Kevin Drum has a weird post up about how there are no big lessons of the 2016 election (at least not at the presidential level), because nothing much actually happened:

Everyone wants to draw big, world-historical lessons from this election. That’s understandable, since the result was the election of an unprecedentedly dangerous and unqualified candidate. But the data just doesn’t support any big lessons. Barack Obama won the popular vote in 2012 by 3.9 points. Hillary Clinton won the popular vote in 2016 by 2.1 points. That’s less than a two point difference, despite the fact that Obama is unusually popular and Clinton had to run after eight years of Democratic rule. In the end, she did slightly worse than Obama, which is about what you’d expect. Unfortunately, a little too much of that “slightly worse” happened to be in three must-win states.

Nevertheless, the identity politics critics insist that the lesson for Democrats is to ditch identity politics. The economic lefties say the lesson is that Democrats need to be more populist. The Bernie supporters are sure that Bernie could have won. The DNC haters think it was a massive FUBAR from the Democratic establishment. The moderates blame extremism on social issues for alienating the rural working class.

These have one element in common: All these people thought all these things before the election. Now they’re trying to use the election to prove that they were right all along, dammit. But they weren’t. This election turned on a few tiny electoral shifts and some wildly improbable outside events. There simply aren’t any truly big lessons to be drawn from it.

On one level, this is very true — but it also proves way too much. If nothing matters at all, and the electorate is so partisan that either party could nominate a headless chicken and come pretty close to what econometric models predict, then why bother having elections in the first place? It’s more an argument against democracy itself than against over-interpreting the 2016 election specifically.

And 2016 is a pretty weird year to describe as featuring “just a few tiny electoral shifts.” After all, Donald Trump won the Republican nomination by running against essentially the entirety of his party’s leadership. And his main competition was another guy running against the entirety of his party’s leadership. The overwhelming majority of GOP primary voters opted against everybody contending for support of the party leadership. That’s not a meaningless fact. Then Trump went on to win an electoral college majority in spite of never uniting his party leadership behind him, being massively outspent, and being opposed by essentially the entirety of the media establishment, and a chunk of the conservative counter-establishment. That’s not meaningless either.

Had Trump lost his three key midwestern states by the same narrow margin that he won them, nobody would be saying, “see? we told you Clinton had this in the bag.” They’d be saying, “holy crap — that was way too close!” Republicans would be furiously debating whether someone less outrageous than Donald Trump but running on a Trumpian platform could win in a walk, or whether they should return to the true Reaganite faith. Democrats, meanwhile, would be fretting about erosion of support in the Midwest and whether they need to shore it up by moving left on economics and/or ditching identity politics, or whether they should focus on “flipping” North Carolina and Florida to compensate for their inevitable losses of Pennsylvania, Michigan and Wisconsin in 2020. Everyone would be debating what the very obvious and meaningful changes in the electorate portend — and in very similar terms to the way they are now.

What is probably a waste of time is focusing too much on the small-ball “lessons” of the Clinton campaign’s arrogance and incompetence. This article about how they lost Michigan is instructive in that regard:

Everybody could see Hillary Clinton was cooked in Iowa. So when, a week-and-a-half out, the Service Employees International Union started hearing anxiety out of Michigan, union officials decided to reroute their volunteers, giving a desperate team on the ground around Detroit some hope.

They started prepping meals and organizing hotel rooms.

SEIU — which had wanted to go to Michigan from the beginning, but been ordered not to — dialed Clinton’s top campaign aides to tell them about the new plan. According to several people familiar with the call, Brooklyn was furious.

Turn that bus around, the Clinton team ordered SEIU. Those volunteers needed to stay in Iowa to fool Donald Trump into competing there, not drive to Michigan, where the Democrat’s models projected a 5-point win through the morning of Election Day.

Michigan organizers were shocked. It was the latest case of Brooklyn ignoring on-the-ground intel and pleas for help in a race that they felt slipping away at the end.

“They believed they were more experienced, which they were. They believed they were smarter, which they weren’t,” said Donnie Fowler, who was consulting for the Democratic National Committee during the final months of the campaign. “They believed they had better information, which they didn’t.”

Flip Michigan and leave the rest of the map, and Trump is still president-elect. But to people who worked in that state and others, how Clinton won the popular vote by 2.8 million votes and lost by 100,000 in states that could have made her president has everything to do with what happened in Michigan. Trump won the state despite getting 30,000 fewer votes than George W. Bush did when he lost it in 2004.

The article goes on to detail exhaustively how Clinton’s campaign in this crucial state ignored all the evidence that it was slipping from their fingers. It’s a hugely damning indictment, and the people who manage and run the next Democratic presidential campaign had better read it.

But it’s not an adequate explanation for Clinton’s loss. Because the Clinton campaign put huge efforts into GOTV in another crucial state — Pennsylvania — that they did see was at risk. And those efforts paid off — Clinton’s vote total was just a whisker shy of Barack Obama’s 2012 winning total in the state, or John Kerry’s winning 2004 total.

And Trump won Pennsylvania by a larger margin than he won Michigan.

The point of an explanation isn’t to provide predictive power, but to help determine what to do next. A predictive model doesn’t need to explain why anything happens, and therefore need not provide any guide to action. “Fundamental models say this election will be close” doesn’t tell you what ground the election will be fought on, or how to maximize your chances of victory fighting on that ground.

The most important lesson of 2016 is not “the country is so partisan that nothing matters” but rather “fundamentals matter way, way more than campaigns.” Trump, after all, made just about every mistake you could possibly make in his campaign. But here we are litigating over which mistake Clinton made that cost her the election, or whether it wasn’t her fault at all but instead the fault of the FBI or the Russians.

But that doesn’t mean that campaigns don’t matter. It means that what matters most about campaigns is whether they understand the fundamentals. The next election will be fought on the ground shaped by the fundamentals of the Trump era. If those fundamentals are meaningfully better than they were in 2016, then it will be an uphill battle for the opposition. If they are stagnant or worse, then the ground will be more favorable. But the fundamentals will determine the shape of the ground, and if the Democrats don’t prepare to fight on that ground then they will not maximize their chances of winning.

Any debate about what the ground looks like and how to fight on it is worth having.