My 2016 predictions were … not the best. The 2018 predictions were better. From last Tuesday afternoon:

I’ll guess a net gain for Dems of 37 seats, then, for no particular reason. Although I like the fact that that number is in between higher- and lower-end estimates being offered by data nerds. RCP analyst Sean Trende predicted a Democratic pick-up of 32 seats a few days ago; Nate Silver’s site is detecting an average Democratic pick-up of 39 in the models it’s running based on the current polling with the potential for much, much more if Democratic turnout is strong nationally. But I’ll be, er, conservative in saying 37.

In the Senate I’m guessing a final split of 52/48 for the GOP. Republican pick-ups tonight: Kevin Cramer in North Dakota (obviously), Josh Hawley in Missouri (although I fear McCaskill’s lucky streak), and I’ll go with Matt Rosendale in Montana as my upset special. Democratic pick-ups: Jacky Rosen in Nevada given her lead in the early voting there and, alas, Kyrsten Sinema in Arizona, who’s stubbornly led in most polls of the state. I think Joe Manchin, Bob Menendez, and Debbie Stabenow all win comfortably and that Joe Donnelly and Bill Nelson win narrowly, although if any of those predictions are wrong Donnelly is easily the likeliest of the five to lose. Ted Cruz and Marsha Blackburn will have more suspenseful nights than everyone expects before (whew) prevailing.

Votes are still being counted in some races (cough) but as things stand as of 1:30 p.m. ET today the GOP holds a 52/47 advantage in the Senate and Democrats have gained 32 House seats. That Senate number will likely settle at 53/47 if and when Rick Scott’s victory in Florida is confirmed. I missed on that race and missed badly on Donnelly’s blowout loss in Indiana, and although Marsha Blackburn did win in Tennessee it wasn’t suspenseful. All told, though, not bad.

Not bad with the House count either. In fact, there’s still a chance that Democrats will end up gaining 37 seats exactly:

The 2018 midterm election looked last Tuesday like a serious but not crippling setback for Republicans, with Democrats making significant gains in congressional races but failing to deliver a comprehensive defeat to President Trump and his party.

The picture for Republicans has gotten grimmer since then, as a more complete tally of votes has come in across the country.

What looked at first like a modest Democratic majority in the House has grown into a stronger one: The party has gained 32 seats so far and appears on track to gain between 35 and 40 once all the counting is complete.

A net gain of just two seats in the Senate and a loss of three dozen seats in the House feels wave-y, if not quite the tidal wave that the GOP enjoyed in 2010. The 2006 midterms were seen as a Democratic wave even at the time, a backlash to the Bush administration’s performance in Iraq and after Hurricane Katrina, and yet Democrats gained just 31 House seats that year. They cleared that benchmark this time, with the counting not yet done. The reason it didn’t feel like a wave last Tuesday night was due to some early bellwether races that went the GOP’s way (e.g., Barr vs. McGrath in Kentucky), Mike Braun’s surprising blowout of Donnelly in Indiana, and the shocking double-barreled upsets in Florida by Ron DeSantis and Rick Scott. But after the shaky start, Democrats began winning the toss-up races in the House; Braun’s promising win wasn’t replicated in battlegrounds like Montana, Nevada, and Arizona; and the nailbiters in Florida overshadowed a rout by Democrats in most statewide races in the Rust Belt, which were supposed to be “Trump Country” after 2016. Even the modest Senate gain came with an asterisk, GOP pollster David Winston told the Times. Huge losses among women, young adults, independents, Latinos, and suburbanites are simply not sustainable, he said. Republicans held the Senate this year because, by chance, nearly all of the battleground races were fought on favorable turf.

The most noteworthy group of the ones I just named are suburbanites. The chatterati expected them to break decisively for Hillary in 2016, enough so to hand her the presidency, but Trump held onto enough of them to enable his white working-class base to put him over the top in key states. Not so this year. It’s often said that “Never Trumpers” don’t exist outside a small segment of conservative media, Michael Warren noted a few days ago, but last Tuesday’s results suggest that’s not true. Some of the Maybe Trumpers in the suburbs seem to have become Never Trumpers. If Trump can’t figure out a way to turn them back into Maybe Trumpers in the next two years, he’s done.

This is a culmination of the strategy Trump brought to his adopted party: actively court the white working-class voter. These blue-collar conservatives had been skeptical of the plutocratic Mitt Romney and the conventional GOP agenda four years earlier, went the thinking, while a straight-talker with a populist agenda could add them to a successful Republican coalition. The new strategy still requires keeping the existing members of the coalition in the tent, however, especially the suburbanites.

But there’s a problem: These suburban voters are not Trump’s base. In many ways they are the segment of the party Trump’s candidacy was designed to ignore, downplay, or even antagonize. Republicans in these districts aren’t as alarmed by immigration. They prefer the benefits of global trade to the protections of tariffs. They bristle at Trump’s coarse style of politics that punches first and asks questions later. For these voters, civility and moral leadership are more than just niceties—they’re motivating issues. The message Trump and the GOP send, either incidentally or intentionally, is that these Americans matter less. Increasingly, they’re listening.

Even if Trump figures out a way to increase working-class white turnout in 2020 and offset losses in the suburbs, as Warren noted elsewhere, that might solve his reelection problem but it won’t solve the GOP’s. Most of the districts that flipped on Tuesday were in suburban districts; Trump might be able to win battleground states by turning out rural voters but that won’t mean a thing for Republican House candidates in suburban areas who end up getting swamped by an anti-Trump backlash. The realistic best-case scenario for Trump in 2020 may be winning a second term but ending up with a fairly solidly blue House that’ll block anything he tries to do for the next four years when it isn’t busy serving him subpoenas. Last Tuesday’s wave was the first step towards that bleak scenario.