Hard as it can be to define the best strategies for one’s party, it’s also imporant–and fun–to mock the other party’s strategic thinking. I had a chance to do that this week at New York:

Hanging over all the audacious steps taken so far this year by Donald Trump and his Republican Party has been the fact that voters will get a chance to respond in 2026. The midterm elections could deny the GOP its governing trifecta and thus many of its tools for imposing Trump’s will on the country. Indeed, one reason congressional Republicans ultimately united around Trump’s Big Beautiful Bill was the sense that they needed to get all the policy victories they could in one fell swoop before the tough uphill slog to a likely midterm defeat began. No one had to be reminded that midterm House losses by the president’s party are a rule with rare exceptions. With Republicans holding a bare two-seat majority (temporarily three due to vacancies created by deaths), the gavel of Speaker Mike Johnson must feel mighty slippery in his hands.

But if only to keep their own spirits high, and to encourage fundraising, Republican voices have been talking about how they might pull off a midterm miracle and hang on to the trifecta. A particularly high-profile example is from former RNC political director Curt Anderson, writing at the Washington Post. Anderson notes the unhappy precedents and professes to have a new idea in order to “defy history.” First, however, he builds a big straw man:

“[I]t’s always the same story. And the same conventional campaign wisdom prevails: Every candidate in the president’s party is encouraged by Washington pundits and campaign consultants to run away from the national narrative. They are urged to follow instead House Speaker Thomas P. ‘Tip’ O’Neill Jr.’s famous axiom that ‘all politics is local’ and to think small and focus on homegrown issues.”

Actually, nobody who was really paying attention has said that since ol’ Tip’s retirement and death. As Morris Fiorina of the Hoover Institution has explained, presidential and congressional electoral trends made a decisive turn toward convergence in 1994, mostly because the ideological sorting out of both parties was beginning to reduce reasons for ticket splitting. And so, returning to a pattern that was also common in the 19th century, 21st-century congressional elections typically follow national trends even in midterms with no presidential candidates offering “coattails.” So in making the following prescription, Anderson is pushing on a wide-open door:

“[T]o maintain or build on its current narrow margin in the House, the Republican Party will have to defy historical gravity.

“The way to do that is not to shun Trump and concentrate on bills passed and pork delivered to the locals, but to think counterintuitively. Republicans should nationalize the midterms and run as if they were a general election in a presidential year. They should run it back, attempting to make 2026 a repeat of 2024, with high turnout.”

Aside from the fact that they have no choice but to do exactly that (until the day he leaves the White House and perhaps beyond, no one and nothing will define the GOP other than Donald Trump), there are some significant obstacles to “rerunning” 2024 in 2026.

There’s a lazy tendency to treat variations in presidential and midterm turnout as attributable to the strength or weakness of presidential candidates. Thus we often hear that a sizable number of MAGA folk “won’t bother” to vote if their hero isn’t on the ballot. Truth is, there is always a falloff in midterm turnout, and it isn’t small. The 2018 midterms (during Trump’s first term) saw the highest turnout percentages (50.1 percent) since 1914. But that was still far below the 60.1 percent of eligible voters who turned out in 2016, much less the 66.4 percent who voted in 2020. Reminding voters of the identity of the president’s name and party ID isn’t necessary and won’t make much difference.

What Anderson seems focused on is the fact that in 2024, for the first time in living memory, it was the Republican ticket that benefited from participation by marginal voters. So it’s understandable he thinks the higher the turnout, the better the odds for the GOP in 2026; that may even be true, though a single election does not constitute a long-term trend, and there’s some evidence Trump is losing support from these same low-propensity voters at a pretty good clip. At any rate, the message Anderson urges on Republicans puts a good spin on a dubious proposition:

“The GOP should define the 2026 campaign as a great national battle between Trump’s bright America First future and its continuing promise of secure borders and prosperity, versus the left-wing radicalism — open borders and cancel culture or pro-Hamas protests and biological men competing in women’s sports — that Democrats still champion. Make it a referendum on the perceived new leaders of the Democratic Party, such as far-left Reps. Jasmine Crockett (Texas) or Alexandria Ocasio-Cortez (New York).”

Without admitting it, Anderson points to the single biggest problem for Republicans: They don’t have a Democratic incumbent president or a Democratic Congress to run against. Jasmine Crockett is not, in fact, running in Pennsylvania, where she is likely unknown, and even AOC is a distant figure in Arizona. Democrats aren’t going to be running on “open borders and cancel culture or pro-Hamas protests or biological men competing in women’s sports” at all. And Republicans aren’t going to be running on “Trump’s bright America First future” either; they’ll be running on the currently unpopular Trump megabill and on economic and global conditions as they exist in 2026. Democrats could benefit from a final surge of Trump fatigue in the electorate and will almost certainly do well with wrong-track voters (including the notoriously unhappy Gen-Z cohort) who will oppose any incumbent party.

Whatever happens, it won’t be a 2024 rerun, and the best bet is that the precedents will bear out and Republicans will lose the House. A relatively small group of competitive races may hold down Democratic gains a bit, but unless an unlikely massive wave of prosperity breaks out, Hakeem Jeffries is your next Speaker and Republicans can worry about what they’ll do when Trump is gone for good.

There’s been a lot of buzz about the world’s wealthiest man pledging to start a third party, so I addressed that dubious proposition at New York:

The feud between Donald Trump and his onetime deep pocket and henchman Elon Musk keeps bubbling up in unpredictable ways. But one fracture point that is potentially bigger than an exchange of insults and conspiracy theories is the Tech Bro’s musings about creating a third national political party. Not because there’s any real popular demand for another party but because Musk’s wealth could give even the dumbest idea wings.

This angle is interesting in part because Trump has himself flirted with third-party talk when it suited his purposes. But you wouldn’t know that from his categorical put-down of Musk’s fantasies over the weekend at Truth Social:

“I am saddened to watch Elon Musk go completely ‘off the rails,’ essentially becoming a TRAIN WRECK over the past five weeks. He even wants to start a Third Political Party, despite the fact that they have never succeeded in the United States – The System seems not designed for them. The one thing Third Parties are good for is the creation of Complete and Total DISRUPTION & CHAOS, and we have enough of that with the Radical Left Democrats, who have lost their confidence and their minds! Republicans, on the other hand, are a smooth running “machine,” that just passed the biggest Bill of its kind in the History of our Country.”

He went on to brag some more about his megabill and to spitball about why Musk might have opposed it, without mentioning Musk’s own argument that it is a debt and deficit nightmare.

The third-party threat was clearly weighing on the 47th president’s mind this weekend. When asked about it by a reporter earlier on Sunday, Trump said, “’I think it’s ridiculous to start a third party,” later noting, “He can have fun with it, but I think it’s ridiculous.”

Musk has fleshed out his fantasy a bit after getting the inevitable endorsement of his efforts from his personal echo chamber on X:

“One way to execute on this would be to laser-focus on just 2 or 3 Senate seats and 8 to 10 House districts.

“Given the razor-thin legislative margins, that would be enough to serve as the deciding vote on contentious laws, ensuring that they serve the true will of the people.”

The idea, then, isn’t to launch a new party through some big, splashy presidential campaign that will capture what Musk has called the “80 percent in the middle” of voters alienated by the Democratic-Republican “uniparty.” That, as it happens, was the vision of the last real third-party builder, Ross Perot, who never made much of an effort to create an alternative ballot line at the state level. Perot failed in no small part because winning or even threatening to win elections in a first-past-the-post system requires the sort of regional voting base he never enjoyed. The more limited strategy Musk seems to be talking about doesn’t require displacing a national party but instead simply exploiting the close competitive balance of the existing two major parties and seizing the margin of control in Congress for leverage purposes. It’s a down-ballot version of what southern segregationists tried to do with regional tickets in the 1948 and 1968 presidential elections: prevent either major-party candidate from gaining a majority in the Electoral College and then shake the parties down for policy concessions. They didn’t fail by much.

So what would Musk’s new party, which he has dubbed the “America Party,” make its be-all-and-end-all demand? Best we can tell, he wants massive reductions in the size and cost of the federal government, along with the attendant public debt. That’s not only a slender reed for a disruptive third party but it’s at least rhetorically identified with the GOP despite that party’s own spotty fiscal record. From a practical point of view, why would some aspiring deficit hawk in any given state or congressional district want to take a flier on a candidacy under the America Party banner when they could just as easily run as a Rand Paul–Thomas Massie fiscal hard-liner in a Republican primary? The only answer I can think of is that it may be a way to gain access to Musk’s money. And it’s unclear at this point how much of his fortune Musk is willing to devote to this effort.

As Nate Silver points out, if Musk could lavishly finance a new party with a broader agenda than bringing back DOGE — say, developing a national AI strategy that could prevent rather than accelerate demolition of the workforce — it might gain some purchase, particularly with young voters who dislike both major parties. But it would require the sort of patience and political sophistication Musk has not in any way displayed up to this point in his career.

More likely, Musk is just the latest in a long list of political amateurs who look at unhappiness with the two-party system and make two major mistakes: (1) they don’t grasp that most self-identified independents are what Silver calls IINOs, independents in name only, who routinely vote for the same major party even when given alternatives; and (2) they assume all these people share the same grievances with the current party system.

The only demonstrated template for third parties in the U.S. is to address an entirely unmet demand. When Republicans broke through in the late 1850s, they were exploiting a situation in which one major party (the Whigs) had already died and the other could not stake out a national position on slavery. At this point, Musk isn’t offering anything voters can’t find in the right wing of the Republican Party or, barring that, in the Libertarian Party. So Trump is correct to argue that his frenemy has “gone off the rails.”

After months of watching and writing about Trump’s huge budget reconciliation bill, I wrote my final assessment today…but then saw a poll that made me rethink the whole thing, and wrote that up at New York:

When top House Democrat Hakeem Jeffries chose to exploit a loophole in the House rules, which allows party leaders to talk as long as they want, to discuss at record length the baleful effects of Donald Trump’s One Big Beautiful Bill, it initially looked like a bit of a publicity stunt, albeit it a good one. It delayed for hours a very big moment of Republican self-congratulation over the final passage of this enormous package of legislation. It probably screwed up a lot of congressional flight arrangements to get members home for Independence Day. And it likely put a few kinks into plans for Trump’s own festive July 4 signing ceremony, wherein the president will surely praise himself, thank his vassals, and tell more than a few fibs about what his grim masterpiece will do.

But beyond all that, it’s becoming clear that Democrats need to do a much better job articulating their take on this bill and its profoundly reactionary effects on the social safety net. To those of us whose job it is to listen to arguments over 940-page bills as they chug through Congress for months, it may seem like congressional Democrats have been grinding away at the message that Republicans are cutting Medicaid to give a tax cut to billionaires. Some of them held up signs about Medicaid cuts when Trump delivered his belligerent address to a joint session of Congress four months ago. Yet a startling new poll shows the extent to which that messaging — and for that matter, Republican messages hailing the megabill as a people-pleasing growth engine for the U.S. economy that doesn’t touch Medicaid benefits at all — isn’t really breaking through, as Sam Stein reported:

Meanwhile, the kind of people Democrats need to reach but often don’t aren’t hearing much at all:

“73% of 2024 Trump supporters who didn’t vote in 2020 and 56% of Biden-to-Trump flippers have heard nothing about the bill. These shares are 20 points higher than their Harris supporting counterparts — indicating the urgent opportunity if Democrats break out of our own media silos.”

Part of the problem, to be sure, is that Trump’s megabill is incredibly broad and complicated, and the budget reconciliation process by which it was developed, debated, and enacted is insanely complex and obscure. It’s all about as remote from the civics-book understanding of how laws are made as you can get, and it has been understandably difficult for Democrats to describe it compellingly in a sound bite, a protest sign, a TV ad, or indeed, in Jeffries’s eight-hour speech. It was designed that way, and that’s why half the public isn’t absorbing anything about it, and a lot of others are simply processing it via big, vague party-driven narratives.

The bottom line is that the struggle to define this consequential legislation has just begun. For Democrats, finding ways to convey the horror the megabill inspires in those who have studied it closely, and the concrete damage it will do to actual people, must continue right up until the midterm elections. Yes, Trump and his allies will do many other things that might galvanize voters, from his reckless foreign policies to his cruel mass-deportation initiative to the lawless conduct he exemplifies and encourages among his appointees. But nothing is likely to match the megabill in magnitude or in the malignancy of its authors. If voters march to the polls in 2026 or 2028 with no better than a rough idea of what it means, America will get more of the same.

After writing my umpteenth explainer of this or that legislative, executive or judicial-branch development of 2025, it hit me there was an obvious pattern that I discussed at New York:

Systemic chaos and confusion have been the reigning leifmotif of the second Trump presidency. If your grasp of the way government is supposed to work is based on Schoolhouse Rock or social studies lessons on “how a bill becomes law,” the last five months or so must have been baffling to you. It’s not an accident, either: Donald Trump’s ways of doing business make the normal business practices of the public sector all but impossible.

Congress has done remarkably little this year other than slowly and erratically tinker on the so-called One Big Beautiful Bill. When reconciliation bills were conceived of in the early ‘70s, they were meant to provide some budgetary discipline to Congress. Over time they have become the principle way for majority parties with governing trifectas to enact massive collections of legislation without fear of a Senate filibuster. Reconciliation bills, put simply, are a way to take power from the minority party. They are tailor-made for a president with authoritarian tendencies and largely undivided party support to ram through a vast agenda. And so, unsurprisingly, the 2025 reconciliation bill is bigger and badder than any of its predecessors. During his first term in 2017, Congress packaged Trump’s agenda into two reconciliation bills. This time he insisted on doing everything at once, creating a truly monstrous megabill.

Budget reconciliation bills have their own incredibly obscure and complex set of rules and procedures that confound everyone but a small tribe of budget wonks. There are multiple layers to the development of a megabill, from a budget resolution that sets targets, to “reconciliation instructions” that tell House and Senate committees what to produce in the way of revenues and spending, to all-night Senate “vote-a-ramas” where the minority party gets to offer dozens of symbolic amendments designed to make the majority look bad. Once the assembled bill reaches the Senate, that chamber’s nonpartisan parliamentarian rules on which provisions fail to meet the criteria for inclusion in such bills. Then the majority party will try to put Humpty Dumpty back together by tweaking the offending provisions, or altering other items to make the overall arithmetic work. All this is happening very quickly, and largely behind the scenes. And then the House has to accept changes made by the Senate or the whole wild process more or less starts over.

Figuring out where things stand at any given moment with a reconciliation bill is all but impossible, so people rely on big partisan narratives. To Republicans, Trump’s Big Beautiful Bill is a carefully wrought device to extend and create tax cuts for people who deserve them; to boost national defense and border security spending; to reverse the “open borders” policies of the Biden administration; and to reduce government “waste, fraud and abuse.” To Democrats it’s an abominable reverse-Robin Hood measure to give tax cuts to the wealthy at the expense of vulnerable poor and middle-class Americans, whose Medicaid, Obamacare, and SNAP benefits will be slashed. Each party accuses the other of fiscal profligacy. Sometimes Republicans jockeying for leverage in their own caucuses temporarily agree with Democrats and yell and posture, but it’s understood that in the end they will toe the party line. No one will read the final bills that Trump will likely push through the House and Senate. So how can citizens comprehend the consequences until the dust clears and the new laws are fully implemented? They really can’t; it’s all based on partisanship, trust, selective information, and gut instincts. Civics in action it isn’t. And it’s all being done this way because Trump has insisted on making the Big Beautiful Bill a testament to his own power and historical significance.

The chaos in the judicial branch is arguably even worse. Since the Trump administration is determined to expand presidential powers to the breaking point, it is inviting and sometimes initiating litigation in the federal courts on a vast array of issues. In general challenges to Trump’s policies have succeeded more often than not in the lower courts. But many issues will only be resolved when they reach a deeply divided U.S. Supreme Court. Indeed, on the final day of that Court’s current term, the conservative majority restricted the ability of lower-court judges to put a hold on the administration’s many power grabs through nationwide injunctions. So more than ever, all roads lead to incredibly high-stakes decisions by the Supreme Court, three of whose members owe their life-time appointments to Trump.

That same Supreme Court, moreover, is increasingly prone to issuing temporary orders with no explanation. This often helps the administration work its will, but leaves a lot of questions unresolved, as the Brennan Center recently explained:

“Today, the justices grant relief in contentious shadow docket cases twice as often as they did just a few years ago. The surge in issuing this relief has coincided with Justices Neil Gorsuch, Brett Kavanaugh, and Amy Coney Barrett joining the Court.”

It’s hard to avoid the suspicion that Supreme Court conservatives are trying to hide their tracks on controversial cases. Just this week the Court permitted the administration to resume deporting immigrants to random and often dangerous countries other than their own without a hearing — and the decision was made with no oral arguments and no majority opinion. Who knows what it means in the long run? Those without a law school education and a huge amount of time on their hands will be lost trying to follow it all.

On top of everything else, the Trump 2.0 executive branch has engaged in an incredible amount of secrecy and procedural chaos, all of it making it hard to know what will happen to the federal government and millions of federal employees. Elon Musk’s DOGE initiative may be winding down a bit with its chief’s recent departure from the administration. But the damage it wreaked throughout the public sector via arbitrary mass firings, lockdowns of agencies and programs, refusals to pay invoices, and sheer terror tactics by embedded teams of totally inexperienced data geeks, will live on for years. And anyone relying on, much less working on, federal programs and benefits will be seeing shadow ninjas jumping out of the walls to blow it all up for a long time.

Post-Musk, Office of Management and Budget Director Russell Vought has emerged as the real quarterback of executive branch aggressive chaos. The steely Christian nationalist and Project 2025 co-creator wants to shrink the federal government radically. He’s renowned for pursing all sorts of obscure power-grabbing measures that only a handful of people fully understand, including spending rescissions (cancellation of previously appropriated funds, which must be approved by Congress), spending deferrals (delays in releasing previously appropriated funds, which can be overriden by Congress), “pocket rescissions” (deferrals issued just before the end of the fiscal year so that Congress won’t have time to override them), and impoundments (blatant refusals to spend appropriated dollars). It’s all in play.

Top it all off with a powerful president who is famously erratic, and it’s hard to see the road ahead, or even the road underneath one’s feet right now. To understand any given issue you need to know where it stands in the latest version of the budget bill; whether it’s subject to federal court litigation; what the executive branch is doing to undermine the prerogatives of Congress and the courts; and what priorities Trump may choose at any given moment. It’s a lot, every minute of every day, and the relentless chaos fostered by the 47th president continues to defy every single precedent.

As an inveterate defender of the value of public opinion research, I sprang into action at New York when critics mocked pollsters for getting the NYC Democratic mayoral primary wrong:

One of the reasons Zohran Mamdani’s smashing victory yesterday created such a sensation is that public polls (overall) did a poor job of predicting the size and shape of both his and Andrew Cuomo’s coalitions in the Democratic mayoral primary. An average of these polls prepared by Race to the WH showed Cuomo with 36.4 percent of first-choice ballots and Mamdani with 28.6 percent. While nearly all of these polls showed Cuomo being pushed into a ranked-choice battle with Mamdani, and a few (notably the final survey from Emerson College) showed the democratic socialist prevailing in the final calculation, only one public survey, from Public Policy Polling, predicted a first-choice Mamdani plurality. No one provided evidence of a Mamdani plurality so large that his nomination was certain without waiting for the final ranked-choice numbers.

The polling miss will be a source of great joy among those who think polling plays too great a role in contemporary politics — or that would like to get rid of polling altogether. The idea that we should rely on hunches, spin, prejudice, and tiny-sample reporting to get a handle on public opinion rather than trying to measure it objectively strikes me as, to use a technical term, willfully stupid. But partly to rebut polling nihilists, and to promote a better understanding of what polls can and cannot be expected to do, it’s worth a look at why they were wrong in New York. Here, a few reasons:

Polls are snapshots of fast-moving races

This explanation may seem too obvious to require explanation, but highly dynamic multicandidate contests often take shape late in a cycle, when ads and get-out-the-vote efforts have reached their maximum impact in both persuading and mobilizing voters. Polls often can’t capture such trends, which is why we often mystify them with terms like momentum. Only a few pollsters (e.g., Marist, Data for Progress, and Emerson) did multiple surveys, making apples-to-apples comparisons possible. There was clearly a Mamdani trend underway, but its size and durability was hard to nail down.

Primary polling is always shaky

It’s a truism in the political-analysis biz that polling for general elections is typically more accurate than polling for party primaries. The reasons are simple enough: Much of the electoral behavior we see each November is entirely predictable thanks to partisan patterns that recur time and time again and change slowly, if at all. Candidate choice in primaries is far less mechanical and thus is more fluid and harder to capture. So fairly sizable polling errors in primaries are actually normal, not a sign that they are for some reason becoming useless or misleading. You should just add a few grains of salt to any primary poll.

Turnout in a race like New York’s is extremely hard to predict

All polls are based on models that make certain assumptions about the shape of the electorate, which in turn depends on turnout patterns. There’s nothing much more difficult to predict than who will show up for a municipal primary in late June of a non-presidential, non-midterm election year. For all the local and national hype, heavy campaign spending, and genuine excitement associated with the New York mayoral primary, turnout was only about 30 percent. The value of precedents was limited; for one thing, the advent of early-voting opportunities (a relative novelty in New York, where there was no early voting prior to 2019) has added a new variable to turnout and likely helped boost the participation of the younger voters who backed Mamdani so decisively. More fortuitously, the extremely high temperatures that afflicted the city on Tuesday likely depressed turnout among the older voters central to Cuomo’s prospects for victory.

Ranked-choice voting complicates everything

This was only the second mayoral primary conducted under the controversial ranked-choice voting system, which allows voters to express second-through-fifth-place candidate preferences, with a postelection calculation eliminating lower preferences until someone wins a majority. While there has been a lot of talk about how RCV might affect who won or lost, it’s equally important to comprehend that the system has a complex and hard-to-measure impact on campaign strategies (especially the cross endorsements that Mamdani used effectively) and how voters process their own decisions, particularly when everyone is still a novice at utilizing RCV. A lot of late-deciding voters were just figuring out how RCV worked, which very likely affected how they responded to polls.

Maybe 2025 is a bad year for pollsters

It’s almost always a bad idea to generalize from any one contest, or any one election cycle, that “the polls” (another generalization) are good or bad. For one thing, they are always useful in providing analytical data, even if they’re “wrong” about predicting winners or winning margins. Some data is better than no data at all. Beyond that core dogma, which cannot be repeated too often, the reality is that polls are more accurate in some election years than others and vary by the type of race involved. Most famously, the 2016 presidential-general-election polls were pretty accurate at the national level but erred decisively in the battleground states that unexpectedly vaulted Donald Trump to the White House. Similarly, 2024 general-election polls were reasonably accurate, but a slight underestimation of Trump’s vote in the battleground states, which tilted all of them into his column, made them seem off. 2020, by contrast, was a “bad” presidential election for polls, in part because the pandemic introduced a lot of uncertainty into turnout patterns and candidate preference alike.

Polling in non-presidential election years is harder to assess because they are less frequent and consistent, and the only “national” numbers predict the overall House popular vote, which is not that tightly related to the number of seats won or lost. But 2022, to cite the most recent example, was a “good” midterm election for polls. There’s no telling what the scattered election landscape of 2025 will ultimately show (there’s no public polling at all in many off-year races), and pollsters are constantly refining their methodologies as well. So the smart thing to do is to wait a beat, or two, or three before making any sweeping judgments about what the New York City results mean for the relevance of polls.