As Election Day loomed in 2012, traffic at the New York Times website spiked, as is normal during moments of national importance. But this time, something was different. A wildly disproportionate fraction of this traffic—more than 70 percent by some reports—was visiting a single location in the sprawling domain. It wasn’t a front-page breaking news story, and it wasn’t commentary from one of the paper’s Pulitzer Prize-winning columnists; it was instead a blog run by a baseball stats geek turned election forecaster named Nate Silver. Less than a year later, ESPN and ABC News lured Silver away from the Times (which tried to retain him by promising a staff of up to a dozen writers) in a major deal that would give Silver’s operation a role in everything from sports to weather to network news segments to, improbably enough, Academy Awards telecasts. Though there’s debate about the methodological rigor of Silver’s hand-tuned models, there are few who deny that in 2012 this thirty-five-year-old data whiz was a winner in our economy.

— Deep Work: Rules for Focused Success in a Distracted World, Cal Newport, pub. Piatkus 2016

I don't understand. What models? What does the author mean by rigor? Does he control the broadcast?


The key to understanding this is to understand model or, in particular, a statistical model.

A model describes a set of data or observations. To take a trivially easy example, if we have a set of data {1,2,3,4,5}, one model is to say that the average is 3.

This might lead a naive modeller to predict that the next observation is likely to be around 3.

A shrewder modeller would see that the data are increasing steadily and predict that the likely next value is 6.

This is the story of Silver. He had a skill in data modelling that was not bound by simple or conventional statistical methods. His models are described as hand-tuned because he developed his own techniques rather than just using conventional models. The conventional models are described in the text as having methodological rigor, meaning that they were standard techniques applicable to many circumstances. The mention of debate means that Silver was criticised for not sticking to the conventionally approved methods. These methods, having been formally and carefully developed, would have been regarded as having intellectual or mathematical rigor.

Rigor = strict precision : EXACTNESS logical rigor

”Tentatively one might suggest that what characterizes science is the rigor of its methodology”

Merriam Webster

But Silver was able to see patterns in data that the conventional methods and others did not see, giving his models a predictive force better than others. And with better prediction comes better economic advantage.

  • Or they were suggesting that maybe he just got lucky, but no-one was willing to state their suspicions quite so blatantly in public. Private, individually developed, non-peer-reviewed models will always be open to challenges questioning their rigor. This is because the vast majority will be seriously flawed. But when one does work, the thing to do is figure out why, not fall further behind every day. Rigor isn't the measure of merit of a model. It is a measure of how well we understand the problem.
    – Phil Sweet
    Jan 1 at 14:42
  • @PhilSweet Yes I agree maybe he did get lucky, but his success means that your hypothesis is difficult to test. Its a bit like successful fund managers: maybe they are good or maybe they are lucky, but who can tell?
    – Anton
    Jan 1 at 16:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.