Friday, March 21, 2014

FiveThirtyEight, data journalism, visualization, and the naysayers

It's taken the naysayers just a few days to react to the launch of Nate Silver's new FiveThirtyEight. The most vocal one has been Leon Wieseltier. It's not surprising. The guy likes to proclaim that this science and data journalism stuff has not place in his cherished turf. For some strange reason he reminds me of a certain editor at El Mundo newspaper who used to walk around the newsroom mumbling that he was looking forward to the day when "this fashionable Internet thing" would go away. This was just ten years ago.

Wieseltier is also fond of misrepresenting his targets' points. This is worrying, as he's the literary editor of The New Republic, so you should expect him to be able to extract meaning from a piece of text. But, as philosopher Daniel Dennett wrote a while ago, Wieseltier "doesn't know, and doesn't even know that he doesn't know."

Wieseltier doesn't understand (or doesn't wish to understand) Silver, either: "(He) cannot conceive of public reason except as an exercise in statistical analysis and data visualization." This is malarkey. What Silver wrote in his opening manifesto and said repeatedly is that even if numbers are a core component of his new website, he's concerned mainly about evidence. And evidence can (and should) be both quantitative and qualitative. That's explicit in his essay and book.

Wieseltier asks rhetorically "I am sorry that (Silver) finds George Will and Paul Krugman repetitious, but should they revise their beliefs so as not to bore him?" Of course not but, again, that's not Silver's point. What he wrote is that journalists should be open to change their thought based on careful analysis. We should stop being driven by anecdote and ad-hoc notions, and become more rigorous. If you think that this is something that all journalists and op-ed writers do, perhaps you've never read Thomas Friedman (have fun with this and this) or Malcolm Gladwell (this,) to name just two popular ones.

Wieseltier ends his lazy article with this question: "What is it about conviction that frightens these people?" It's not conviction that frightens us. What frightens us is convictions based on prejudices, biases, and ignorance. How can you fight against them if it's not by wielding qualitative and quantitative methods? Wieseltier doesn't say, perhaps because he may not be able to tell the difference between a median and a mean.


A different line of criticism of the new FiveThirtyEight is more interesting. It troubles me, for instance, that Silver hired Roger Pielke to write about science. This is a mistake. I also confess that I'm a bit disappointed with the stories published in the first few days. I am not the only one. We were expecting more depth in many of the stories, perhaps. But, hey, let's cut these folks some slack. It's been just a week!


  1. Excellent summary, but if you're concerned about journalistic integrity, it's probably best not to cite Joe Romm for evidence that the Roger Pielke Jr. hiring was a mistake. Romm routinely and rudely smears those who don't agree with him (including respected science journalists like Andy Revkin), and trashes any study that doesn't validate his opinion. I know that Pielke Jr. can be kind of a smarmy, frustrating figure, but you're not getting the full truth if you just rely on Romm. For example, see Revkin's coverage on the California drought debate between Pielke and Holdren.

  2. Thanks for the comment and the link. I will take a look at it. I didn't link to Romm, by the way, but to Kroh, who quotes him. I should have included something else, perhaps by Michael E. Mann, about him (for instance I linked to that article, in fact, because Kroh quotes Mann, too.

  3. Oops -- I didn't check the author line. ClimateProgress has traditionally been Joe Romm's blog, and I didn't notice they opened it up to new authors, so I just assumed it was a Romm post. Climate politics are tough, and I've been following it for years as people sling mud back and forth. I come from the perspective of a scientist (ecologist) who wants climate action, but finds the attack dog style of Romm unproductive.

    I'm waiting to pass judgment on Pielke's column in FiveThirtyEight until he publishes a few columns, and then hope to judge it based on the data he presents. His first column covers what I expected it would - the relationship between climate change and natural disasters. I think part of the reason those like Climate Progress dislike Pielke is because he points out that there's not always a relationship between natural disasters and climate change, and if we care about reducing damage to people and property, there are adaptation and mitigation measures we can take that aren't linked to cutting carbon consumption. For those who want to link every bad thing to climate change, they see Pielke as undermining their message.

    A good example is tornadoes. Right now, the science says there's no relationship between increased strength or number of tornadoes. When there was a bad tornado season a couple years back, Romm jumped all over it, while Andy Revkin pointed out that, "it is irresponsible not to mention the need to reduce inherent and avoidable human vulnerability to tornadoes in the crowding South, particularly in low-income regions with flimsy housing." Revkin gives a great wrap on the relationship between tornadoes and climate change here. And this is often what Pielke hits on -- the data currently say that if we want to save human lives right now, we can't deny adaptation and mitigation strategies.

    This is all a long winded way to say that I think it's a good thing if Pielke Jr. brings a data-driven column that explores the grey areas of climate change and science to FiveThirtyEight, as things aren't always as black and white as some would make them seem. That seems to be right up Nate Silver's alley. And if Pielke Jr. makes assertions that aren't supported by the science and data, I hope (and expect) he gets called on it, in a productive, non-vindictive way.

  4. It seems that folks criticizing 538 for not being deep/rigorous enough are pointing to the DataLab pieces ( see Tyler Cowen ( and Paul Krugman's ( responses, which both do this.

    The DataLab pieces are explicitly stated to be brief/timely counterpieces to the site's featured articles: they're not intended to offer deep analysis or heavy context. They even have separate social feeds (@fivethirtyeight vs @datalab538). If people are going to criticize the site's offerings for not being deep enough, don't focus on the pieces that are intended to be breezy!

  5. What I meant is that I was expecting more longform (data) journalism. But, as I wrote, it's too early to judge the website properly. So far, so good. I follow it closely