I worked with the Decision Desk HQ crew Monday night to help assemble Iowa caucus data*. I can tell you that the Microsoft application deployed to handle both parties caucuses, to use a technical term, sucked.
It was buggy, inconsistent (different people had different updates of the same data using the same application), prone to crashing, browser killing (we tried Firefox, Chrome and Safari and all of them at some point stopped responding), and surprisingly difficult to use. For a company with Microsoft’s pedigree, I expected more.
The biggest problem I saw was that the Democrats didn’t report vote counts. I now know that this is by design.
Discrepancies can occur in official elections, and caucuses are not even official election events run by the secretary of state’s office, noted Dennis Goldford, a Drake University professor who closely studies the Iowa caucuses.
“The caucus system isn’t built to bear the weight placed on it,” he said. “There aren’t even paper ballots (in the Democratic caucuses) to use for a recount in case something doesn’t add up.”
By contrast, the Republican caucuses were a dream, although hampered by the same terrible Microsoft app. In fact, the Des Moines Register results page crashed throughout the night using Microsoft’s “API” (application program interface) data feed. DD used manual input and beat the automated AP results most of the night. That’s how bad the Microsoft “solution” was.
We’ve got two competing stories as to who really won Iowa on the Democrat side (I’m not even going to touch Trump’s call for a do-over). Quin Hillyer at National Review reported that Sanders really won, because Hillary’s six coin flip wins (the odds of that being 63:1) for county delegates gave her an edge in “delegate equivalents.”
But I’ve seen nobody point out what should be obvious: If “delegate equivalents” are supposed to fairly represent the actual voting behavior of caucus attendees, even down to narrow fractions, then in terms of actual votes, Sanders slightly defeated Clinton. The final count of delegates to the
state[county] convention (aside from the seven won by Martin O’Malley) was Clinton 699, Sanders 695. But by actual voter decisions, the count was Sanders 695, Clinton 693, and six ties.
The Iowa Democratic Party does not have comprehensive records on how many coin flips/games of chance were held Monday evening. However, they do have partial records.
More than half of the 1,681 Democratic caucuses held Monday night used a new Microsoft reporting app. Of those, there were exactly seven county delegates determined by coin flip. The remaining precincts did not use the Microsoft app, and instead used traditional phone-line reporting to transmit results. In these precincts, there no are records of how many coin flips occurred. There’s only anecdotal information on these precincts.
In most endeavors of life, such a situation would be known as (what the military calls) a “Charlie-Foxtrot.” Or a fustercluck.
But barring the opaque mysteries of Democratic Party religious rites and oaths, Clinton and Sanders ended in a tie, which means Sanders won. Hillary can declare victory all she wants, but she didn’t win. The big loser in all this is Microsoft, which should be barred for all eternity from handling election data.
I like the caucus system because it forces voters to take a bit more effort and participate in actual candidate selection, versus a drive-by early vote on the way to Wendy’s. But if caucuses are like the one held Monday in Iowa, let’s stick to primaries and let the state and county election officials, who know their ass from a hole in the ground (generally at least), deal with the results.
*Decision Desk called the GOP race at 9:41 p.m., nearly an hour before anyone else. Check them out for the NH primary. I volunteer for them, so my gushing is really based on believing in the project, not financial gain.