Clinton Campaign May Have Been Too Smart to Win

An insanely close loss could be explained by many things, but the faulty self-confidence of “Brooklyn” may have ultimately betrayed Hillary Clinton. Photo: Brendan Smialowski/AFP/Getty Images

The hardest thing about explaining very close election defeats is that you can talk yourself into believing any number of factors, large or small, could have made the crucial difference. Hillary Clinton’s defeat came down to 100,000 votes in three states out of nearly 130 million cast nationally. That’s a deficit so tiny that the search for a single culprit will probably take political detectives down the dark road to madness.

It is understandable that within the ranks of the Clinton campaign’s own high command, it is an article of faith that FBI director James Comey cost HRC the presidency by making the ridiculous email-server issue the dominant subject of conversation during the crucial last days of the campaign. It was an external event that came out of the blue (or more specifically, out of the fevered libido of Anthony Weiner) and reinforced doubts about Clinton’s trustworthiness among a small group of undecided voters in key states.

But an equally plausible case can be made that the Clinton campaign itself was so convinced it could not lose that it missed the danger signs emanating from the three Clinton firewall states (Michigan, Pennsylvania, and Wisconsin) that eventually awarded Trump the presidency despite a 2 percent deficit in the national popular vote. Focusing on what happened in Michigan, Politico’s Edward-Isaac Dovere provides a wealth of anecdotal evidence of a national campaign that could not believe and thus did not pay attention to distress signals about “base” turnout and defections to Trump.

It is illuminating to compare Dovere’s piece to a September Politico article by Shane Goldmacher in which Clinton campaign operatives celebrate the transcendent genius of its data-analytics arm, anchored in what worked for Barack Obama in 2012. Clinton analytics chief Elan Kriegel was the hero of the story, and readers came away with the impression that nothing could happen anywhere in the U.S. electorate without a Kriegel-developed algorithm instantly kicking in to adjust campaign resources and sustain the flight path to victory.

Unsurprisingly, as Dovere found, this all-knowing analytics system and the iron self-confidence it inspired made “Brooklyn” (Clinton’s national headquarters) largely impervious to feedback:

The anecdotes are different but the narrative is the same across battlegrounds, where Democratic operatives lament a one-size-fits-all approach drawn entirely from pre-selected data — operatives spit out “the model, the model,” as they complain about it — guiding [campaign manager Robby] Mook’s decisions on field, television, everything else.

Now the Clinton campaign was not unique in its reliance on a “model” for understanding election dynamics. One of the big trends since 2012 among political practitioners and observers alike has been the gradual displacement of random-sample polling with models of the electorate based on voter-registration files, supplemented by tracking polls of this fixed universe of voters. This approach tends to create a more static view of the electorate and its views, and probably builds in a bias for thinking of campaigns as mechanical devices for hitting numerical “targets” of communications with voters who are already in your column. You could see this new conventional wisdom (and the pseudoscientific certainty it bred) in pre-election models published by Bloomberg Politics and in an Election Day modeling experiment conducted by Slate. Having invested heavily in its own “model” for what it needed to do when and where, the Clinton campaign was naturally resistant to conflicting signals from the ignoramuses on the ground.

It is in that respect that just about everyone within and beyond the Clinton campaign erred in crediting it with a state-of-the-art “ground game” worth a point or two wherever it was deployed. Clinton had lots of field offices, to be sure. She had more money for get-out-the-vote operations. Team Clinton did much, much more targeted outreach to key voters in key states than did Team Trump. But in the end “Brooklyn’s” decisions were based on assumptions that had very little to do with actual developments on the “ground;” its hypersophisticated sensitivity to granular data about many millions of people made it fail to see and hear what was actually happening in the lead-up to the election.

For now it probably doesn’t matter whether it was James Comey or the campaign’s faulty self-confidence that cost Clinton the election. But when it comes time to build the next presidential general-election campaign, the people setting up the organization and paying the bills might want to rely a bit less on any system that values analytical omniscience at the expense of a willingness to change the game plan if there are signs that that is needed.

If, God forbid, I were running a future presidential campaign, the headquarters would be plastered with posters emblazoned with the title of the autobiography of the great, data-driven baseball manager Earl Weaver: “It’s What You Learn After You Know It All That Counts.”

Clinton Campaign May Have Been Too Smart to Win