…after weeks of competition, we ignore the #s and go by the order in which they sang…producers know this…"  This was not a voting-system failure, but rather a failure of viewers to pay attention to the numbers.  (Some people would say the producers are at fault for the inconsistent order — even though Ryan repeatedly reminds viewers to pay attention to the numbers.)  This issue is mentioned here first because some reports of voting irregularities do not clarify the nature of the irregularity, which means the complaint is not necessarily about the voting system.
• People in the California "Bay Area" — which is adjacent to James Durbin's home town of Santa Cruz — have reported not being able to get through on the correct phone numbers for James Durbin.  As one fan wrote: "I tried for three hours to vote for James, kept getting 'all circuits are busy.'"  Another fan wrote: "Every week since voting started, I called to vote for JAMES on 2 phones continuously for 2 solid hours with NO problems.  But Wednesday, on 1 phone I kept getting a busy signal & the other a message 'all circuits are busy-try your call later'.  Only a handful of my votes went through. James was robbed of my votes!"  Yet another fan wrote: "Can you please tell me why I was not able to get through to vote for James but turned around and called back to vote for Scotty to test it and it went straight through!!!"
• Some people have reported that when they attempted to vote online for James Durbin the confirmation was for Lauren Alaina.  As one fan wrote: "When I clicked to vote for James, the picture that popped up afterwards[sic] was Haley’s."  (Note that numbers are not used in online voting, so the number-related confusion is not related.)

The apparent widespread failure of online voting is not surprising.  When Constantine Maroulis was prematurely eliminated in Season 4, it occurred because of a widespread failure of AT&T's phone system, which was unaccustomed to handling hundreds of thousands of people dialing the same phone number at the same time.  Now, in this first year for American Idol's online voting, it isn't surprising that the servers (which are the computers at the other end of an Internet connection) are unable to handle the tremendous volume produced by hundreds of thousands of votes coming in at the same time.  (In fact, the most common way to attack a website is to bombard it with as many simultaneous users as possible, which is called a "denial-of-service" or DoS attack.)

Keep in mind that the number of voters, and the number of times each person votes, significantly increase as the weeks progress.  (Similarly the number of voters in the VoteFair poll usually increases each week.)  This means that what has worked in previous weeks may break down as the finale approaches.

As for the phone-system problem, the large concentration of phone calls from the same area — namely the San Francisco Bay Area — can easily overload the limited number of "central offices" where the phone lines connect to the main phone-system network (and where a limited number of servers may be handling the calls).  In contrast, votes for Scotty McCreery from people in the Southern states are coming from a wider geographic area (although people in Nashville and Atlanta might have experienced a similar busy-signal congestion).

Although it is not likely to have happened, another possibility is that a phone-system operator in the Bay Area could have effectively disabled one (or both) of the phone numbers assigned to James Durbin.  This capability is intended as a defense against a "denial-of-service" (DoS) attack on a phone number (such as when a large corporation is getting lots of angry calls that block their normal phone service).

It's worth mentioning that phone-system providers (like AT&T) are legally required to keep their phone service running 99.999 percent of the time.  That's why phone systems are so reliable.  In contrast, internet connections do not have such requirements for reliability.  Remember that the digital age is young, so there are still lots of bugs that need to be worked out.

Internet connections are much more vulnerable to reliability problems, and — unlike modern phone systems — they can be attacked by people who understand, and have access to, their inner workings.  So it is very possible that American Idol's online voting system was hacked into by someone who successfully blocked, or reassigned, votes intended for James Durbin.  Another possibility is that someone who controls a portion of the Internet, or controls servers that host the American Idol voting, could have thought the servers were being attacked (in a "denial-of-service" attack), and unknowingly assumed they needed to limit access to those servers in order to protect other users accessing other websites (which is the standard way to handle a DoS attack).

What really happened?  The American Idol show never openly admits to voting mistakes, so we may never know.  What is certain is that if the vote counting, and the voting method, had been fair, James Durbin would not have been eliminated!

Top 4 results

James Durbin was the most popular singer in this week's VoteFair poll, yet he got the fewest official (first-choice) votes.  What happened?  A combination of these factors probably accounts for his premature elimination:

• Vote splitting between Haley Reinhart and James Durbin partially accounts for his premature departure.  They are both "edgy", and they provide an alternative for those who don't (as much) like Scotty McCreery.  This means that if Haley Reinhart were to be eliminated next week, those voters would be without a singer in the finale that they strongly like.  There was also vote splitting between Lauren Alaina and James Durbin (as alternatives to Scotty), but their similarity is less obvious.  The pairwise counting used in the VoteFair poll automatically takes such secondary preferences into account, and the results tell us what we already knew, which is that the real most-popular choices are Scotty McCreery and James Durbin.
• Apparently the fans of James Durbin were not as dedicated in their voting, compared to the dedication of the fans of the other singers.
• The VoteFair poll here only allows one ballot per participant, whereas official voting allows many votes per person.  As pointed out here before, this difference can easily make this poll's first-choice-only counts different from the official vote-counting results.
• Perhaps the fans of James Durbin primarily relied on online voting, which limits voting to 50 votes per Facebook account.
• The fans of Haley Reinhart and Lauren Alaina may have used text-message voting more than the fans of James Durbin.  This is an advantage because it can be done faster than online voting (because of the need to type the "captcha" characters).  The cost of text-message voting works to the advantage of affluent voters, and the fans of James Durbin probably tend to be less affluent.
• The fans of Haley Reinhart and Lauren Alaina may have used phone voting more than the fans of James Durbin.  This year phone voting can make a bigger difference than in previous years because online voting has freed up the phone lines (compared to previous years when the phone lines were saturated, and each singer got close to the same number of phone votes).
• Male County-music fans may have heavily voted for Lauren Alaina (while female Country-music fans probably provided plenty of votes for Scotty McCreery).
• Recent reminders that James Durbin is married probably reduced the number of votes from female teenagers.
• Christian churches may have organized their votes to be concentrated on Scotty McCreery and Lauren Alaina, more than James Durbin (whose affinity for heavy-metal music appears to be stronger than his affinity to Christianity).
• The voters in Southern states probably concentrated their votes on Lauren Alaina and Scotty McCreery.
• This week's first song sung by Haley Reinhart — Michael Jackson's "Earth Song" — probably resonated with lots of young voters, who are discouraged by the destructive choices made by U.S. (and European) politicians during the past few decades.  (Young people don't yet realize that the flawed vote-counting methods used in U.S. and European elections are to blame for electing special-interest puppets who are disliked by most voters of all ages.)
• Software geeks might have written software that can automatically read the letters that must be typed for each online vote, and that software would have been used more by male fans, who would tend to favor the two female singers.  Such software would make it much easier to reach the 50-vote limit.  (If such software hasn't yet been written and distributed, it will become available in the near future because this isn't a terribly difficult software-development task — for the single-style "captcha" images currently used in American Idol voting.)
• There is also the possibility that some other voting factor will become known later.  (Last year it was a Facebook page posting the wrong number for phone voting.  The year before it was organized power text-message voting, with assistance from AT&T employees).

Top 4 commentary

If this week's voting is done on the assumption that no one is safe, then Lauren Alaina is most likely to go home.  If she were the only Country-music singer, then she would have a significant advantage, but most Country-music fans will be supporting Scotty McCreery.  Lately, Haley Reinhart has been growing in popularity — especially among guys — so that makes James Durbin and Scotty McCreery less safe than in previous weeks.

A recommended new rule to improve fairness

The producers of American Idol are considering changing next season's voting rules to counterbalance what they believe is a bias in favor of male singers.  The most obvious example of this bias appears to be this year's very early elimination of Pia Toscano.  Yet, as explained below, she was not really the least popular, just the singer with the fewest first-choice-only votes.

Fortunately there is a simple way to improve the fairness of eliminations, without taking control away from us, the voters.  It would be a new rule.  (It would be an added rule, not a replacement for the "judge's save.")  Here is the rule, conveyed in words as they might be written on Ryan's prompter:

Ryan:  “Last season [Season 10] there were some very big surprises in the results.  Two singers who had never been in the bottom three suddenly got the fewest number of votes.  Apparently lots of people, when they voted, thought those singers were safe, so they voted strategically to try to control the order of elimination.  But those singers were not safe.  The first time it happened, with Casey Abrams, the judges used their one save.  The second time it happened, with Pia Toscano, the judges could not do anything.  Her elimination was especially shocking because many people thought she had a very good chance of winning.  So, this season, we are adding a new rule.  Starting with the top ten, if a contestant ends up with the fewest votes, and they have never been in the bottom three, the judges can send home someone else in the bottom three.  However, the singer they send home has to have previously been in the bottom three.  So what does this mean for you when you vote?  It means you need to keep your favorites — that's plural, favorites — out of the bottom three.  For our judges, they still have the same "judge's save" that they can use just once (at any time, up to the top-four week).  Now they can also keep things fair if a front-runner — someone who has never been in the bottom three — suddenly gets the fewest votes.  If this rule had been in effect in previous seasons, the judges could have saved not only Pia Toscano, but also could have saved Siobhan Magnus in season 9, (Michael Johns in season 7), Chris Daughtry (and Mandisa and Kellie Pickler) in season 5, Constantine Maroulis in season 4, and Tamyra Gray in season 1.”

The singers who might have been sent home earlier, include contestants such as Tim Urban (in season 9), Syesha Mercado (in season 7), and Kristy Lee Cook (in season 7).

This rule would not have helped Jennifer Hudson because she had been in the bottom three for two previous weeks.  Also, the rule would not have helped in Season 6 when Gina Glocksen and Chris Sligh were eliminated before less-popular Sanjaya Malakar because Sanjaya was not in the bottom three when these other two were eliminated.

(Additional details about this recommended rule:  If the judges express interest in using this rule to save a singer, Ryan will reveal who has the second-lowest number of votes so the judges can take that information into account.  This rule cannot be used in the top-four or top-three weeks.  Reminder:  This rule is in addition to, not a replacement for, the "judge's save".)

If the judges think the voters are biased against female singers, the judges can use this rule to protect a popular female singer who they think is better than a bottom-three male singer (who has previously been in the bottom three).

This simple rule would be much fairer that adopting the approach used on So You Think You Can Dance.  On that show the judges need to compensate for the fact that on performance night each dancer is dancing in a different style from the other dancers.  In contrast, on American Idol, every singer can choose from the same songs (or from any song style in the case of the year-of-birth week).  In other words, there are no "kiss-of-death" song styles on American Idol the way there are "kiss-of-death" dance styles such as the Viennese waltz.

As for the gender bias, what's really going on is that female fans are better networked than male fans, and young females encourage each other to vote, vote, vote for their favorite singers, which means they are probably voting more times than male fans.  And of course young females prefer male singers.  Is that bias unfair?  Not necessarily.  Keep in mind that the voting is really trying to measure how many of the show's fans will attend each singer's concerts, buy each singer's music, and influence others to like those singers.

As described here repeatedly, there is an even fairer way to improve the voting.  However, it would be difficult for Ryan to explain, challenging for some people to understand, and less dramatic for the result-night shows.  This fairer approach would be to allow each Facebook-authenticated voter to cast a single 1-2-3 ballot — in addition to their usual first-choice-only "votes" — and count those 1-2-3 ballots using pairwise counting.  Pairwise counting would correctly identify which of the bottom three singers is really the least popular.  And it has the advantage of using only voter preferences, so that the judges don't get excessive influence.  As can be seen in each week's VoteFair poll results, this pairwise-counting approach correctly identifies the real popularity of all the singers.

Whatever changes the producers make, the change must not take control out of the hands of the voters, such as by letting the judges cast some votes, or asking the judges to assign scores.  After all, the popularity of American Idol comes from the fact that we, the viewers, fully control the outcome (within the limits of first-choice-only voting).  The rule recommended here keeps control in the hands of the voters, and only gives some control to the judges when there is an inconsistency in the voting results, namely when a front-runner (who has never been in the bottom three) suddenly gets the fewest number of (first-choice-only) votes.  (And that only happens because the show does not reveal ranking results above the bottom three.)

Top 5 results

Jacob Lusk has been at the bottom of these VoteFair polls for the last five weeks, so his elimination was long overdue.  Apparently he has very dedicated fans.  However, the same number of fans cannot outvote other voters when they concentrate their votes on a smaller number of contestants.  Another way to look at this is that it demonstrates — in reverse — what happens in politics; adding more candidates to a political race makes it easier for a money-backed candidate to win (if only first-choice votes are collected), whereas having fewer candidates (especially just two) makes it harder for a money-backed candidate to win a primary election.

Top 5 commentary

This coming week, either Jacob Lusk or Lauren Alaina is likely to be eliminated.  Which one?  It's difficult to predict because there are too many conflicting considerations.

Top 6 results

No big surprise this week.  Weeks ago Casey Abrams was at the bottom, and only stayed because of this season's "Judges save".

Votes for Scotty McCreery from Country-music fans

In last week's results, Scotty McCreery dropped in popularity (from first place to second place), yet he is still likely to win this competition.  Why?  At the moment, Country-music fans are keeping him out of the bottom three, and that's all they need to do.  During the final few weeks, and especially in the finale week, Country-music fans are likely to use networking (radio shows, concerts, TV shows, and more) to get Country-music fans — including those who do not watch the show — to vote for Scotty McCreery.  These votes, combined with many other viewers who prefer Scotty McCreery over James Durbin, will easily tip the balance in Scotty McCreery's favor.

Also consider that Scotty McCreery is much more than just a good Country-music singer, as evidenced by his performance of an Elvis Presley song.

Top 7 results

No surprise this week.  With twice as many males, vote splitting among males makes them more vulnerable.

Top 7 commentary

If Wednesday's VoteFair poll results — including first-choice-only votes — are representative of actual voting, either Jacob Lusk or Stefano Langone will be eliminated next.

Top 8 results

In previous weeks Paul McDonald was getting votes from people who follow the recommendations of the "vote for the worst" website.  This effect concentrates votes, which counteracts the vote-splitting effect.  As happened in previous seasons, after a very-popular singer was eliminated, those voters stop supporting that "worst" choice.  It happened again as a result of top-rated Pia's very-early elimination.  As a clarification, the "vote for the worst" site does not choose the worst contestant.  (They used to do it that way, but after one or two weeks that contestant would get eliminated, so now they choose mid-level singers who are weak singers but strong entertainers.)  This week's bottom three was realistic because there is now a clearer gap between the best singers and the weakest singers.

Top 9 shocking results; why Pia Toscano was eliminated early

What accounts for this week's shocking results?  Here are the reasons why third-most-popular Pia Toscano was unfairly eliminated:

• First and foremost, American Idol uses an unfair voting method; they simply count "first-choice" votes and fail to collect information about secondary choices.  The result is another dramatic reminder that the person with the fewest "votes" is not necessarily the least popular!
• Two weeks ago, when Casey Abrams ended up at the bottom, we got shocking results because voters neglected their secondary choices.  This week voters over-reacted and focused too much attention on supporting vulnerable singers, without also supporting the best singers, particularly Pia Toscano.
• American Idol does not share voting results about the most popular singers.  This leaves voters without the information they need to compensate for the unfair "first-choice-only" voting method.  Expressed in voting terms, voters are attempting to do "strategic voting" without knowing how other voters are voting.  Waiting for someone to end up in the bottom three is like watching for icebergs in the fog; by the time you know, it may be too late.

There are likely to be claims that Pia's lack of warmth, or her emotional restraint, or her lack of performance skill, or some other characteristic is to blame for her early departure.  Although that accounts for why she is not quite as popular as Scotty McCreery and James Durbin, that explanation fails to account for the real cause of her early elimination, which is American Idol's flawed voting method.

Was it a mistake for the judges to use the "judges save" on Casey Abrams, and not have it available this week?  No.  Consider that "saves" wouldn't be needed if American Idol used a fair voting method.  Also, needing to have the judges involved so much is further evidence that the voting method is flawed.

Did we, the voters, make a mistake in our voting?  It's not possible for voters to ensure fair results when the voting method is flawed.  Yes, "strategic voting" involves risks (as demonstrated by the elimination of Pia Toscano), but unfairness is also very likely when there is no strategic voting (as happened when Casey Abrams ended up at the bottom).  Until 1-2-3 ballots are used, strategic voting is necessary to get fair results, yet difficult to do when there is no information about how others are voting.

With Pia Toscano out of the contest, the finale is likely to be between Scotty McCreery and James Durbin, and Scotty is likely to win.

What can be done to prevent this (the very early elimination of a top singer) from happening again?  American Idol needs to allow online voters to also cast 1-2-3 ballots!  That would allow pairwise comparisons that reveal which singer is "pairwise" really less preferred than each of the other singers.  Specifically, a "pairwise" comparison would have revealed that Pia Toscano is more popular than Jacob Lusk, and that Pia Toscano is more popular than Stefano Langone, and which of the remaining two, Jacob Lusk or Stefano Langone is actually the least popular (assuming that these three really were the least popular).

Hopefully this event will be a wake-up call for what's going on in American politics.  The same voting unfairness (namely using only first-choice votes) happens in primary elections, but the unfairness goes unnoticed because the winner of a primary election is always from the correct party.  Financially backed special interests have learned how to use vote splitting in both primary elections so that, regardless of which party wins the general election, the result is the election of a "special-interest puppet" instead of a "problem-solving leader."  Tragically, after a few decades of "special-interest puppets" being in power, and special interests getting what they want, the result is our now-collapsing economy.  (Wise strategies have been neglected, while shortsighted "tug-of-war" contests have been given excessive funding and attention.)  Solutions to the problems we face are waiting for us.  To adopt them, we, the vast majority of voters (of both parties, plus those of us who don't like any of the parties) must insist on using 1-2-3 ballots and pairwise counting, first in small organizations and non-political places like American Idol, and then in local elections, and eventually in electing problem-solving leaders to Congress.

Top 9 commentary

Paul McDonald and Jacob Lusk have been losing appeal, and along with Stefano Langone, are vulnerable this week.  Vote splitting among the more-numerous males (six, compared to three females), combined with a better-than-usual performance from Haley Reinhart, further makes these three guys vulnerable.

Now that Casey Abrams is out of the bottom three, it's natural to wonder, "What happened to put him at the very bottom last week?"  Apparently lots of voters assumed that other voters were giving him the votes he needed to stay.  This is what happens when voters focus on their one or two favorites and do not also vote for their secondary choices.  It wouldn't happen if official voting (and government elections) used 1-2-3 ballots (the kind used here at VoteFair.org) instead of single-mark ballots.

When former American Idol contestant Constantine Maroulis performs Thursday night, keep in mind that, as explained below, he got so many phone votes that there was a phone-system failure and it caused him to get the fewest official votes, so he was eliminated prematurely.  (That was before the "judge's save" was created to deal with such surprises.)

Once again, a "thank you" to Ed at www.IdolHeadEd.com for sending voters to this VoteFair poll and making the results more accurate.  Check out his site if you want more American Idol commentary.

Top 11, second week, results

The week's VoteFair poll results, with 1,279 ballots (after removing a few ballot-stuffing attempts), almost exactly matched the official voting results!  In the VoteFair results, Naima Adedapo was least popular and Thia Megia was next to the bottom, and these two were eliminated.  Paul McDonald, who was in the bottom three, was one spot up in the VoteFair rankings, tied with Jacob Lusk.

Rankings shifted a bit this week.  In particular, Haley Reinhart, who was tenth two weeks ago, did better than usual, without which she probably would have been eliminated instead of Thia Megia.

Earlier in the week Thia Megia appeared to be more popular than next-to-last, but remember that out-of-country voters can vote in this VoteFair poll whereas, apparently, only votes from within the U.S. count in the official results.  Also, many votes in this poll were cast after Wednesday's performances, and her performance was not a strong one.

Top 11, second week, commentary

Based on recent results, Lauren Alaina is a possible surprise elimination.  If you like her, but not as your first choice, and your first choice is one of the popular singers — namely James Durbin, Pia Toscano, Scotty McCreery, Casey Abrams, or Thia Megia — then you can also cast some votes for Lauren Alaina as a ("strategic voting") way to keep her from being eliminated too early.  (She is not popular enough to win.)

Besides this possible surprise elimination, the singers Paul McDonald, Jacob Lusk, Haley Reinhart, Stefano Langone, and Naima Adedapo are all vulnerable.

Last week the judges said to "Vote for your favorite", and the results were unfair.  For increased fairness, also vote for your secondary favorites, especially if they are not getting lots of first-choice votes in this VoteFair poll (as indicated by the graph length in the "Traditional vote count (for comparison)" column).  This strategy helps to reduce the unfair effects of vote splitting.

As the contest evolves, it becomes clearer that Scotty McCreery and either Pia Toscano or James Durbin are likely to be in the finale.  The reason for Scotty McCreery being likely to be in the finale is that country-music fans are well-networked and will be concentrating their votes on him, and they can outvote pop-music fans who are split as to who they prefer.  Note that in past years apparently the phone numbers for the country singer have been posted and announced to country-music fans, so that country-music fans can cast their huge number of votes without watching the shows (prior to the finale).

Top 11, first week, results — dramatic vote splitting!

What a dramatic example of vote splitting!  In hindsight we can look at this week's poll results (after removing lots of ballot-stuffing attempts) and see that Casey Abrams did not have very many more first-choice votes than six other singers who are at the bottom in first-choice votes.  Yet he is the fourth-most popular singer!  Once again this difference demonstrates that first-choice votes (alone) are not sufficient to correctly identify the least-popular choice!

Lately, each week's early out-of-country votes for Thia Megia, plus ballot-stuffing attempts, have thrown off the early results in this VoteFair poll, but after the fans of other singers have voted here, and duplicate votes have been removed, the results would have shown that everyone except Pia Toscano, James Durbin, and Scotty McCreery were vulnerable.

The fact that Thia Megia was in the bottom three suggests that the official vote counters are taking into account that the online votes — which presumably include lots of votes from the Philippines and other non-U.S. countries — do not match the phone-based votes (which are from within the U.S.).

The judges were wise to use the judge's save on Casey Abrams, but now they cannot rescue any other victims of vote splitting.  As a result, some unfair eliminations are likely in future weeks.

The "where" of ballot-stuffing attempts

Interestingly, this week's ballot-stuffing attempts (multiple votes from the same voter) came — first and foremost — from the Philippines.  Secondarily they came from states that are known for unfair election results, namely Florida, Texas, California, New York, Pennsylvania, and Ohio.  (Additional ballot-stuffing attempts came from Georgia, Connecticut, Washington (state), Colorado, Kansas, and New Hampshire.)

Top 11 commentary

Lots of votes in this poll (including repeat votes that have been removed) are from the Philippines, and we don't know how the official vote-counters are dealing with out-of-country votes, so don't assume that Thia Megia being the most popular in this poll means that she will actually get the most official votes.  What it does reveal is the importance of networking, where fans telling other fans (of the same singer) to vote here (and in the official results) is a powerful influence.

With so many choices (12), vote splitting continues to be likely.

Top 12 results

Once again we got fair results.  Karen Rodriguez was the least popular singer — if we take into account that bottom-ranked Naima Adedapo has noticeably more first-choice votes than Karen.

Why did one of the judges even consider using the "judge's save"?  It should be reserved to rescue someone worthy (a possible winner) from a "surprise" elimination.

Why a weekly poll?

In response to a reader who asks why this poll does not start after the singers have sung, here are some reasons the VoteFair poll is done on a weekly basis:

• One of the purposes of this poll (besides teaching people how voting should be done) is to predict "surprise" eliminations.  Predictions require advance information.  As a reminder, you can do these predictions by comparing popularity rankings (as determined using 1-2-3 ballots and VoteFair Ranking calculations) with first-choice-only preferences (which is what the official vote-counters use).  Specifically, in the results column labeled "Traditional vote count (for comparison)," look for a singer whose bar length is significantly shorter than the bar lengths of the singers near the bottom; that singer (or those singers) are more popular, but do not have enough fans who are voting for them as their top choice.
• Although performances affect popularity and voting, most voters (both here and in the official voting) tend to have the same favorites from week to week.  Yes, there are exceptions, yet the rarity of such exceptions underscores this tendency.
• Immediately after the performances we are busy casting official votes, rather than voting in American Idol polls.
• Before the official voting starts, it's important to know how others will be voting, and who is most vulnerable.  This allows strategic voting, which is an unfortunate necessity when official voting uses single-mark ballots.  (This issue is related to why primary elections were added before "general" elections, namely to prevent vote-splitting surprises that used to be common in general elections.)

Online voting: it removes phone-system bottlenecks, and provides useful new info

To clarify, this year's choice to allow online voting is a good one.  It takes the load off phone lines so that phone-line bottlenecks no longer cause all the contestants to get about the same number of phone votes.  That's been a major unfairness (that gave an advantage to text-message voters) for many years.

Online voting now allows the official vote counters to learn each online voter's relative preference for the singers.  The number of times the voter votes for each contestant reveals a limited way of ranking the contestants.  This newly available information can be used to do pairwise comparisons.  This means they can verify that each contestant near the bottom is (pairwise) preferred over the person being eliminated by a majority of online voters.   This is a simplified version of what's done in this VoteFair poll.

Top 12 commentary

Based on recent results, Karen Rodriguez and Haley Reinhart are the most vulnerable.

Thia Megia has risen in first-choice votes this week, apparently due to her well-networked fans telling each other to vote here.  If they are equally well-networked in promoting her in official voting, she will do quite well.  As often happens in cases like this, after she rose to the top of these popularity results, she drifted back down because the fans of other singers increased their voting here.

Online voting from outside the U.S.

In previous seasons, fans outside the United States could not vote.  Now, through online voting, they can vote.  (If the official vote counters reject votes from outside the U.S. — which can be tracked either through Facebook info or through the IP address — foreign voters can go through a proxy service.)  As long as the number of foreign votes is not too high, this shouldn't be a significant unfairness.  And presumably the winners will be touring outside the U.S., and selling music outside the U.S., so the show's producers may not be concerned about this outside influence.

Top 13 results

Fair results once again!

Top 13 commentary

(The Top-13 VoteFair poll was split into females and males because 13 choices exceeds this website's software limit of 12 choices.)

Based on recent VoteFair ranking results, this week the most likely elimination is Ashthon Jones.  Next most likely is Karen Rodriguez, which would be a surprise because she is more popular than Haley Reinhart and Naima Adedapo who have more first-choice fans.

Among the guys, the most popular choices are Casey Abrams and Paul McDonald.

Top 24 results

We got fair results!  The five most popular males and the five most popular females — according to this week's VoteFair poll after removing repeat (more-than-twice-this-week) voters — won the top 10 spots.

Allowing online voting improved fairness because it bypassed the phone-based bottleneck — which for many years has been identified here as unfair.

Allowing the judges to pick wild-card contestants was another wise decision because it could have been used to compensate for vote-splitting effects.  This time there was no dramatic vote splitting, so the judges were able to choose marginal contestants (who are likely to be eliminated soon).

Top 24 commentary

Males:  If the results here are representative of official-voting preferences, Brett Loewenstern is not as popular, but his personality makes him more memorable than some others, so he may squeak into the top 5 males.  The other four likely vote-based finalists are Casey Abrams, Scotty McCreery, Jacob Lusk, and James Durbin.

Females:  If the (preliminary) results here are representative of official-voting preferences, Naima Adedapo is not too popular among most voters, but because of vote concentration (the cousin of vote splitting) she may squeak into the top 5 females.  The other four likely vote-based finalists are Pia Toscano, Lauren Alaina, Thia Megia, and maybe Haley Reinhart.

 

---

Like these American Idol polls?  Spread the word!  Also you are invited to voice your opinion — about VoteFair ranking (not about contestants) — on the Testimonials page.