Covering the 2016 election was the most intellectually challenging thing I've ever done as a writer.

By my count, I wrote 135 columns this year, with 110 of them focused on some aspect of the presidential contest. That would have been difficult during any election year. But of course 2016 wasn't just any election year.

This was the year the impossible happened. Or rather, the year something highly implausible did — not once, but twice. First Donald Trump managed to win the nomination of the Republican Party, and then he managed to win the presidency.

Watching it unfold in real time, needing to analyze and make sense of it all, and then form sharp, intelligent, engaging opinions about it, and endure the critical assault of readers, three times a week — it was dizzying, arduous, and incredibly fulfilling. For as long as I live, and however the Trump administration unfolds, I'll look back on this past year as one of the seminal experiences of my life.

Which isn't to say I'm proud of everything I wrote this year. Far from it.

The columns that hold up best, I think, are those that attempted to assess the Trump phenomenon dispassionately, placing it in an historical, philosophical, and global context — and those that took on the media for its often unhinged coverage of the campaign. Some of my horse-racey predictions proved to be pretty accurate — declaring on March 16 that Trump was unstoppable for the GOP nomination, highlighting Hillary Clinton's serious vulnerabilities in general, and against Trump in particular. My favorite column of the year sought to explain the Brexit vote (and related developments throughout the West, including the rise of Trump) in terms of a broad-based rejection of progressive ideology.

But there were also plenty of howlers. There always are some, which is why I've made a tradition of devoting a year-end column to highlighting what I got wrong and reflecting on why. But this year, with so many columns written in response to ongoing, unprecedented events, there were more messes than usual.

Sometimes the mistake was a product of bad timing and deadline pressure. That was certainly the case with my foolish column about Ted Cruz's speech at the GOP convention. Going into the evening, I planned to write about how Cruz's remarks would set up his 2020 campaign for president by positioning him as a (slightly more mainstream) populist who could appeal to many of Trump's most passionate supporters the next time around. (Assuming, of course, that Trump would lose in November. I'll return to that presumption in a moment.) But then Cruz came out and knifed Trump live on national television by refusing to endorse him and telling audience members to "vote your conscience."

What to do? Why just slightly adjust my argument, of course! How brilliant of Ted! He'd made it possible to run in four years as Trump 2.0 while also insulating himself from the disaster of Trump 1.0. What a master political tactician!

It was a stretch when the column appeared the next morning. By the time Cruz ended up endorsing Trump in mid-September, it had become completely ridiculous. Far from proving himself a skilled strategist, Cruz had demonstrated his political incompetence and, for a man who makes a fetish of upholding principle, an almost comic incapacity to take a stand and stick with it.

Then there are the numerous columns in which I thought through the post-election landscape — every one of them based on the assumption that Trump would lose on Nov. 8. Not all of these are worthless. If you're willing to bracket that faulty premise, the dynamics I discuss — within each party and between them, and in the country's political culture more widely — are still valid in places.

But that isn't true at all of my Sept. 1 column — the one that deserves every bit of the opprobrium that's been directed its way. And there's been an awful lot of it. Why? Because that's the column in which I not only forecasted a Clinton victory but foolishly offered an extremely specific prediction that Donald Trump would "lose in the biggest landslide in American history."

From the morning the column appeared all the way down to Election Day, Trump supporters laid into me for what they saw as a piece of writing that exemplified all that was delusional about the liberal media. Trump would win. He'd show me. And even if he lost, it wouldn't be a landslide, let alone the biggest one in American history.

How did I get it so wrong? There were several factors. First, I really did think Trump would lose. I blame his victory mainly on Hillary Clinton's poorly run campaign, and ultimately on the Democratic Party's decision to rally behind a candidate under investigation by the FBI. Trump should have lost. He was ignorant, offensive, inept. An appealing, scandal-free opponent should have sunk him easily.

Fine. But where on earth did I come up with "the biggest landslide in American history"? That was a product of the moment the piece was written — and my willingness to go out on a limb.

Throughout the middle and end of August, Trump was languishing in the polls, averaging at around 40 percent, with a few big-name polling outfits showing him down in the 35-38 percent range. Meanwhile, in the two weeks leading up to this column, Trump had hired Kellyanne Conway to be his campaign manager and then made some statements that sounded like he was backing away from his draconian anti-immigrant position. What if that were the future of the Trump campaign? If he alienated his most passionate supporters by moderating on immigration while remaining ignorant, offensive, and inept (as he surely would), that could be a recipe for a loss in Alf Landon territory (36.5 percent).

So that's what I predicted. It was risky. And very, very wrong. I've learned my lesson.

Crow doesn't taste good, but it can be healthy. Provided it's a meal eaten rarely.