An Incredibly Serious Investigation
The case of the overstuffed hoagie bracket
Everything was going according to plan — until it wasn’t.
We created the Italian Hoagie Bracket to honor a favorite local lunch option, highlight a few businesses, and have a bit of fun. In response, thousands of hoagie eaters flocked to answer the question that has consumed Philadelphians for decades: Who serves the best Italian hoagie?
We watched the votes roll in. But then we saw some voting irregularities. Without a Hoagie Electoral Commission, we responded the only way we could: by spending way too much time investigating it.
As someone on Reddit said of the Hoagie Bracket: “This is the kind of value-adding, investigative local journalism that we need more of. I am not kidding.”
Be careful what you wish for.
During the first day, the delis received winning bracket submissions at a predictable rate.
However, a couple of days in, a deli from the middle of the pack saw multiple voting surges.
On the fourth day, another massive batch of votes gave it a clear lead over the competition. This pattern continued throughout the voting.
Our initial working theory was that the deli was somehow incentivizing customers to vote. For instance, offering a discount if a customer completed a bracket in-store. But, these batches came outside of operating hours.
Something didn’t add up.
We called a meeting of our crack team to a top secret, underground bunker that houses The Inquirer's Department of Hoagie Bracketology.
The Inquirer’s Data & Democracy Reporter, Jonathan Lai, analyzed the hoagie bracket combinations — there are 32,768 possible ways to fill it out — and when those brackets were submitted.
In one of several tests, he looked at the winner of each bracket and the winner of the next bracket submitted. Put simply: If you roll a 16-sided die, it shouldn’t keep coming up 4. And that’s true for 15 of the 16 delis: When a deli wins a bracket, it usually doesn’t win the next one.
There’s just one deli that stands way, way out. When it wins a bracket, it also wins the next one a whopping 60% of the time. The die is loaded.
Mark Loomis, our Director of Analytics, found the next clue. “Def a few sus [visitors],” he said, pressing the giant red ENHANCE button.
He discovered that a single iOS device had visited the page 742 times. Another had registered 323. The timestamps for these votes were irregularly spaced, indicating that all 1,000-plus brackets — the brackets that were skewing our data — were submitted by a person, not a bot.
Somebody, somewhere, was trying to fix the hoagie bracket.
Now at this point you — like a very vocal minority of our investigative team — may be thinking this has all the makings of a six-to-eight episode crime podcast. But there’s one glaring problem: There’s no crime. This is a made up contest to compare sandwiches. The first rule of hoagie bracket is there are no rules.
We placed no restrictions on how often people could vote and established no penalties for shops trying to influence the outcome. For example, Fink’s blatantly engaged in a free-hoagies-for-votes campaign that would make Al Schmidt blush, while Middle Child took to posting A HUMBLE MESSAGE IN ALL CAPS on Instagram.
We’re not going to name the deli because why would we? No rules were broken.
Also, there’s no proof that these votes came from the store itself. When we called the deli owner to ask if this was his handiwork, he seemed surprised. “I honestly don’t have a clue... I guess it’s a good thing people are voting for us,” he said. He also seemed incapable of such an advanced vote-rigging operation, having said that he entrusts his social media to someone who “does all my hashtags.”
Maybe these bulk votes were the work of that hashtag-doer, or maybe a private citizen just REALLY loves this deli’s hoagies. Frankly, there’s something endearingly Philly about spending hours manually entering brackets over and over and over again. Bots are for New Yorkers. Here, we just hit refresh obsessively.
Still, this contest was designed to determine the favorite hoagie across our readership, and one person casting this many votes is enough to skew the results. Late in the process, we discovered another bulk vote, about 400 entries, for a different deli. In each case, we removed the excess votes to better reflect the will of the people.
It took a team of devoted journalists, some intricate calculations, and more time than our bosses would have liked, but now you can sleep (and eat) soundly, knowing our hoagie election was secure.
Staff Contributors
- Design and development: Sam Morris
- Data Reporting: Jonathan Lai and Mark Loomis
- Editing: Kate Dailey