It’s always worth looking at electoral history before looking ahead, and I did so today at New York.

[T]he [2018] Senate landscape seems impossibly difficult for the Donkey Party, which is defending 25 of 33 seats up next year, ten of them in states carried by Donald Trump last year. And of the Republican incumbents who are defending their seats, only two appear vulnerable at this point (though both, Jeff Flake and Dean Heller, are now really vulnerable, facing both primary and general-election opposition). Everyone understands that the pro-GOP landscape and the anti–White House midterm dynamics are pushing these races in different directions. But it’s hard to get a handle on the magnitude of these contending forces.

Larry Sabato’s Crystal Ball filled in some of our knowledge gap today, with a study by Geoffrey Skelley of Senate races in midterms dating all the way back to 1914. The numbers are pretty clear: “91% (287 of 314) of non-presidential party incumbents won reelection in midterms.” That’s much better than the 75 percent midterm reelection rate of senators from the presidential party.

Lest you imagine this is all a vestige of a bygone era of wide-scale ticket-splitting, there’s this factoid:

“If anything, out-party incumbents losing in a midterm is becoming less common: In six of the last eight midterms, including the last three (2006, 2010, and 2014), no such incumbent lost reelection.”

Here’s the fun part of the Crystal Ball analysis:

“Applying the historical averages to next year’s Senate elections would result — drumroll please — in a net party change of…zero seats. If 91% of the Democrats/Democratic-caucusing independents are reelected, that would be 23 out of 25, and if 75% of the Republicans are reelected, that would be six of eight, leading to no net change.

“Without making any predictions, such a scenario is plausible: Democrats could lose two of the incumbents defending dark red states, in states such as Indiana, Missouri, Montana, North Dakota, and West Virginia, but otherwise hold everything else, while Republicans could lose the only two seemingly vulnerable seats among their much smaller stable of incumbent-held states: Arizona and Nevada.”

It is clear that the tendency to lump all of the so-called Trump Ten senators into the same endangered category has never made much sense. Trump carried Florida, Michigan, Pennsylvania, and Wisconsin by narrow enough margins that even the mildest midterm anti–White House breeze could carry Democratic incumbents over the finish line. Ohio (an 8-point Trump margin in 2016) is more promising for Republicans, with Indiana (a 19-point margin) and Missouri (18 points) even more promising, and Montana, North Dakota, and West Virginia (20-point-plus margins) really ripe targets. But Democratic incumbents in all these states have their own distinctive strengths, and at the moment not one of them is trailing according to the authoritative Cook Political Report.

There is one more imponderable to throw into the mix: Will the tendency of demographic groups that have recently become central to the Democratic coalition to significantly underperform in midterms change? There is reason to believe, in part based on special elections since Trump took office, that the spirit of resistance to the 45th president will offset some if not all of that phenomenon. But it’s something to keep in mind.

In general, the shape of the Senate after 2018 is going to be conditioned by a variety of competing and conflicting factors. One thing, however, is very clear: Had Hillary Clinton won last November, we would probably be looking at big Senate — and House — gains for the GOP next year. Using the in-party/out-party historical percentages, here’s what you’d get according to Crystal Ball’s Kyle Kondik:

“Just going by the averages, we’d expect the Democrats to lose a half-dozen seats and the GOP to lose one or even zero incumbents. And the net Democratic losses probably would have been even bigger because it’s easy to imagine one or more red state Democrats seeing the writing on the wall and retiring in advance of a tough midterm, giving the Republicans an easy open-seat pickup or two.”

In other words, had Clinton won, we might be talking about a GOP drive to get a 60-vote supermajority in the Senate. There are always silver linings for the party that falls short in presidential elections.

As part of our continuing look ahead at 2018, I offered these thoughts about 2017 developments at New York.

Projecting what will happen in midterm elections is always tricky. Yes, the party controlling the White House almost always loses House seats — though that didn’t happen in 1998 and 2002. Sure, the president’s approval ratings seem to have a very significant impact on the White House party’s losses or (in rare occasions) gains. But predicting approval ratings is tough.

But one bit of objective data you can track that’s not simply a matter of projections is the performance of the two parties in special elections heading toward the midterms. And while no one special election necessarily has predictive value, if you add them up it starts mattering.

That’s what Harry Enten did today at FiveThirtyEight, and better yet, he scored all these elections like a bookie would score a sporting contest: with a “spread.” In this case the spread was the partisan “lean” of every congressional or state legislative district that’s had a special election so far this year, based on its performance in the last two presidential elections, with the last one weighted most heavily (as one would expect). There have been five U.S. House special elections this year. Democrats beat the spread in all five, by an average margin of 16 percent of the total vote (ranging from 6 percent in Georgia’s sixth district to 23 percent in Kansas’s fourth district). That’s a lot.

There have also been 25 state legislative special elections this year. Democrats beat the spread in 21 of them, by an average of — again — 16 points. None of the four races where Democrats failed to do better than the partisan lean were competitive contests.

Enten notes that Trump’s approval ratings would historically suggest an 11-point GOP deficit. The generic congressional ballot has been showing a 9-point GOP deficit at present. So the special elections are showing a stronger swing to Democrats than the standard data points.

In any case, it represents a lot of arrows pointing in the same direction.

Like most everyone else, I was shocked by the president’s nonchalant and false-equivalence-laden initial reaction to this last weekend’s white riot in Charlottesville, and wrote a take for New York that delves a bit into the history involved:

There is a sinister congruence between the president’s reaction to the white-supremacist riot in Charlottesville yesterday and the object the rioters assembled to defend: the city’s doomed Robert E. Lee statue. Both are manifestations of Neo-Confederacy, the fierce, century-long effort of the Southern ruling class to normalize white racism so long as it did not degenerate into extralegal violence.

Like most of its counterparts across and beyond the South, Charlottesville’s Lee statue was not erected during the Civil War or in the period when ex-Confederates might be expected to remember the famed military leader of the planter’s rebellion. It was commissioned in 1917 and erected in 1924 as a monument, not to the Confederacy, but to the rebellion’s posthumous victory over Reconstruction and the Civil War amendments to the Constitution.

The Neo-Confederates and their many Yankee sympathizers viewed Jim Crow as a peaceable compromise between slavery and racial equality. In that regime’s latter days, whole generations of white Southern politicians posed as civil upholders of law and order equally opposed both to civil-rights “agitators” and to the white-trash hoodlums of the Ku Klux Klan, who were successors to the white terrorists that the “better element” of Southerners strongly supported during and immediately after Reconstruction. These politicians often condemned, as Donald Trump did yesterday, the “egregious display of hatred, bigotry and violence, on many sides,” that threatened the quiet tyranny of Jim Crow. They preferred a “civil” resistance to equality in Congress and the courts, and in the genteel Citizens’ Councils that, as one sociologist aptly put it, “pursued the agenda of the Klan with the demeanor of the Rotary Club.”

The Neo-Confederates lost the battle against civil rights, but maintained a cultural rearguard for many years. You heard at least a faint echo of their words every time a conservative Southern politician hailed “law and order,” or attacked “the welfare,” or demanded maximum incarceration of African-American “predators.” This sort of politics maintained an unmistakable connection to nostalgia for the Old South, an imagined tranquil place of good manners and interracial understanding.

But until Donald Trump’s election, it seemed Neo-Confederacy had finally about run its course. The display of Confederate regalia on state flags, public buildings, and even football mascots gradually became distasteful, even to many conservative politicians. In 1993, when Georgia governor Zell Miller (later a conservative hero) proposed getting rid of the Confederate symbolism on Georgia’s state flag (imposed not during or after the Civil War, but during the period of white resistance to desegregation), soon-to-be House Speaker Newt Gingrich supported him. And by the time South Carolina governor Nikki Haley finally had the Confederate Battle Flag taken down from the statehouse after an outburst of racist violence — nearly 20 years after a Republican predecessor had proposed the same thing — it aroused little public opposition.

Yes, even hard-core conservatives began to understand that Confederate insignia were not just parts of history that today’s Southerners should “cherish,” to use the president’s startling allusion to the Lee statue, but part of a retroactive effort to whitewash history in the pursuit of racist lies.

But then, in the blink of an eye, the backlash to acts of simple racial decency began. It was not confined to Donald Trump’s campaign, but in many corners of the right, hostility to “political correctness” — defined as sensitivity to the fears and concerns of, well, anyone other than white men — became a hallmark of the “populist” conservatism Trump made fashionable and ultimately ascendent.

And so the relatively uncontroversial movement to get Jim Crow era Confederate insignia and memorials out of the public square and back into museums and history books suddenly faced renewed opposition — not just from the motley crew of open white supremacists who viewed the 45th president as their hero, but from politicians who saw a broader constituency for a brand-new era of white backlash. It is no mistake that Corey Stewart — who was Trump’s 2016 Virginia campaign chair until he was dumped for excessive public hostility to anti-Trump elements in the Republican National Committee — seized on the decision of the Charlottesville City Council to remove the Lee monument as an example of contemptible “political correctness” in his surprisingly successful 2017 GOP gubernatorial campaign (which fell just short of upsetting the heavily favored Ed Gillespie). Stewart, who is now running for U.S. Senate, not only defended Trump’s refusal to distinguish between white supremacists and their opponents in the Charlottesville violence, but took it to the next level in an interview with Breitbart News:

“We have the violent left which recently attacked a U.S. Congressman [Steve Scalise], which has been attacking Trump supporters across the country, and I never hear Democratic politicians condemning them. I’m not going to play their game. I am not going to condemn anyone other than the criminal. We always have to protect citizens who are trying to exercise their First Amendment rights. From my perspective, there were a lot of left-wing agitators who violently attacked citizens who were trying to espouse their views last night and today.”

So there you have it: Not only were the Nazis, Klansmen, and other white supremacists who chose Charlottesville for their big rally simply trying to “espouse their views,” but also conservatives should not say a discouraging word to them so long as they aren’t “the criminal,” however that is defined.

Perhaps the president will eventually be convinced by his more politically pragmatic allies to draw a clear line between himself and the racists who revere him. It would be immensely more valuable if he condemned not just idiot Klansmen and neo-Nazis but demagogues like Stewart whose idea of “Trumpism” is to champion any and all types of white backlash to “political correctness” and “the Left” as legitimate.

As we now know, Trump did briefly (and not very convincincly) clarify his feelings about the white supremacists of the Unite the Right demonstration. And then he showed his truer colors in a subsequent press conference wherein he expressed solidarity with the “fine people” amongst the Klansmen and neo-Nazis who just wanted to revere the memory of Gen. Lee. And so he rejoined the neo-Confederacy after a couple of days apart.

As always, Ron Browntein slices and dices public opinion data in an especially useful way, as I was happy to share with readers at New York:

It’s generally conceded (though not by the president himself, who argues the size of his rallies is the best indicator of his popularity) that Trump’s already weak public standing has softened somewhat. But there is less agreement about where he stands with this or that part of the electorate — particularly his “base,” however that is defined.

But now Ron Brownstein, working with some previously unpublished Gallup data from 13 current and projected battleground states (Arizona, Colorado, Florida, Georgia, Michigan, Minnesota, Nevada, North Carolina, Ohio, Pennsylvania, Texas, Virginia, and Wisconsin), has given us the best snapshot yet of exactly where, geographically and demographically, Trump stands as compared to his performance last year. It shows that he is hemorrhaging the most support from college-educated white voters, though he’s lost a crucial bit of his strength from the non-college-educated white voters often considered his “base.”

“Trump’s approval rating among college-educated whites has declined relative to his 2016 vote in all 13 states. In seven of those states, his approval rating stands at least 10 points lower than his vote –a list topped by North Carolina and Florida (both 19 points lower), Georgia (18 points lower), Ohio (15 points), Virginia (12 points), and Michigan and Minnesota (11 points each.) His approval rating among these white-collar whites reaches above 50% only in Texas and Georgia, and exceeds 45% in just two other states, Nevada and Arizona. In seven states, his approval among these well-educated white voters has tumbled to 40 percent or less. (That includes four states essential to his victory: North Carolina, Michigan, Pennsylvania and Wisconsin.)”

Some might object that Trump managed to win last year despite poor approval numbers then. But that cuts both ways: In 2016, Trump was competing with a pol who was nearly as unpopular as he was. Hillary Clinton will not be on the ballot or in the consciousness of voters in 2018 —or presumably in 2020, either.

“These anemic numbers suggest how much Trump benefited last fall from the doubts these voters also held about Clinton; standing alone, without her as a foil, he’s facing much harsher assessments. (In a separate national Quinnipiac University poll released last week, 59% of college-educated whites said they “strongly” disapproved of Trump’s performance.) “I think the doubts about her blocked how big the potential was for those voters to vote against Trump,” says long-time Democratic pollster Stanley B. Greenberg.”

Relatively speaking, Trump’s support is holding up better among white working-class voters, the core of his base. But there are still problems:

“[Trump’s] standing still represents an erosion from his 2016 vote among blue-collar whites in 12 of those states; in five of them, he’s declined by double-digits. Perhaps most important are the trends in four of the Rust Belt states that proved decisive last year: Michigan, Ohio, Wisconsin and Pennsylvania. In all of them exit polls found Trump won between 62 and 64% of non-college whites. In each case, that was a substantial increase from Romney’s performance with those voters in 2012, when former President Barack Obama carried the states.”

In those four crucial states, which were where the 2016 upset was consummated, Trump’s white-working-class approval numbers are down around the levels of support Romney achieved in losing all four in 2012.

We sometimes forget in our awe at Trump’s astonishing rise to the presidency that it was a near thing, particularly in the general election. He doesn’t have much margin for error, and nor does his party in a midterm election that usually represents a referendum on the party controlling the White House (and in this case, the much-despised do-nothing Congress as well). The drops in support among white voters in battleground states that Brownstein documents aren’t massive, but they could spell the difference between Republican success and calamity just down the road.

It’s awfully early to be thinking about the 2020 election, but potential candidates are already seeing the next president of the United States in their bathroom mirrors, so observers must take notice, as I did at New York:

[T]here’s nothing unusual about speculation surrounding two fairly obscure Democratic House members, Seth Moulton of Massachusetts and Tim Ryan of Ohio. Compadres in an unsuccessful revolt against Speaker Nancy Pelosi last November (Ryan was the alternative candidate, Moulton a vocal supporter), they are both already being mentioned in connection with 2020…. They will join another colleague who has attracted some national attention, Cheri Bustos of Illinois, as featured speakers at September’s Polk County Steak Fry, in Des Moines. This event is an effort to revive the annual Steak Fry event former senator Tom Harkin used to host in his home town of Indianola, which was a big-time magnet for future Democratic presidential candidates.

Everybody should get used to the idea of a putative 2020 Democratic field the size of an Iowa cornfield. There are three potential candidates who would each become front-runners in the 2020 race if they decide to run: Bernie Sanders, Joe Biden, and Elizabeth Warren. There are questions as to whether any or all of them are too old, particularly for a party looking to show it’s shed the barnacles exposed by the 2016 Clinton campaign. Sanders will be 78 when the 2020 Iowa Caucuses are held; Biden will be 77, and Warren will be 71 (for that matter, Donald Trump will be 73). But none of them has any particular reason to diminish their influence by declining interest in 2020. Their long shadows will make it harder for little-known alternatives to emerge. But the possibility of retirement or illness among the Big Three keeps open the gate to dark-horse fantasies.

If Sanders, Biden, and Warren do fall by the wayside, Democrats may suddenly find themselves with a presidential field that resembles the mob that ran for the GOP nomination in 2016. And as National Review’s Jim Geraghty reminds us, that did not work out to well for those Republicans who kept expecting someone to emerge to knock off Donald Trump, right up to the moment he was nominated:

“One chunk of the field convinced itself there was an ‘establishment lane,’ leaving Jeb Bush, John Kasich, Marco Rubio, and Chris Christie all elbowing each other for the same base of support that proved insufficiently influential. On the other side, Ted Cruz, Rick Perry, Bobby Jindal, and Scott Walker tried to occupy the ‘conservative lane.’ Ben Carson and Carly Fiorina competed with Trump for an ‘outsider lane.’ But in the end, it turned out there were no real lanes, just a traffic jam. Every non-Trump candidate’s determination to be the last one standing against Trump was the strategic miscalculation of the cycle.”

There is no Donald Trump analogue gearing up for a presidential run on the Democratic side in 2020, so far as we know. But if one emerges, she or he will be helped enormously if there is a large field against which to pose as Gulliver among the Lilliputians. Geraghty counts 18 possible candidates right now, and while some will certainly not run, others may come out of the woodwork if the Big Three give the race a pass. Indeed, the Trump precedent has made the narcissism of long shots seem a lot more reasonable.

There are some things about the Donkey Party’s procedures that might help mitigate the risk of an accidental nominee. Most importantly, Democrats award delegates on a strictly proportional basis, making the occasional sweeps that helped Trump win in 2016 impossible. The Democratic practice of awarding nonelected “superdelegates”— if they preserve it — is also a hedge against a hostile-candidate takeover of the party.

But even if they have no reason to fear the fate that befell the GOP in 2016, Democrats should begin to think through the practical consequences of a very large presidential field. If nothing else, the two-tier debates Republicans were forced to undertake would be very controversial in a party where last year’s debate scheduling and formatting were a huge bone of contention. And an undifferentiated glut of candidates might be good for party unity but not so hot for voter interest.

Maybe one or two of the Big Three (it’s hard to envision all of them running) will enter the 2020 race and either lock up the nomination early or at least cull the field of electoral weaklings. If not, then just two short years from now, at the Iowa State Fair, the candidates may be so thick on the ground that you won’t be able to stir ’em with a stick.