The Delegate Fight: 2016

[Erc]

March 26 Democratic Primaries

Alaska, Hawaii, and Washington qualify as a "regional cluster," and receive a 15% delegate bonus.

Alaska (D)

Overview
20 Delegates (0.42% of total)
Closed Caucus
10 "District"
4 At-Large
2 PLEO At-Large
4 Superdelegates

Details

The 10 "District" delegates are apportioned based on the statewide caucus vote.  The caucuses also elect delegates to the State Convention (May 14-15), which elect the 4 At-Large and the 2 Pledged PLEO delegates.

Superdelegates

Clinton (1): Kim Metcalfe

Sanders (1): Vice Chair Larry Murakami

Uncommitted (2): Chair Casey Steinau, Ian Olson

Useful Links
The Green Papers: AK
AK Delegate Selection Plan

Hawaii (D)

Overview
35 Delegates (0.73% of total)
Closed Caucus
6 At-Large
3 PLEO At-Large
16 by CD
10 Superdelegates

Details

6 At-Large and 3 PLEO delegates will be apportioned based on the statewide caucus result.  8 delegates per CD will be apportioned based on the result in each CD.

Superdelegates

Clinton (6): Sen. Brian Schatz, Sen. Mazie Hirono, Rep. Mark Takai, Jadine Nielsen, Lt. Gov. Shan Tsutsui, Russell Okata

Sanders (2): Rep. Tulsi Gabbard, Doug Pyle

Uncommitted (2): Gov. David Ige, Stephanie Ohigashi (Clinton 2008)

Useful Links
The Green Papers: HI-D
HI Delegate Selection Plan

Washington (D)

Overview
118 Delegates (2.47% of total)
Half-Open Caucus
22 At-Large
12 PLEO At-Large
67 by CD
17 Superdelegates

Details

The CD delegates are allocated based on the caucus vote in each CD: 12 in CD 7; 7 in CDs 1,2,6,9; 6 in CDs 3,8,10; 5 in CD 5; 4 in CD 4. The 22 At-Large and 12 PLEO delegates are allocated proportionally based on the number of CD delegates each candidate won.

Superdelegates

Clinton (10): Gov. Jay Inslee, Sen. Maria Cantwell, Sen. Patty Murray, Rep. Jim McDermott, Rep. Derek Kilmer, Rep. Rick Larsen, Rep. Dennis Heck, Rep. Susan DelBene, Rep. Adam Smith, Rion Ramirez

Uncommitted (7): Ed Cote, Juanita Luiz, Sharon Mast, David McDonald, Chair Jaxon Ravens, Valerie Brady Rongey, Lona Wilbur

Useful Links
The Green Papers: WA-D
WA Delegate Selection Plan

[Erc]

March 12 Results

In addition to the Guam and Marianas Islands results noted earlier...

Wyoming elected 9 Cruz, 1 Trump, 1 Rubio, and 1 Uncommitted delegate at its County Conventions, as well as delegates to the state convention.  We may be able to forecast the results of the state convention based on these results in the next day or so.

D.C. narrowly voted for Rubio over Kasich, with Cruz and Trump falling below threshold.  Rubio wins 10 delegates (and his 2nd Rule 40 state), Kasich wins 9.  

D.C. is the last contest of this cycle that is required to be proportional.  States have the option to use Winner-Take-All starting on Tuesday, and 4 of the 5 states voting that day use it at least on some level.

In Iowa, the results are in:

Clinton 704 - Sanders 700 - O'Malley 1 - Uncommitted 1

Sanders and company did well, it seems, but Clinton now has clinched a majority in Iowa (by 1 delegate) and I am awarding her the final delegate in Iowa.

[cinyc]

D.C. is the last contest of this cycle that is required to be proportional.  States have the option to use Winner-Take-All starting on Tuesday, and 4 of the 5 states voting that day use it at least on some level.

The Commonwealth of the Northern Mariana Islands is also voting on Tuesday - and the winner of the Republican presidential preference poll takes all 9 delegates there.  So it's 5 out of the 6 jurisdictions that use WTA on some level.  And the CNMI will be the first to do so, given the time difference.

[An American Tail: Fubart Goes West]

March 26 Democratic Primaries

Alaska, Hawaii, and Washington qualify as a "regional cluster," and receive a 15% delegate bonus.

So, just for being in the "same region," these three states all get 15% more delegates than they would otherwise?

[ElectionsGuy]

March 26 Democratic Primaries

Alaska, Hawaii, and Washington qualify as a "regional cluster," and receive a 15% delegate bonus.

So, just for being in the "same region," these three states all get 15% more delegates than they would otherwise?

Pretty stupid, but it should help Bernie.

[Shadows]

March 26 Democratic Primaries

Alaska, Hawaii, and Washington qualify as a "regional cluster," and receive a 15% delegate bonus.

So, just for being in the "same region," these three states all get 15% more delegates than they would otherwise?

Pretty stupid, but it should help Bernie.

The appropriations of delegates often breaks with population, many states Clinton won big, shdn't have had so many delegates. This is what the system is.

You lose some, you win some.

[Erc]

Iowa County Conventions

Mostly, the results lined up with expectations based on the caucus results, but there were some exceptions:

Projected: Clinton 699 - Sanders 690 - Unallocated 17
Actual: Clinton 704 - Sanders 700 - O'Malley 1 - Uncommitted 1
Change: Clinton +5, Sanders +10, O'Malley +1, Uncommitted +1

Black Hawk County

Projected: Sanders 36 - Clinton 32 - Unallocated 1
Actual: Sanders 35 - Clinton 34
Change: Clinton +2, Sanders -1

Clinton did much better than expectations; not only did O'Malley & Uncommitted swing to her, so did some of Sanders' support.

Bremer County

Projected: Sanders 5 - Clinton 6 - Unallocated 1
Actual: Sanders 6 - Clinton 6
Change: Sanders +1

O'Malley chose Sanders here.

Butler County

Projected: Sanders 3 - Clinton 2 - Unallocated 1
Actual: Sanders 4 - Clinton 2
Change: Sanders +1

Sanders won the coin flip.

Cerro Gordo County

Projected: Clinton 11 - Sanders 11
Actual: Clinton 12 - Sanders 10
Change: Clinton +1, Sanders -1

Clinton gained support from Sanders here.

Chickasaw County

Projected: Clinton 4 - Sanders 2
Actual: Clinton 3 - Sanders 3
Change: Clinton -1, Sanders +1

Sanders gained support from Clinton here.

Crawford County

Projected: Clinton 2 - Sanders 2 - Unallocated 1
Actual: Clinton 3 - Sanders 2
Change: Clinton +1

O'Malley chose Clinton here.

Decatur County

Projected: Sanders 1 - Clinton 1 - Unallocated 1
Actual: Sanders 2 - Clinton 1
Change: Sanders +1

O'Malley and Uncommitted sided with Sanders and won the coin toss.

Delaware County

Projected: 3 Sanders - 4 Clinton - 1 Unallocated
Actual: 4 Sanders - 4 Clinton
Change: Sanders +1

O'Malley sided with Sanders.

Dickinson County

Projected: 3 Clinton - 4 Sanders
Actual: 4 Clinton - 3 Sanders
Change: Clinton +1, Sanders -1

Sanders support bled to Clinton.

Fremont County

Projected: Sanders 1 - Clinton 1 - Unallocated 1
Actual: Sanders 2 - Clinton 1
Change: Sanders +1

O'Malley sided with Sanders.

Henry County

Projected: Sanders 3 - Clinton 3 - Unallocated 1
Actual: Clinton 4 - Sanders 3
Change: Clinton +1

Clinton wins the coin flip.

Jackson County

Projected: Sanders 4 - Clinton 4 - Unallocated 1
Actual: Clinton 5 - Sanders 4
Change: Clinton +1

O'Malley sided with Clinton (or Clinton won a coin flip).

Jasper County

Projected: Sanders 8 - Clinton 9 - Unallocated
Actual: Clinton 10 - Sanders 8
Change: Clinton +1

O'Malley did not break hard enough for Sanders for him to pick up the last delegate.

Jefferson County

Projected: Clinton 2 - Sanders 7
Actual: Clinton 3 - Sanders 6
Change: Clinton +1, Sanders -1

Sanders support bled to Clinton.

Johnson County

Projected: Clinton 37 - Sanders 55
Actual: Clinton 38 - Sanders 54
Change: Clinton +1, Sanders -1

Sanders support bled to Clinton.

Keokuk County

Projected: Sanders 1 - Clinton 2 - Unallocated 1
Actual: Sanders 1 - Clinton 3
Change: Clinton +1

O'Malley sided with Clinton or she won the coin flip.

Linn County

Projected: Sanders 63 - Clinton 57 - Unallocated 1
Actual: Sanders 63 - Clinton 58
Change: Clinton +1

O'Malley did not side with Sanders.

Marion County

Projected: Sanders 6 - Clinton 6 - Unallocated 1
Actual: Sanders 6 - Clinton 7
Change: Clinton +1

O'Malley did not break hard enough for Sanders.

Mills County

Projected: Clinton 3 - Sanders 2
Actual: Clinton 2 - Sanders 1 - O'Malley 1 - Uncommitted 1
Change: Clinton -1, Sanders -1, O'Malley +1, Uncommitted +1

Clearly, the Sanders camp tried something here, and it may have worked.  The delegation going into the convention was Clinton 23 - Sanders 22.  I'm guessing the Sanders camp split into groups of 7, 7, and 8, each of them making viability and getting a delegate.  Don't know whether the Clinton camp attempted to fight them for control of one of the delegations, or whether they were taken off guard.

Monroe County

Projected: Sanders 1 - Clinton 1 - Unallocated 1
Actual: Sanders 2 - Clinton 1
Change: Sanders +1

O'Malley sided with Sanders.

Plymouth County

Projected: Clinton 3 - Sanders 4
Actual: Clinton 3 - Sanders 6
Change: Clinton +1, Sanders -1

Sanders support bled to Clinton.

Polk County

Projected: Sanders 105 - Clinton 121 - Unallocated 2
Actual: Sanders 113 - Clinton 115
Change: Sanders +8, Clinton -6

O'Malley sided with Sanders, who also bled a significant amount of support off from Clinton.

Pottawattamie County

Projected: Sanders 16 - Clinton 15
Actual: Sanders 14 - Clinton 17
Change: Sanders -2, Clinton +2

Sanders bled significant support to Clinton.

Scott County

Projected: Sanders 41 - Clinton 40 - Unallocated 1
Actual: Sanders 42 - Clinton 40
Change: Sanders +1

O'Malley did not side with Clinton.

Warren County

Projected: Sanders 10 - Clinton 12
Actual: Sanders 9 - Clinton 13
Change: Sanders -1, Clinton +1

Sanders support bled to Clinton.

Webster County

Projected: Sanders 7 - Clinton 9
Actual: Sanders 8 - Clinton 8
Change: Sanders +1, Clinton -1

Clinton support bled to Sanders.

Wright County

Projected: Sanders 2 - Clinton 3
Actual: Sanders 3 - Clinton 2
Change: Sanders +1, Clinton -1

Clinton support bled to Sanders.

Analysis to follow.

[Erc]

Iowa County Conventions: Summary

Three main sources of differences between my initial projections and the final count:

Assigning of "Unallocated" Delegates

17 delegates had unsure outcomes due to the presence of O'Malley or Uncommitted delegates (15), or due to effective ties requiring coin flips (2).

Clinton won 8 of these, while Sanders won 9 (they split the coin flips).  Not a great result for Sanders.

Enthusiasm Gap / Delegates Switching Sides

In 18 cases, there were enough no-shows or people changing their minds to flip delegates that had been projected from one side to the other.

On net, Sanders gained 2 delegates from Clinton this way.  He did really well in Polk County (Des Moines), where he picked up at least 6 delegates from Clinton, but on net bled to Clinton in the rest of the state.

Tactical Voting

As discussed above, a clever play by the Sanders camp in Mills County moved a delegate from Clinton and Sanders to Uncommitted and O'Malley (who had no initial representation).  Who actually chose the Uncommitted and O'Malley delegates is a different question, of course.

Final Results

Clinton 704 - Sanders 700 - O'Malley 1 - Uncommitted 1.

Note that if Clinton loses even 1 delegate, she is no longer guaranteed the win at the State Convention.  This flip could only come from someone switching sides or Clinton folks failing to show up.  It's entirely possible, but these should be really committed folks at this point.  As a result, on the main page I'm calling the final delegate for Clinton, though I will be marking it with an asterisk (as with the other caucus/convention states).

COIN FLIP CONSPIRACY

If Clinton had lost the coin flip in Henry County, she'd be at 703 delegates and would not have a majority.  Of course, she also lost two other coin flips today, but who's counting.

[Lief 🗽]

As discussed above, a clever play by the Sanders camp in Mills County moved a delegate from Clinton and Sanders to Uncommitted and O'Malley (who had no initial representation).  Who actually chose the Uncommitted and O'Malley delegates is a different question, of course.

Holy crap that's so sneaky. So it's probably 704-702.

[Erc]

Actually, there seems to be some controversy regarding the Polk County Convention (which Sanders did unexpectedly well at).

The initial counts for Clinton were even worse, showing Sanders beating her there despite Clinton's pretty decisive win there on caucus night.  As a result, the Clinton folks asked for a credentials check of the delegates (numbering 1200 in total!), which dragged on the process for many, many hours.  Some Sanders supporters simply left (due to other commitments) as time passed, and the final count showed Clinton ahead, though still short of what would have been expected from caucus night results.

More details can be found here.

If this is disputed going forward (which it very well could be, considering accusations directly from Sanders' Twitter that "Effort under way right now by @hillaryclinton and party allies to steal Polk County Iowa conv. election Bernie won earlier today."), the outcome of that final delegate may once again be in dispute.

For all the outcry here, I have to say Clinton was in the right to ask for credentials checking considering the surprising nature of the result.  Even if all the credentials were in order, the difference in the result means---even for the results the Sanders camp is saying is crying foul about!---at least 35 Clinton supporters switched to Sanders (or 70 Clinton supporters didn't show up).  That's not all that many out of a total of 1200 delegates, I suppose, but it's still a lot!

That's a lot of people who were committed enough to Clinton on February 1 to go to a convention on a random Saturday the next month to support them, who then decided to change their minds and vote for the other guy.  Everywhere else in the state the swings were way smaller (though of course those counties were smaller); the only county to see more than a 1-delegate swing was Pottawattamie, and that in the other direction.

Was there a concerted effort by the Sanders camp on February 1 in Des Moines to elect stealth Sanders supporters as Clinton delegates to the Polk convention?  (In the manner of Paul's myriad efforts to this effect in 2012 on the GOP side)?  I don't know, but this doesn't smell right to me either way.

EDIT: More discussion in this thread.

[Ebsy]

Actually, there seems to be some controversy regarding the Polk County Convention (which Sanders did unexpectedly well at).

The initial counts for Clinton were even worse, showing Sanders beating her there despite Clinton's pretty decisive win there on caucus night.  As a result, the Clinton folks asked for a credentials check of the delegates (numbering 1200 in total!), which dragged on the process for many, many hours.  Some Sanders supporters simply left (due to other commitments) as time passed, and the final count showed Clinton ahead, though still short of what would have been expected from caucus night results.

More details can be found here.

If this is disputed going forward (which it very well could be, considering accusations directly from Sanders' Twitter that "Effort under way right now by @hillaryclinton and party allies to steal Polk County Iowa conv. election Bernie won earlier today."), the outcome of that final delegate may once again be in dispute.

For all the outcry here, I have to say Clinton was in the right to ask for credentials checking considering the surprising nature of the result.  Even if all the credentials were in order, the difference in the result means---even for the results the Sanders camp is saying is crying foul about!---at least 35 Clinton supporters switched to Sanders (or 70 Clinton supporters didn't show up).  That's not all that many out of a total of 1200 delegates, I suppose, but it's still a lot!

That's a lot of people who were committed enough to Clinton on February 1 to go to a convention on a random Saturday the next month to support them, who then decided to change their minds and vote for the other guy.  Everywhere else in the state the swings were way smaller (though of course those counties were smaller); the only county to see more than a 1-delegate swing was Pottawattamie, and that in the other direction.

Was there a concerted effort by the Sanders camp on February 1 in Des Moines to elect stealth Sanders supporters as Clinton delegates to the Polk convention?  (In the manner of Paul's myriad efforts to this effect in 2012 on the GOP side)?  I don't know, but this doesn't smell right to me either way.

Just to clarify a few things, only about 1000 credentialed delegates showed up (not unusual at all), and some of the controversy was that alternates were tallied when they were not supposed to on the first count, leading to the number of Sanders supporters being over-represented.

[Erc]

Just to clarify a few things, only about 1000 credentialed delegates showed up (not unusual at all), and some of the controversy was that alternates were tallied when they were not supposed to on the first count, leading to the number of Sanders supporters being over-represented.

That makes a lot of sense, thanks for bringing me back from the deep end there.  We'll see how this plays out.

[Erc]

As discussed above, a clever play by the Sanders camp in Mills County moved a delegate from Clinton and Sanders to Uncommitted and O'Malley (who had no initial representation).  Who actually chose the Uncommitted and O'Malley delegates is a different question, of course.

Holy crap that's so sneaky. So it's probably 704-702.

It's something I'd thought about, but I never thought anyone would have the balls to pull off.  If the Clinton camp saw this coming (they probably didn't), they could have countered it though.

Assuming everyone showed up to the convention and no one switched sides, initial count was:

Clinton 23 - Sanders 22

Viability (15%) is 7.  The Sanders camp splits into three groups:

Clinton 23 - Sanders 8 - O'Malley 7 - Uncommitted 7

If Clinton sees this coming, she can counter it by moving her own supporters to O'Malley or Uncommitted (but not both):

Clinton 15 - Sanders 8 - O'Malley 15 - Uncommitted 7

This gets complicated quickly, but the Clinton delegates, if they played it right, might have been able to still pick 3 delegates, even though they weren't technically Clinton delegates.

Most likely they just simply got played, though.

[Ebsy]

So, pending whatever craziness happens in the wake of Polk County, Clinton leads by just as many state/district delegates as she did before the conventions. Both will probably bleed some support going into the statewide convention, so we really don't know who will get the final delegate.

[Erc]

So, pending whatever craziness happens in the wake of Polk County, Clinton leads by just as many state/district delegates as she did before the conventions. Both will probably bleed some support going into the statewide convention, so we really don't know who will get the final delegate.

The technical reason I held off calling it on the front page is we didn't know how O'Malley supporters were going to break.  Now we do, and it seems they split pretty evenly (actually in favor of Clinton in Polk, if you believe it).

Sanders could very well flip the state at the state convention, and I'll note that on the main page; but for the purposes of the count I'm going to come in line with literally every other source and give Clinton a 23rd delegate.

[Mr. Morden]

What happens with Carson's delegates when he endorses TRUMP? Do they go to TRUMP?

Nope. Candidates do not have control over their delegates in that fashion.

At the moment, it seems that his 6 delegates in Iowa and Virginia will be bound to him on the first ballot (Virginia is far less certain than Iowa).

In Nevada, he has two delegates, which he can choose to release entirely, or reallocate them 1 to Trump and 1 to Rubio.

None of these delegates are actually chosen by the Carson camp, so Carson will presumably have very little influence over any delegates he releases.

Are there any states where Rubio has delegates, where he'd lose them on the first ballot if he drops out of the race this week?  I'm thinking that he may want to "sort of drop out but not really", by ceasing to campaign but keeping his delegates in hand in order have leverage over forcing a contested convention down the road, should that opportunity arise.  If "suspending his campaign" would result in his delegates in certain states becoming free agents, then maybe he'll figure out a way to suspend his campaign without legally calling it that.

[The Other Castro]

Honestly, to me this whole caucus process seems to serve only as a confrontation arena to increase negative rhetoric and turn people with very similar political ideologies against one another. In the words of the immortal John Oliver, "How is this still a thing?"

[Mr. Morden]

In the words of the immortal John Oliver, "How is this still a thing?"

How did he manage to become immortal?

[The Other Castro]

In the words of the immortal John Oliver, "How is this still a thing?"

How did he manage to become immortal?

My bad, I meant the immortal words of John Oliver. It's been a long day.

[Mr. Morden]

Another delegate question:

Illinois (R): March 15

Overview
69 Delegates (2.79% of total)
Open Primary
15 At-Large (Winner Take All)
54 District (directly elected)

Delegate Allocation and Selection

A presidential preference poll is on the ballot; the winner of this poll wins all 15 At-Large delegates.  12 of these are chosen at the State Convention on May 22.

Additionally, voters directly vote for 3 delegates running for a slot within their CD; the delegate candidates have their Presidential preference listed on the ballot.  All current candidates have a complete slate of delegates.  The top three delegate vote-getters in each CD get their ticket punched to Cleveland.  Voters are not obliged to vote for delegates who match the candidate they voted for in the preference poll.  Often, there are some personally popular delegate candidates who can get elected on the strength of their name alone (e.g. in 2008, Dennis Hastert, a Romney delegate, was elected in his CD despite a McCain win there in the preference poll).

In the 2012 results here on Atlas:

https://uselectionatlas.org/RESULTS/state.php?year=2012&fips=17&f=1&off=0&elect=2

it says that in the 2012 GOP primary results in Illinois, there were 15 delegates "unallocated".  Is that a mistake?  Or does the delegate allocation process allow such a large number of unallocated delegates in Illinois?  (Were these district level delegates that expressed no presidential preference on the ballot?)  Would we be likely to see a repeat of that this time, or are the rules now different?

[Smash255]

Another delegate question:

Illinois (R): March 15

Overview
69 Delegates (2.79% of total)
Open Primary
15 At-Large (Winner Take All)
54 District (directly elected)

Delegate Allocation and Selection

A presidential preference poll is on the ballot; the winner of this poll wins all 15 At-Large delegates.  12 of these are chosen at the State Convention on May 22.

Additionally, voters directly vote for 3 delegates running for a slot within their CD; the delegate candidates have their Presidential preference listed on the ballot.  All current candidates have a complete slate of delegates.  The top three delegate vote-getters in each CD get their ticket punched to Cleveland.  Voters are not obliged to vote for delegates who match the candidate they voted for in the preference poll.  Often, there are some personally popular delegate candidates who can get elected on the strength of their name alone (e.g. in 2008, Dennis Hastert, a Romney delegate, was elected in his CD despite a McCain win there in the preference poll).

In the 2012 results here on Atlas:

https://uselectionatlas.org/RESULTS/state.php?year=2012&fips=17&f=1&off=0&elect=2

it says that in the 2012 GOP primary results in Illinois, there were 15 delegates "unallocated".  Is that a mistake?  Or does the delegate allocation process allow such a large number of unallocated delegates in Illinois?  (Were these district level delegates that expressed no presidential preference on the ballot?)  Would we be likely to see a repeat of that this time, or are the rules now different?

Illinois had 15 unpledged or Superdelegates in 2012. 

As per Green Papers

You must be logged in to read this quote.

[Erc]

What happens with Carson's delegates when he endorses TRUMP? Do they go to TRUMP?

Nope. Candidates do not have control over their delegates in that fashion.

At the moment, it seems that his 6 delegates in Iowa and Virginia will be bound to him on the first ballot (Virginia is far less certain than Iowa).

In Nevada, he has two delegates, which he can choose to release entirely, or reallocate them 1 to Trump and 1 to Rubio.

None of these delegates are actually chosen by the Carson camp, so Carson will presumably have very little influence over any delegates he releases.

Are there any states where Rubio has delegates, where he'd lose them on the first ballot if he drops out of the race this week?  I'm thinking that he may want to "sort of drop out but not really", by ceasing to campaign but keeping his delegates in hand in order have leverage over forcing a contested convention down the road, should that opportunity arise.  If "suspending his campaign" would result in his delegates in certain states becoming free agents, then maybe he'll figure out a way to suspend his campaign without legally calling it that.

Yep, though the language is often unclear.  Here's my earlier post on Rubio's delegates, which I will update with his more recent results soon.

They are technically only released upon Rubio suspending his campaign in Louisiana, Nevada, and Alaska, although we can only really be certain about Louisiana.  There may be a loophole in Nevada (technically they're only released if you suspend after the State Convention), but there's also a loophole to prevent a zombie campaign (they are released if you "otherwise discontinue [the] campaign").  In Alaska, the language for automatic reallocation is not very clear, but "maintaining an active campaign" appears to be the requirement.

Not sure whether that sort of weird limbo would be worth it to keep his 5 delegates in Louisiana.  In all other states, delegates are released only if he withdraws, officially releases them, or is "not a candidate" before the National Convention (which we wouldn't know technically until the first ballot).

[Erc]

Some more details on the continuing Yob legal battle.

[jimrtex]

Another delegate question:

Illinois (R): March 15

Overview
69 Delegates (2.79% of total)
Open Primary
15 At-Large (Winner Take All)
54 District (directly elected)

Delegate Allocation and Selection

A presidential preference poll is on the ballot; the winner of this poll wins all 15 At-Large delegates.  12 of these are chosen at the State Convention on May 22.

Additionally, voters directly vote for 3 delegates running for a slot within their CD; the delegate candidates have their Presidential preference listed on the ballot.  All current candidates have a complete slate of delegates.  The top three delegate vote-getters in each CD get their ticket punched to Cleveland.  Voters are not obliged to vote for delegates who match the candidate they voted for in the preference poll.  Often, there are some personally popular delegate candidates who can get elected on the strength of their name alone (e.g. in 2008, Dennis Hastert, a Romney delegate, was elected in his CD despite a McCain win there in the preference poll).

In the 2012 results here on Atlas:

https://uselectionatlas.org/RESULTS/state.php?year=2012&fips=17&f=1&off=0&elect=2

it says that in the 2012 GOP primary results in Illinois, there were 15 delegates "unallocated".  Is that a mistake?  Or does the delegate allocation process allow such a large number of unallocated delegates in Illinois?  (Were these district level delegates that expressed no presidential preference on the ballot?)  Would we be likely to see a repeat of that this time, or are the rules now different?

The rules are different. I think that this is due to a tightening of RNC rules - if you have a preference poll, you have to use it. Illinois is a bit odd in that delegate candidates appear on the ballot (along with their presidential preference), and voters vote for the delegates from their CD. They could conceivably split their vote (since three delegates are elected, each voter may cast three votes). Some voters will likely only vote for one delegate.

In 2012, two Santorum delegates won a seat because of a split vote among Romney delegate candidates. It appears that there may be some advantage to having a recognizable name. In 2012, Ethan Hastert (Dennis Hastert's son) finished quite a bit ahead of the other Romney delegate candidates. A Darrin LaHood had strong support as a Gingrich delegate. This may have cost Romney a delegate, since a Santorum candidate took the delegate slot.

In some cases, the last delegate candidate for a presidential candidate had a big drop off, in other cases the first candidate had more votes, and the other two about the same. Voters also chose the alternate delegates.

In 2012, there was a beauty contest, but it was ignored. In 2016, it will be used to pledge the statewide delegates. It appears that the delegate ballot will be down-ballot following senatorial and congressional nominations. Some voters will likely skip the delegate ballot.

If stereotypes are valid, this may hurt Trump. The Cruz campaign has probably done a better job of targeting supporters, and will make sure they understand the ballot. If Trump relies more on a mass appeal, it may be difficult to communicate the nuances.

[Erc]

For those really interested in what happened in Polk County, here are two really long articles about them:

The Polk County Democratic Convention Fiasco
Polk County Democrats Convention Turns Into A Total Disaster

Interestingly, it seems like the last round of recounts hurt Clinton, as many of her delegates were older people who had to leave for medical reasons.  It didn't seem to help that the Sanders folks happened to get the more comfortable of the two rooms.

Overall, a huge, embarrassing mess.  Credentials challenges seem possible at the State Convention, which could prove ugly.