【Dear Utol (2025): Totoy Bayo Episode 38】

On Sunday night,Dear Utol (2025): Totoy Bayo Episode 38 the Patriots completed the most improbable comeback in Super Bowl history.

Which is to say: It wasn't impossible. Sure, a few pundits predicted the Patriots were done, but they're pundits. The number crunchers never ruled the Patriots out—even though, to many, it seemed like they did:

This Tweet is currently unavailable. It might be loading or has been removed.

Forecasting and data visualizations grow trendier by the year, as the pace of the business speeds up, and more people crave more information in more easily digestible forms. The New York Times, ESPN (and its ownership of FiveThirtyEight), Vox, and plenty other outlets have built teams dedicated to taking in data, and trying to communicate its meaning to the general public, as quickly (and digestibly) as they can.

But as illustrative as these kinds of graphs can be, they're also misleading—or at least, misleading for a large swath of the public that has a fundamental misunderstanding of what this kind of data represents. It's a problem that's existed for decades—weather forecasters know this better than most—that's now beginning to poison the public's faith in data analysis more broadly.

Monday night provided the newest example that, while talented at crunching numbers and producing eye-pleasing results, these outlets are coming up short in adequately communicating just what they're finding. And they might be doing more harm than good in the rush to put those findings out.

To err (on probability) is human

People are generally okay at math. That said, they're terrible at numeracy, or: understanding numbers in context. It's a problem that speaks to the bigger problem of how people understand the kind of data work that's become trendy, said Joe Steinhardt, an assistant professor at Michigan State University, where he studies risk perception. And, he went on, it's starting to have an effect on how people view expertise of all kinds.

"The more information we're getting, people are having the wrong takeaway," he said. "I think they're losing faith in math and science. We're seeing that argument or we're seeing [people] just throwing their hands in the air."

The main issue has to do with confusion over the difference between probability and prediction. Pundits tend to predict something will happen. Forecasters, on the other hand, look backwards at data to understand current circumstances—say, being down 28-3 in the third quarter of a football game—and then, the probabilityof various things happening.

These forecasts aren't predictions. They are guides. Based on the data available, a team will comeback from that kind of deficit a certain percentage of the time. It's not that it will or won't; it's just about likelihood. The Patriots comeback was unlikely, which is what the in-game probability showed. That does not mean it couldn't ever happen.

Even people that understand this concept can find themselves "tricked" by it, in the sense that they see the forecast as a guide to what's inevitable in the future. Something that has a three percent likelihood isn't that uncommon. The odds of hitting a single number in roulette are 2.6 percent.

This nuance is easily lost. The Super Bowl graph above immediately led people to compare the Patriots comeback to Donald Trump's election victory over Hillary Clinton. The comparison's an easy one; Clinton was favored by just about all the big statistical models. These models, the logic seems to go, are then useless or at the very least unreliable if the less likely event happened.

This Tweet is currently unavailable. It might be loading or has been removed.

The data meeks shall inherit the earth

It's worth exploring how we got here.

In the early 2000s, professional sports helped introduce the world of statistics began to mainstream into the mainstream. In 2003, Michael Lewis published Moneyball, which chronicled how the Oakland Athletics front office used a different approach to analyzing baseball statistics to find undervalued players. The book helped kickstart a broader fascination with how professional sports teams had begun using advanced forms of data analysis to build teams that continues into the present day.

Five years later, Nate Silver would first gain attention for his use of demographics data to predict the outcomes of elections. In 2012, he would introduce the major turning point for the growing world of internet-native, data-based prognostication. Silver's model correctly predicted the presidential outcome for all 50 states.

Suddenly, data wasn't just informative. It could predict the future—or at least that seemed to be the growing misconception about what Silver was actually doing.

Meanwhile, the growth of digital media was met with demand for journalism that looked good on the internet.

"It just so happens that we are now in a media age in which web media in particular can cater towards a hyper-educated audience and find an audience for it, if that's their choice," said Chadwick Matlin, senior editor for culture at FiveThirtyEight. "Also, the web is becoming an increasingly visual medium and one in which charts and data can help offer analysis and ways in which to make content more engaging for readers."

Matlin noted that this doesn't necessarily mean that readers are getting the message. The average person is quite bad at performing basic risk analysis or understanding probability. This is a relatively new discovery, one which helped form the basis of rise of behavioral economics, which looks out the big ramifications for people making bad decisions. Not coincidentally, this is a concept that Michael Lewis is helping popularize through a book.

"Numbers are very good at tricking us and making us believe in certainty when actually the numbers themselves aren't expressing certainty," Matlin said.

Shoot the messenger

Four years after Silver helped make data journalism mainstream, the mystique's been shattered. Silver (and plenty of other lesser-name brand data journalists) drew the ire of nearly everyone following the election of Trump, who'd been widely forecast as the underdog. To the uninformed, these forecasters were snake oil salesman who might have gotten lucky. To the informed, they were charlatans who hadn't adequately communicated to the public what their work truly meant.

This, despite FiveThirtyEighthaving given Trump one of the better chances of winning.

This Tweet is currently unavailable. It might be loading or has been removed.

This Tweet is currently unavailable. It might be loading or has been removed.

The mistrust that's developed around data journalism isn't a standalone issue. Journalism as a whole is suffering through a crisis of confidence that's fueling the acceptance of questionable information, and the very real problem of very fake news. It comes despite the advent of powerful computing technologies that provide us with a massive leap in the evolution of analysis of past events.

Sadly, that analysis doesn't mean much if it's seen as a way to predict the future.

"They either lose faith in the science, which again has never been more accurate, or they lose faith in the idea of even using numbers in the first place," Steinhard said. "Then they go back to things where they like feel it in their bones."