DialIdol moved to its off-season home today and was down a spell while that happened.

Forums were archived as a part of this move.
 

Predictions were too close to call - but you know you were ignoring all my warnings about margin of error - right? Well I wasn't! I fully expected (and wanted (and predicted)) Adam to win. I was prepared for all the DialIdol haters and their mean-spirited emails - really. I was as shocked as you and I not only visit the only site in the world that must have been suggesting a Kris win - I run it! That result was unexpected and the only word I can think to describe it is - unbelievable.

Congratulation Kris!
 

---

As for YourPredictions, mattfriske won the season among those having made every prediction - congratulations Matt! I did not make every prediction (oops!) and got this week wrong too!
 

---

Many people emailed me that the GeoPredictions page wasn't working correctly. I've looked and looked and looked again and it is working correctly. I think the confusion comes from the default view of GeoPredictions showing the state color as the predicted loser - not the predicted winner. You have to use the drop-down to show the predicted winner. If you were confused - double-check and let me know if I am crazy.
 

Predictions are too close to call this week! This is the first time this has ever happened for an American Idol finale in the history of DialIdol.com. No matter what conclusions you draw from DialIdol's predictions - think again!

Only 1.1% separate Kris and Adam - this is too close to call no matter how you calculate margin of error and no matter how many directions you apply it.

Kris seems to have that inconsequential, 1.1% lead over Adam and he held that lead pretty much the whole night. DialIdol measured the busy signal every four minutes last night. What was interesting was that Kris seemed to consistently beat Adam by that slim margin for the first two hours of voting - as opposed to them alternating victories. This is a sign that the data you see is accurate but - again - there is no way to call it accurate when only 1.1% separates them!

Adam seemed to fair better on the West coast winning all three Pacific states and edging Kris more often than not. Traditionally these Pacific states have a hard time over-riding what happened in the rest of the country but with scores this close it is simply too close to call!

This should make for one heck of a finale!

Predictions were too close to call and so there could be no wrong predictions this week though I'm sure you'll have your own opinions about that, which of course you are entitled too.

Only nine tenths of one percent separated Kris and Danny which is incredibly close - too close to be meaningful even though this week it worked in DialIdol's favor. :)

---

As for YourPredictions... 24 people tied for 100% this week. Among those having made all predictions - mattfriske remains in the top spot and probably has mathematically won the season since he remains 11% ahead of beth.

For the first time in a few weeks things seemed to go smoothly Tuesday night with consistent data streaming in the whole night. Also consistently, Allison battled Kris for the bottom spot and sure enough, Allison ended up at the bottom of DialIdol's predictions chart - and American Idol's results.

As far as DialIdol was concerned - though Allison consistently came in last - Kris was consistently just a tad ahead of her and he himself was in just about as much danger as Allison. Adam and Danny were notably ahead of them both - both overall and consistently throughout the night.
 

---

As for YourPredictions... A whopping 10 people tied for 100% this week with 51 people picking Allison to go home correctly! Among those having made all predictions - mattfriske remains in the top spot and probably has mathematically won the season since he is 11% ahead of beth.

Predictions were too close to call and so there could be no wrong predictions this week though I'm sure you'll have your own opinions about that, which of course you are entitled too.

Interestingly Ryan Seacrest said to Jamie Foxx that Adam and Matt were the bottom two (approximately 47 minutes into the show). This is unusual for American Idol because usually they do not even hint that the person sent back to safety came in third last. This means that Danny and Allison finished either first or second, Kris finished third, Adam finished fourth and of course Matt finished last.
 

---

As for YourPredictions... Quixotecot won this week with a 91.11 score. Among those having made all predictions - mattfriske remains into the top spot.

Predictions were too close to call and so there could be no wrong predictions this week though I'm sure you'll have your own opinions about that, which of course you are entitled too.

Interestingly Ryan Seacrest said to Jamie Foxx that Adam and Matt were the bottom two (approximately 47 minutes into the show). This is unusual for American Idol because usually they do not even hint that the person sent back to safety came in third last. This means that Danny and Allison finished either first or second, Kris finished third, Adam finished fourth and of course Matt finished last.
 

---

As for YourPredictions... Quixotecot won this week with a 91.11 score. Among those having made all predictions - mattfriske remains into the top spot.

Predictions were too close to call and so there could be no wrong predictions this week though I'm sure you'll have your own opinions about that, which of course you are entitled too.

Lil Rounds got voted off despite being at the top of DialIdol's chart... Margin of error obviously came into play, but if you read my posts yesterday you saw that there were oddities and warnings relating to Lil's prediction as early as one hour into the vote on Tuesday night. In my judgment most of the discrepancy would have been resolved if Lil Rounds dialing volume via DialIdol even came close to that of the other contestants. Instead she got just 35% of the calls that the next lowest contestant got. That is hugely different and difficult to overcome the massively different sample sizes.
 

---

As for YourPredictions... babyvfan, pws53069, Turr and lindzhun74 all got 100% this week. Among those having made all predictions - mattfriske moves into the top spot.

Make no mistake about it - the predictions are simply too close to call this week. No one is predicted to be flat-out safe for the first time this season therefore DialIdol can't be wrong because it really isn't making any prediction.
 

---

I ran through a simulation of last nights vote - filtering out the dialers I had found last night who seemed to be getting a disproportionately high number of busy signals for Lil (even during the 9pm hour where filtering wasn't in place yet) and the resulting prediction still had Lil solidly in the second row on the chart though teetering on the brink of the third row where Danny was. Everything else was unchanged order wise. Also the predictions were still too close too call - much closer in fact. Here is what it looked like when I filtered these dialers out.

I also ran the original formula from the start of the season (with the filter on the Lil dialers) and there wasn't much change there.

And here is the original method without any filtering.

Finally, here is last nights method (the current method) without the filtering I put in place (at 10pm); interestingly Lil Rounds is actually conclusively predicted to be safe.

In all cases Kris and Allison were always on the bottom two rows and Lil never placed lower than the second row which I found interesting if not surprising.

It doesn't change the fact that DialIdol's official prediction looks like this.

Or the fact that their predictions are all simply too close to call. That will make for an even more interesting night tonight - I can't wait!

It looks like there is some weird stuff going on with Lil Rounds prediction this week. I did some analysis during the first hour of voting and found three of 24 people dialing for her with busy percents way out of whack. Two of those three were calling from the same area code. I filtered those dialers out at about 10pm but the "damage" was done. In a way I believe this has more to do with a disproportionate number of callers for her compared to the rest of the contestants. She only got 40% of the calls that the next highest total. This makes her data very unreliable and probably over-inflated. If there are going to be any surprises they will probably come from her. I believe she is ranked higher on the chart than she the data suggests she should be.
 

DialIdol will be supporting the judges save this week. The vote from the first week with seven contestants will be known as week 7.1 and the second week of seven contestants will be known as week 7.2 - the current week. It also works out that week 7.1 was going to send one contestant home and week 7.2 is sending two contestants home.

Assuming American Idol is resetting the vote count this week, scores will start from scratch building on week 7.2, leaving scores from week 7.1 intact. If American Idol decides to carry votes over from last week (an assumption DialIdol is not making) then scores will build on 7.1 and week 7.2 will be deleted. If American Idol doesn't clarify this by the time the vote begins predictions will be built on week 7.2 which assumes the votes have been reset to zero.

The client application used to dial will NOT know anything about weeks 7.1 and 7.2. The client application will see one instance of week seven. The "Call Log" section of the client application will NOT reset your call totals to zero - the call log section will combine weeks 7.1 and 7.2 into one week seven.

I did not get a chance to look over last week's data relating to Danny/Lil as noted below but then again the weekend is not over yet. :)
 

Anoop, Lil and Matt were all predicted to be in danger of being in the bottom three and in danger of being voted off. Lil and Anoop were in the bottom three and Matt was the bottom vote getter so all three of those predictions were accurate. I understand that Lil was high up on the chart but margin of error is very real and proving such this week.

As I monitored the vote last night I saw some very weird stuff going on with Danny and Lil's predictions. Lil seemed to be getting the most busy signals from 9:30 to 10:30 (Eastern) while Danny was getting the fewest busy signals. I'm not quite sure what happened here but I'll look into it.
 

---

The oddities with Lil and Danny caused me to forget to submit my own predictions in YourPredictions - I completely forgot until the show started and it was too late. rib777 and MGFAN8675 didn't forget and got the highest scores for the week. mattfriske moves into first among those having made every predictions this season.
 

---

As for the judges save, American Idol hasn't made it clear whether the votes carry over from this week to next or the start from scratch. DialIdol is prepared for both but it is assuming that American Idol is starting from scratch.
 

This marks the first week this season where the person listed on the bottom of DialIdol's predictions didn't leave the show. That trivial fact doesn't change DialIdol's accuracy for the week or season though.

Anoop, Lil and Scott were all predicted to be in danger of being in the bottom three and in danger of being voted off. Lil and Anoop were in the bottom three and Scott was voted off so all three of those predictions were accurate.
 

---

As for YourPredictions : Turr, bigdave7, pappy3c and turner.rikki got 100% accuracy this week - congrats to them!

mattfrise moved into the lead for those who've made every prediction this season. I barely edged metallicat1 for the number two slot event though metallicat1 has made more correct predictions overall.

Kris Allen's prediction had me analyzing the data a bit more than usual. The fact that he performed second and performed poorly (according to the judges) didn't seem to matter. Kris' prediction started weak and got a little stronger as the night went on. This directly plays into the new formula I implemented beginning with the top 13. That formula put more emphasis on the periods of time when there were a lot of busy signals (the beginning of the vote).  Historically this proves to be more reliable but it is hard to ignore the fact that 30 minutes into the voting Kris began outperforming a few other contestants and Scott MacIntyre took over as the bottom dweller. The formula puts more emphasis on those first 30 minutes - thus why Kris is lower.

For giggles I ran the last nights data through the old formula (which doesn't emphasize based on busy signals) and here is what the prediction was.

Notice that Scott gets bumped down from third last to last and Kris gets bumped up a notch. Anoop is out of the bottom three altogether and Matt joins Scott & Kris in the bottom three. Interesting eh? This week will say a lot about the formula adjustment I made but no matter what happens I think this formula is sticking around because, it made last weeks data more accurate - among many other weeks.

My point here is that Scott or Matt are names that could surprise you tonight. In reality though it seems to come down to Scott or Kris being sent home and that is how I picked them in YourPredictions.

Just to be clear though, DialIdol is officially predicting  that anyone except Adam is in danger of being voted off tonight. The reason for that is because of margin of error and the close nature of their scores. Also, anyone except Adam could be in the bottom three. Adam is the only one DialIdol predicts is absolutely safe.

Allison Iraheta was predicted to be in danger of being in the bottom three and she was so that counts as being accurate. Anoop Desai was also predicted to be in danger of being in the bottom three and he was so that counts as being accurate. Megan Joy Corkrey also was predicted to be in danger of being in the bottom three and of course in danger of being voted off and she was so that counts as being right.

Incidentally, Anoop definitively came in second last which means Allison came in third last. Their predicted ranges spanned these slots so that was accurate as well, albeit not picture perfect. Still - not a bad way to rebound eh?
 

---

As far as YourPredictions go - only two people got 100% accuracy this week - Gokeyrules and babyvfan - congratulations! UVSinger knocked me off the top spot among those who have made every prediction this season and I fell all the way to fourth.

I ran this seasons data through a bunch of different formulas - focusing especially on the week with ten contestants left and while there were scenarios that got Matt a slot or two lower it came at the expense of Megan Joy getting bumped into the bottom slot. In the end I believe I found a tweak that improves accuracy though this tweak wouldn't have made Matt's prediction any better. Scott/Megan Joy would have swapped spots putting Scott more firmly in the bottom three though Megan would still have been there. The same formula when applied to other weeks would also improve accuracy.

Specifically the tweak is that intervals are now four minutes long instead of three. The idea is that there is more data per interval now and thus more accurate. Also, the client application submits its data every 165 seconds - this change will help ensure all dialers are getting their data in all intervals. This change will result in higher margins of error - but not by much and the trade off in "row accuracy" (the row a contestants prediction sits in relating to their actual result) makes it worth it.

All in all I really can't see anything in the data that would have made Matt's prediction more accurate. Quite honestly all I could see were reasons Megan Joy should have gone home. Who knows - maybe I am not looking in the right places? :)

 

Matt Giraud was four tenths of one percent away from being predicted to be in the bottom three but he was NOT predicted to be in the bottom three. We know that he was in fact in the bottom three, which means that DialIdol was wrong about that prediction, the first incorrect prediction of the year.

Scott MacIntyre was predicted to be in the bottom three (and in danger of being voted off) and was actually in the bottom three so that does count as being accurate.

Michael Sarver was predicted to be in danger of being voted off (and of course in the bottom three) and he was so that too counts as being accurate.

Incidentally, this week margin of error did in fact apply in both directions so there are no margin of error “asterisks” this week and it will remain that way the remainder of the season.
 

---

In terms of YourPredictions, beth posted the best score for this week joining eight others in getting all three predictions correct. Among those who have made every prediction this season I am the leader of the pack. :)

 

Yes - Michael Sarver was predicted to be in the bottom three - and since he was that counts as being accurate. Allison Irraheta was also predicted to be in the bottom three and she was so that is also accurate. Stunningly, Alexis Grace was predicted to be in danger of being voted off - and even more stunningly - she was, which makes DialIdol three for three this week.

Personally I found Alexis' prediction very hard to swallow and I spent most of yesterday sweating it. As the results show neared I analyzed data from GeoPredictions and convinced myself Michael Sarver was going to throw a monkey wrench into things. In the end he did - but not enough.

Also, if I were the judges I would have saved Alexis - period. I am pretty sure they wanted to save her but I am also pretty sure they want to maintain some drama for future results shows and apparently that matters more.

Finally, just to let everyone know - margin of error will be switching back to applying in both directions next week - as it should be (in my humble opinion).

---

As for YourPredictions - Coit won the week - getting all three predictions correct and having the most precision. My predictions on DialIdol were never updated to reflect my discoveries about Michael Sarver so I got his prediction wrong but I do rank first among those who've made predictions every week this season. :)

Everyone's combined predictions got Alexis wrong ... and right. She and Michael were predicted to be the bottom two but their averages prevented them from reaching slot eleven - an oddity.

Looking at the predictions might surprise a few people this week - myself included. The biggest surprise might be how specific DialIdol is being this week. The specificity is a result of the formula change I made to DialIdol which went into effect last week. That formula is expected to be more accurate - but here is the thing : it is my opinion that margin of error should apply in both directions but it is currently only applying in one direction. I changed MOE to only apply in one direction last year as a stop-gap measure aimed to reduce the number of ties. I should have changed it back to applying in both directions when I implemented the new formula but I choose not to. That may very well play into this weeks accuracy which will definitely chart the course for the remainder of the season.

If MOE was applying in both directions everyone except Danny, Anoop and Lil would be in danger of being voted off and would show up as yellow.

Regardless - DialIdol will be graded on MOE applying in one direction as it is seen now and has been seen since last night and last year even.

I looked at the data pretty closely and even ran it through the old formula. The data suggests both Allison and Alexis are in trouble but Megan might surprise you. She got absolutely no West coast support which dropped her rank signficiantly. It really is a dead heat between the three of them but weak showing by Scott and Michael might put them in danger too. Scott & Michael's strength was clearly superior to Alexis and Allison - but not by so much that I would count them out. Hence the reason for this update.

What a week! Phone number changes, rule changes and still the show was predictable! :) DialIdol predicted that Jasmine Murray was in danger of being voted off and she was so that was accurate. Jorge Nunez was also predicted to be in danger of being voted off and he was so that was accurate. Heck most everyone even predicted that American Idol would be adding some sort of immunity save - so that was accurate too! :)

For the record - I love the judges save and have wanted that kind of rule for a few years now. I wish they took it a step further and gave the judges one save per judge per season rather than giving all the judges one unanimous save per year. Nevertheless - bravo - one of the few changes about this season I actually like!

One thing that confused me - I believe Ryan said the rule is effective through the top five. He also said it could have been used to save Chris Daughtry. One problem - Chris Daughtry was eliminated when there were four contestants left. Four contestants is less than five so in theory the rule would not have been allowed to save him.

---

As for YourPredictions - UVSinger remains the best predictor for the entire season but seven others who joined the fun later are currently ahead of UVSinger. I rank eleventh - after having rightfully felt skeptical about Allison's prediction and opting to pick Jasmine to get the boot - like many of you. Collectively - everyone's predictions were more accurate than DialIdol - firmly planting Jorge & Jasmine in the bottom two rows.

American Idol confirmed that a different phone number will be used to vote for one of the contestants tonight because they do not own 1-866-436-5713 or 1-866-436-5700 for that matter. DialIdol.com originally brought this problem to light on Saturday. Many other popular AI sites caught on yesterday like votefortheworst.com and mjsbigblog.com and I am beginning to think that it took that trail for American Idol to figure out they didn't own a 13th number.

There are some important things to note! First and most important...

DialIdol will be 100% compatible with this new phone number no matter what it is - even if it is revealed at the bitter end of the show. Phone numbers can change all night long and DialIdol will react to it - the client application checks for new numbers each time it begins dialing whether that is via the auto-start feature or when you click the DialIdol button.

I also can't help but note my substantial concern with this solution. Adding a unique thirteenth number, as opposed to changing ALL the numbers is unfair to that contestant AND more likely to result in accidental calls to the 1-866-436-5713 number. I only hope cooler heads prevail and the powers that be at American Idol realize this and change ALL the numbers for tonight only.
 

---

On another note - DialIdol will be assuming that two contestants will be voted off tonight - while this may or may not happen (I suspect it will not) - I want DialIdol to error on the side of communicating that the second lowest vote getter (and contestants within range of second last place) are in danger of being voted off as well. This is subject to change based on feedback and other factors.

Yes, DialIdol is supporting Dancing with the Stars this season! http://stars.dialidol.com will reflect that during the show when they reveal how they are going to cope with losing two contestants at the last minute. At this point it looks like Holly Madison and Melissa Rycroft will be the replacements but their partners are still fuzzy.

1-866-436-5713 and 1-866-436-5700 continue to connect you to non-American Idol phone conference. (Call at your own risk - parents - parental guidance is suggested). It is unclear how this will effect voting on Tuesday but I believe it will be significant no matter what.

This news somewhat overshadows the new formula DialIdol will begin using this week, which is detailed in Saturday March 7, 2009's entry. I expect that this formula will prove to be the most accurate yet!

For quite some time I have wanted to make two general improvements to DialIdol's Official Predictions. The first thing I wanted to do was place more emphasis on periods of time where there were more busy signals (right after the show ends) than on those were there were very few busy signals (right before voting ends). The second thing I wanted to do was prevent those periods of time with very few busy signals from diluting the scores into ties. This week I am introducing a new formula for calculating DialIdol's Official Predictions that will achieve both of those things.

Of course achieving those two things must not get in the way of making accurate predictions. This new formula has been tested against data from this season and previous seasons and has proven to be more accurate than the old formula.

All in all this new formula should result in more meaningful predictions. Less ties, higher scores and more accurate predictions.

---

I intend on detailing the changes further at a later date but...

It used to be that a person's score was :

(B / T) * Z

where B = Busy signals, T = Total calls placed and Z = time zone weight

Now it is :

(B / T) * (Z * M)

where B = Busy signals, T = Total calls placed, Z = time zone weight and M = the highest busy percentage for all contestants for that period of time (interval).

(Actually DialIdol doesn't and hasn't in quite some time used B / T; instead DialIdol averages everyone's busy percentages to calculate a contestants busy percent.)

On the back end DialIdol then takes all the scores for all the intervals, adds them up then divides by the sum of total weight (Z * M).

 

Again this week DialIdol wasn't visually perfect but it still achieved 100% accuracy. Scott MacIntyre was absolutely predicted to be safe and he was so that was accurate. Lil Rounds was also absolutely predicted to be safe and she was so that was accurate. Jorge Nuñez's prediction was right when you factor in margin of error and DialIdol does.

With respect to Jorge this week - if you checked out my personal predictions prior to the show you saw that I was predicting Jorge to be #2. I did this based on DIHardGraphs which showed Jorge with a lot more busy signals early on which tends to matter more. His support died in the second hour of voting when everyone's support died. During the second hour there are so few busy signals that it really isn't as reliable as when there are many. I'm working on a fix for this but in the mean time margin of error and DIHardGraphs will have to do.

---

In terms of YourPredictions - things are tightening up; jomagi still hasn't missed a prediction but didn't predict with 100% precision and the score suffered. victortperry is now #1 but victortperry wasn't around for the first week (with 36) so his data isn't complete. For what it is worth - I rank #12 having missed Kris Allen's prediction last week.

All in all things are tightening up and it is turning out to be a lot of fun!

Well this week wasn't visually perfect but DialIdol still achieved 100% accuracy. Adam Lambert was absolutely predicted to be safe and he was so that was accurate. (Incidentally was there ANY suspense at the end when it came down to Adam/Nick? Kris versus Nick would have been more suspenseful to me...) Anyway... Allison Iraheta was also absolutely predicted to be safe and she was so that was accurate. Kris Allen's prediction was right when you factor in margin of error and you must factor that in - I told you not to get used to it last week! :)
 

---

In terms of YourPredictions - there is only one person who has gotten every prediction accurate this season - congratulations jomagi!

In terms of the seasonal rankings - I will be working on a way to filter out scores that better reflect those who have participated for most of or all of the season.

Perhaps you have seen the infomercial on TV for Magic Jack, a USB device that you plug into your computer and then plug a standard phone into that to receive a phone line. It's like $40 the first year and $20 thereafter - dirt cheap. Well this little piece of Voice Over IP heaven works with DialIdol. An analog modem is still required but a land line is not. This would be a good fit for cell phone only homes that also have a broadband Internet connection. If you fall into that category and want to use DialIdol you can consider a Magic Jack.

To make the Magic Jack work with DialIdol plug a standard phone cord into the Magic Jack and run the other end of it into the input port on an analog modem - it should be that simple. If your modem has two jacks - one input and one output you can also run another phone card from the output jack on the modem to a phone (portable preferably) and use the portable phone anywhere in your house to make calls whenever you aren't using DialIdol to vote for your favorite contestants.

Let it be known that I do not work for Magic Jack nor was I approached by Magic Jack in any capacity. I am not receiving any compensation for this note nor do I expect any and if any comes I will disclose it here on this page. I am not really endorsing the Magic Jack but I will say that what they advertise on TV is for the most part true. It is that easy and the call quality is decent - similar to that of a cell phone. I do not think it can be your only telephone connection but if you want to use it to save on long distance or preserve cell phone minutes I think it is a great way to do that.

Despite only a few requests I decided that DialIdol must better support this top 36 format American Idol has adopted. As such, the Predictions page has been modified to better support this format. From now on when you look at the girls/guys tables the color they are shown in is the absolute, best-case  color that represents their predicted fate. If a contestant shows up in yellow here - that means they have a shot at making it - either because they are tied for first in their gender group or because they are the next highest vote getter.

Also, a new table has been added - underneath the gender tables there will be another table that shows the contestants in the running for the next highest vote getter. This is to better help you understand what the predictions look like when they are combined. To the left of that table is some verbiage that will hopefully explain the page better.

---

I have added some code to validate YourPredictions better which will result in a fairer competition.

Happy Birthday Dad!

---

So it seems like DialIdol started the season off on the right foot. Stephen Fowler was absolutely predicted to be voted off and he was so that was accurate. Danny Gokey was absolutely predicted to advance to the top twelve and he did so that too was accurate. The bonus this week is that margin of error didn't matter much (don't get used to it) and Alexis Grace and Michael Sarver made it into the top twelve as well, their predictions weren't as absolute though so you have to take those with a grain of salt.

---

In terms of YourPredictions, almost one hundred people participated and those will be interested to know that the three predictions of the contestants who made it into the top twelve will count (Alexis, Danny and Michael). Also - Stephen Fowler's prediction will count too. Perhaps this shouldn't be (in fact it was not my original intent) but the way DialIdol calculates DialIdol's official predictions requires that his prediction count - otherwise the official predictions page would reflect an incorrect prediction for Stephen. I might look into changing this but right now this is how it has to be. All the rest of the predictions won't count as right or wrong, however all predictions including those that you aren't graded on count towards your weekly precision and will impact your score. The more precise you were in your prediction - the better your score but of course you do have to get the four predictions that counted right - or some combination of them.

Overall nine people got 100% this week and are all tied for first place; check out the list of the nine on YourPredictions.

Incidentally - DialIdol's official predictions are also scored on YourPredictions and they came in ninety fourth out of ninety five. This is in large part because most everyone was more precise than DialIdol. I don't think I ever mentioned it before now but I do not expect DialIdol to be at the top of YourPredictions - largely because of the wider range. It will be interesting though to see how the season goes.

There is also a special user called Everyone that is the average of Everyone's predictions and that got all four predictions right but because it was not as precise - it ranks near the middle of the pack.

If you see my note below you'll understand why I feel the need to clarify DialIdol's predictions.

The predictions page perfectly displays who DialIdol thinks might be the top girl/guy and you can use the Key on that page to help you understand that.

The next highest vote getter isn't so clear so let me help to clarify. Basically DialIdol is predicting that no girl will be the next highest vote getter and only one girl will be moved to the next round by the vote from last night. Also, Stephen Fowler isn't in range of being the next highest vote getter.

Let's Recap

• Danny Gokey is predicted to be the top guy.
• The girls are so close that really any of them could be the top vote getter but the reality is Alexis Grace has the best chance - probably significantly so.
• No girl is predicted to be the next highest vote getter because none of them are in range of Michael Sarver.
• The battle for the next highest vote getter is between : Michael, Anoop, Ricky & Brent. For those of us from Cleveland - we will be welcoming Stephen Fowler back because DialIdol is predicting he has no chance of advancing via the vote.

The judges pick is irrelevant as far as DialIdol's accuracy is concerned - let's hope American Idol makes it clear who that pick is and doesn't pull some hocus pocus of - these four Idols are moving on.

Incidentally is has not been clear to me when the judges will make their wildcard choices... It could happen each week and make for a more interesting results show or it could happen after all 36 have performed which makes more sense since the judges could then pull more than one contestant per group of 12 - this would be my preference but I get the sense that they are going to pick one per week... Why? Because they need to fill three spots and coincidentally there are three results shows. If they waited until the last week you'd think there'd be a wild card pick per judge - four spots to fill.

---

There was a bug that prevented people from making their predictions for a brief period of time today. That has been resolved.

---

There was a bug preventing forum post counts from increasing; that has been resolved but since the fix was pretty major I am worried there may be new bugs so be on the lookout and let me know if you find any.

---

I believe DialIdol is having trouble sending some emails so if you are registering for the forums and don't receive your activation email - email me and I will auto-approve you. This may be particularly troublesome for hotmail.com addresses.

Happy Valentine's Day!

So let's call these first three weeks of the season "beta" weeks in terms of DialIdol - ok? Here's why...

First of all I am introducing a major new feature this year - YourPredictions. This will allow you to logon to DialIdol.com and make your own predictions! In the spirit of DialIdol.com you will predict a range that you think each contestant might fall into each week and track your accuracy throughout the season. You'll also be able to see how you rank versus others and DialIdol itself. This major new feature has been tested as much as I can test it but I am sure there are bugs abound so please bare with me while we work through them.

---

Registering on the forums is going to be more critical to DialIdol.com now because that same registration is used to make YourPredictions so it is good that the forums are now back open for new registrations. I finally upgraded the forum engine to a newer version with bug fixes and hopefully I've closed a few security vulnerabilities that were allowing hackers to spam the forums.

---

This format American Idol has choose for the first three weeks of the season is sure to cause some confusion - especially on DialIdol. American Idol will advance the top girl/guy vote getter and the next highest vote getter. That means each week for the first three weeks, three contestants will make it into the top 12. America will pick 9 of the top 12 contestants and the judges will pick three more from the rejected 36 to join them in the top 12. OK - now that we all understand how it will work - let's me explain where the confusion will come in on DialIdol. The top girl/guy will be portrayed very clearly on DialIdol.com as usual - none of this confusion hocus-pocus applies to the top girl/guy vote getter. The confusion comes in for the next highest vote getter. This contestant will be shown in second place on either the girl/guy side but it will not be clear whether the girl or guy is in the lead. Additionally it is possible that someone shown in second place really has no chance of advancing.

For example in the following fictional prediction scenario it is obvious that Amy/Adam are predicted to be safe. However it will not be obvious that DialIdol is predicting that Bob/Chuck stand absolutely no chance of advancing to the next round. When you look at the Predictions page you will see that Bob/Chuck are predicted to be in danger of being voted off - even though they are actually absolutely predicted to be voted off (they will be in yellow - not red).

DI Rank	Girls	Score	DI Rank	Guys	Score
1	Amy	65	1	Adam	55
2-3	Beth	54	2-3	Bob	35
2-3	Cindy	53	2-3	Chuck	34
4-6	Diane	20	4-6	David	20

I thought long and hard about how to portray DialIdol's predictions and made the decision that DialIdol would not be tweaked to portray this conundrum. There are several reasons for this but the biggest is that this is only an issue during the vote. After the vote I will analyze the scores and I will determine whether the next highest vote getter is a girl or guy and I will change DialIdol to show that five contestants of that gender will be going home instead of four which will correct it from then on.

There are other reasons including likelihood of encountering this scenario, the fact that no matter what it will be confusing and the time that would have been required to code this.

I may find time to work on this aspect or tweak the way I communicate it but for now this is the way it is.

---

DialIdol works with Magic Jack but you still need an analog modem. To use DialIdol with Magic Jack run a standard phone cable from Magic Jack to an analog modem in your computer (then optionally run another phone line from your modem to your phone). DialIdol will then use your physical modem to dial contestants of your choice over the Magic Jack IP system.

This all assumes the Magic Jack phone network permits the calls to go through. So far my testing indicates they do but it has never been tested during a real vote.

---

You won't care about this unless I screwed up but I have tweaked DialIdol's engine to make it a little lighter on its feet - this could lead to weird things outside of YourPredictions - anywhere on the site - so be on the lookout for weird stuff.

DialIdol is moving to its in-season home today so there was an outage.

The web site is still running on code that doesn't include the new feature I have been working diligently on the past month. That code will be posted this weekend.

The forums will reopen for new registrants this weekend too and they will become more critical to DialIdol too - albeit indirectly.

So it is freezing cold, actually well below freezing, here in Cleveland, Ohio - the perfect weather for working on DialIdol! ...and I am hard at work too! I'm adding a major new feature this year - the ability for YOU to make your own predictions! I believe a few of you have suggested this over the years and I finally figured out how cool this could be! Most of the code to enter/view your own (and other people's) predictions is already done and now I am working on the statistical analysis type stuff for it. You know - like who are the best predictors? The worst? Where do you rank compared to everyone else? Cumulatively are the people's predictions accurate? And of course - how do all of these predictions compare to DialIdol's predictions and the actual results!

I also think I am going to dial down the support DialIdol has for the first three weeks of voting (prior to the top twelve). If you aren't familiar, American Idol starts voting with 36 contestants. The first week the first batch of twelve perform and the top guy, top girl and next highest vote getter move on. This continues for three weeks which means the American Idol viewing audience picks nine of the top twelve. The judges pick the remaining three "wild card" contestants from those that didn't make it through by viewers choice. I think I have explained this sufficiently and it is relatively simple - except when it comes to predictions. In order to visually represent every possibility with this kind of format I'd potentially have to confuse the heck out of you! All things considered I may not support the format completely and instead clarify the predictions with an update on this page after the voting ends. For the most part what I have in mind will work for the casual visitor and the more advanced visitor will absolutely be able to figure out who is in danger. The place I will probably dial stuff down at is the coloring of contestants names. The top guy/girl will be clear but the next highest vote getter won't be. I am working towards this goal unless I can figure out how to simply explain all possibilities which I haven't given up on but as you can see by the length of this paragraph - this is no small chore!

Anyway - that is what's news! I really don't know why I bothered to write this as this code base won't be posted to the production Internet site for weeks yet and will likely be diluted by other updates when it does hit! ...actually I do know why - I am excited for this coming season and to see DialIdol's new features at work!

Well it is getting to be that time again! American Idol's eighth season begins next week on January 13! DialIdol won't kick into high gear until voting begins on February 17. The format for voting during the early rounds has changed this year and DialIdol will be covering them. Right now I am working on getting that all squared away.

The forums remain closed to new guests due to some spam issues that occurred last summer. I will be working to address that soon too.

Stay tuned!

• Updates from 2008
• Updates from 2007
• Updates from 2006
• Updates from 2005