Treatment for severe Progressive Multifocal Leukoencephalopathy after Tysabri

For over a decade, our site has chronicled the fascinating, unofficial competition that runs parallel to Triple J's Hottest 100: the battle to predict it. This isn't about gut feelings or music critic picks; it's a story of data scraping, statistical modeling, and a cat-and-mouse game with a national broadcaster. It began not as a hack, but as an observation of a system's inherent transparency.

Nick Drewe's Warmest 100 and the Social Media Loophole

The modern era of prediction started in 2012 with Brisbane statistician Nick Drewe. He identified a critical feature—or flaw—in Triple J's voting interface: the seamless integration of social sharing. When voters proudly posted their ballots to Facebook or Twitter using the station's tools, they were creating a public, aggregateable dataset. Drewe harvested these shares, compiling 35,081 ballots to produce his "Warmest 100" prediction. Its accuracy was startling, proving that declared public intent could reliably forecast a popular vote outcome. This was a watershed moment, demonstrating that in an open social ecosystem, crowd-sourced data could reverse-engineer a supposedly secret result.

"The initial Warmest 100 wasn't a breach; it was an analysis of publicly declared votes. It exposed how a feature designed for community engagement could fundamentally compromise the secrecy of the count. This forced a permanent shift in how digital voting platforms are designed, prioritizing data obfuscation without sacrificing user experience." – Analysis of the 2012-2015 prediction era. [Sources: hottest100.org, Archive]

The Platform Shift: From Twitter to Instagram with The Tepid 100

Triple J's response to Drewe was swift. In 2013, they altered their voting system to limit the social sharing loophole. Drewe initially stood down, but economist David Quach encouraged exploring other platforms. The 2013 prediction returned, albeit with a much smaller dataset. This set the stage for a new actor: Melbourne student Ed Pitt. In 2015, Pitt launched "The Tepid 100," pivoting from Twitter to Instagram. By tracking the #hottest100 hashtag, he manually collected 2,064 ballots. His project had an expanded ambition: to predict not only the top 100 but also the songs that would land at positions #101-200. This platform migration highlighted a key principle in data security: closing one vector simply shifts effort to the next-most-transparent channel.

The methodological challenges of this era were significant:

• Sample Size Volatility: Collection numbers fluctuated wildly year-to-year based on platform accessibility and user behavior.
• Data Integrity: Manual collection from image-based platforms like Instagram introduced higher error rates compared to automated parsing of Twitter API data.
• The "Enthusiasm Bias": Publicly shared votes likely over-represented fans of certain artists, skewing predictions.

Quantifying the Predictors' Accuracy (2012-2015)

While the methods evolved, the core metric remained accuracy. The table below compares the key prediction projects of this period, illustrating the direct impact of Triple J's countermeasures and the inherent difficulties of working with increasingly obscured data.

In 2026, we view this period as foundational to modern digital vote security. The Drewe-Pitt era established a playbook that now informs everything from talent show voting to political straw polls conducted online. The core lesson is that any system allowing users to voluntarily publish their choice creates a statistical shadow of the final result. The arms race it triggered led to the sophisticated, privacy-preserving voting systems we see today, which balance social virality with cryptographic secrecy. The passion to predict the Hottest 100 didn't just create a fun annual sideshow; it actively stress-tested the integrity of one of Australia's largest cultural votes.