Election Night is upon us, with all of its finger-gnawing anxiety, its cortisol-driven fear, and, for roughly half the country, the possibility of ecstatic relief after another surreal presidential campaign.

Results could take days, even weeks, to shake out. But the state of the race could also reveal itself surprisingly quickly. At 7 p.m. eastern time tonight, polls will close in the battleground state of Georgia. At 7:30 p.m., polls will close in North Carolina, another crucial toss-up. Both swing states are known for counting their ballots quickly, due to state laws that allow them to tally early and mail-in votes before Election Day.

[Read: Election anxiety is telling you something]

So when will we know the results, how can we sensibly extrapolate the early returns, and—perhaps most important—what information and analysis should we ignore? David Wasserman, a political analyst with the Cook Political Report, joined my podcast, Plain English, to explain how to watch the election returns like a pro—without falling for false hope or conspiracy theories. Here are three tips for following Election Night without losing your mind.

1. This might sound weird, but don’t expect this election to be as close as 2016 or 2020.

Wait, what? Aren’t Kamala Harris and Donald Trump essentially tied in national and swing-state polling averages? Didn’t Nate Silver put the odds that Harris will win this election at an exquisitely decimaled number between 50.00 and 50.99 percent? Isn’t there a nonzero chance that both candidates win 269 electoral votes?

Yes, yes, and yes. “This is the closest election in polling that I’ve covered in my 17 years, but that doesn’t mean it’s going to produce the closest result,” Wasserman told me. The 2016 and 2020 elections were absurdly close contests, both of them decided by about 78,000 votes. But, he said, “even elections as balanced as 2024 aren’t likely to hinge on 80,000 votes distributed across a handful of states.” Close polling does not predict historically close elections.

To understand what Wasserman means, perhaps a sports analogy is useful. Both sports betting and political polling try to express uncertain future events in the language of probabilities. The 2016 and 2020 elections were a bit like Super Bowls that went into overtime—something that’s happened only twice in the game’s six-decade history. Let’s say the next Super Bowl, in 2025, looks like a statistical dead heat, with two 13–4 teams with the exact same point differential. Let’s furthermore say that Vegas sportsbooks throw up their hands and declare the game a “pick-’em,” meaning neither team is favored to win. Even with all of this balance, it’s still very unlikely that the game will go to overtime, because so few games ever go to overtime. It’s the same with this election. We are still a normal polling error from either Trump or Harris winning the seven closest swing states, which would be a decisive victory.

[Brian Klaas: The truth about polling]

We don’t know how to forecast future events in any language outside of probabilities, and it’s hard to make peace with a world of probabilities. If you flip a coin 10 times, the median outcome is five heads and five tails. But you shouldn’t expect that 10 flips will yield five heads, because that outcome has less than a 25 percent chance of occurring. You’re actually three times more likely to get a number of heads other than five. So don’t get too invested in any particular electoral map. It’s very unlikely that your highly specific prediction will come to pass, and that includes an election decided by 80,000 votes.

2.  Ignore the exit polls.

Exit polls are exciting, because they provide a morsel of data during a highly anxious evening when audiences and news organizations are starving to know what’s going to happen in the next four hours, or four days. But there’s nothing particularly special about an exit poll. In many ways, it’s just another poll, but with a larger—and possibly misleading—sample. Exit polls might actually be less useful than other public-opinion surveys, Wasserman said, because the majority of voters now cast their ballots before Election Day.

If you’re watching a newscast that’s making a huge deal out of exit polls, it might have more to do with the need to fill time before we get actual election results. Rather, if you want to get an early sense of how things are trending on Election Night, the best thing to do is focus on county-level results that report the complete tally of votes. That means you’ll also want to avoid being overconfident about election results that are incomplete.

3. For the earliest bellwether counties, watch Nash, Cobb, Baldwin, and Saginaw.

By the end of the night, we’re likely to have nearly complete results from counties in Georgia, North Carolina, and Michigan. Here are a few to watch:

Nash County, North Carolina

If you’re looking for a coin-flip county in a coin-flip election, it’s hard to find a better one than Nash, just outside North Carolina’s Research Triangle. According to Wasserman, the county has been decided by fewer than 1,000 votes in every presidential race since 2004. In 2016, out of about 47,000 votes counted, Trump won by fewer than 100 ballots. In 2020, out of about 52,000 votes counted, Joe Biden won by fewer than 200 ballots. If Harris keeps Nash in the Democratic column, it would suggest that she can fight Trump to a draw in poorer areas while she racks up votes in Raleigh, Durham, and Chapel Hill.

[Listen: It could all come down to North Carolina]

Cobb County, Georgia

Metro Atlanta makes up most of Georgia’s vote, and Cobb County is packed with the sort of highly educated suburban dwellers who have shifted left in the Trump years. In 2012, Mitt Romney won Cobb by more than 12 percentage points. In 2020, Biden won the country by 14 points. For Harris to win the election, she’ll need double-digit margins in highly educated counties like Cobb across other swing states.

Baldwin County, Georgia

Although most eyes will be on Atlanta’s Fulton County, Wasserman told me that he’ll also be scrutinizing smaller and midsize Georgia counties, such as Baldwin County. Just outside Macon, in the middle of the state, Baldwin County is about 40 percent Black, and as a college town, it has a lot of young people. In 2016, Baldwin voted for Hillary Clinton by 1.7 percentage points. In 2020, Biden won it by 1.3 points. If Trump breaks through in Baldwin, Wasserman said, “it would be a sign that Harris is perhaps underperforming in both turnout and vote preference among younger Black voters and young voters” across the country.

Saginaw County, Michigan

How will we know if polls yet again undercounted Trump’s support among white men without a college degree? By looking at working-class counties like Saginaw, where Democrats won cycle after cycle before 2016. No Republican presidential candidate had won the plurality of votes in Saginaw since 1984, until Trump carried the county by just over one percentage point against Clinton, only for Biden to claw Saginaw back into the Democratic column by a mere 0.3 percentage points in 2020. “This is a place where organized labor powered Democrats to victory for many years,” Wasserman said. “If Trump wins Saginaw by five points, it’s going to be very difficult for Harris to overcome that.”

In a recent interview, the former Republican presidential candidate Vivek Ramaswamy made an offhanded comment that connected a few dots for me. Ramaswamy was talking with Ezra Klein about the potential for tens of thousands of government workers to lose their job should Donald Trump be reelected. This would be a healthy development, he argued. It could happen, he said, by reinstituting the Trump executive order Schedule F—which stripped certain civil servants of their job protections, allowing them to be fired more easily—and installing a government-efficiency commission to be led by Elon Musk. Ramaswamy said Trump should get rid of 75 percent of federal-government employees “on day one.” Up for debate, he argued, is whether some of those people would eventually be rehired. “That’s not the character of, certainly, what Elon did at Twitter, and I don’t think it’s going to be the character of what the most important part of that project actually looks like, which is shaving down and thinning down the bureaucracy.”

Ramaswamy’s invocation of Twitter is meaningful. In 2022, after acquiring the social network, Musk infamously purged Twitter’s ranks and fired 80 percent of its employees in the first six months, and then made a series of management decisions that ultimately threw the company into further financial disarray. Listening to Ramaswamy speak and hearing the respect in his voice as he cited the centibillionaire’s tenure, it became clear that he sees a blueprint for the Trump administration. Should Musk be appointed as a federal firing czar, it will likely not be because of his electric cars or rockets or internet-beaming satellites: It will be because he acted out the dream of draining the swamp, albeit on a smaller scale. Musk’s purchase of Twitter is not just a Republican success story; it is the template for the MAGA federal government. Even Musk’s mom said as much in a recent interview with Fox News: “He’s going to just get rid of people who are not working, or don’t have a job, or not doing a job well, just like he did on Twitter … He can do it for the government, too.”

Musk’s argument for gutting Twitter was that the company was so overstaffed that it was running out of money and had only “four months to live.” Musk cut so close to the bone that there were genuine concerns among employees I spoke with at the time that the site might crash during big news events, or fall into a state of disrepair. “I am fully convinced that if Musk does what he is saying he will do, it will be an absolute shitshow,” a trust-and-safety engineer at a different tech company told me in 2022. Musk did fire most of the trust-and-safety employees, as well as those in charge of curation and “human rights,” and the Machine Learning Ethics, Transparency, and Accountability team. The purge of these people in particular delighted some right-wing commentators, who saw Musk’s dismissals as a long-overdue excision of the woke bureaucracy inside the company. “Nothing of value was lost,” one MAGA account tweeted at the news of the firings.

[Read: I watched Elon Musk kill Twitter’s culture from the inside]

Twitter did not self-destruct as my sources feared it would (though parts of it have, perhaps most memorably when Musk tried to host Spaces events with Trump and with Florida Governor Ron DeSantis, only for them to glitch out). Small-scale disruptions aside, the site has mostly functioned during elections, World Cups, Super Bowls, and world-historic news events. But Musk’s cuts have not spared the platform from deep financial hardship. His chaotic managerial strategy for Twitter has been to rebrand the site as X, alienate many of its most important advertisers, institute a dubious paid subscription program, and dabble in AI features in the hopes of someday turning the platform into an “everything app.” The end result has been calamitous for the company’s bottom line. Soon after taking over, ad revenues plummeted 40 percent, and the bleeding hasn’t stopped. According to estimates, last year, X lost about 52 percent of its U.S. advertising revenue. A recent Fidelity report suggested that the company may have lost nearly 80 percent of its value since Musk bought it (for arguably way more than it was worth). If this keeps up, some have speculated that Musk may have to sell some of his Tesla stock to keep the company afloat. Musk’s financiers have also been left with massive loans on their balance sheets in what The Wall Street Journal has called “the worst buyout for banks since the financial crisis.”

Trump and Ramaswamy don’t seem to care about any of this. What matters is that Musk has turned X into a political weapon in service of the MAGA movement. X, as I wrote last week, has become a formidable vector for amplifying far-right accounts and talking points; it is poisoning the information environment with unverified rumors and conspiracy theories about election fraud. The far-right faithful do not care that his platform has occasionally labeled pro–Kamala Harris accounts as spam, temporarily banned journalists, restricted accounts that have tweeted the word cisgender, and complied with foreign-government requests to censor speech. Nor do Republican lawmakers seem to care that Musk is wielding his platform to get Trump elected, even after they spent the better part of a decade outraged that tech platforms were supposedly biased against conservatives. Their silence on Musk’s clear bias coupled with their admiration for his activism suggest that what they really value is the way that Musk was able to seize a popular communication platform and turn it into something that they can control and wield against their political enemies.

This idea is not dissimilar from the vision articulated by the Heritage Foundation’s Project 2025, the conservative policy proposal to reshape the federal government in a second Trump administration. Project 2025 is a dense, often radical, and unpopular set of policy proposals that, as my colleague David A. Graham notes, “would dissolve the Education Department and the National Oceanic and Atmospheric Administration, slash Medicare and Medicaid, ban pornography, establish federal abortion restrictions, repeal some child-labor protections, and enable the president to lay off tens of thousands of federal career workers and replace them with political appointees.” Put another way: If Trump were elected and decided to make Project 2025 a reality, his administration would take an existing piece of bureaucratic infrastructure, strip it of many of those who can check its power, and then wield that power to ideological ends and against their political enemies.

The parallels between this element of Project 2025 and Musk’s Twitter are stark. They should also be alarming. The federal government is not a software company, nor should it be run like one. Perhaps there is bloat in our departments and agencies, but civil servants labor over daily technical problems that are crucial to a functioning country—such as census taking, storm tracking, and preparing for pandemics. To simply cut these people with abandon (and replace others with political appointees) could have severe consequences, such as stifling disaster response and increasing the likelihood of corruption.

Consider also the financial dynamic. Last week in a virtual town hall, Musk said that the Trump administration’s second-term agenda—which includes tax cuts, slashing the federal budget, and tariffs on imports, “necessarily involves some temporary hardship,” but would ultimately result in longer-term prosperity. “We have to reduce spending to live within our means,” Musk added. The line is similar to his justification for the layoffs at Twitter, which at the time he called “painful” and necessary so that Twitter could balance its budget. But Musk bought the platform with no idea of how to turn it into a profitable business. His primary interest seems to be prioritizing shitposting and trolling rather than finding advertisers or making good on his ideas to turn X into a WeChat-style commercial app. Musk has never appeared interested in understanding the mechanics of a social network or the complexities of content moderation or even the specifics of the First Amendment. His incuriousness about the thing he ended up in charge of has been exceeded only by his desire to use it as a personal playground and political weapon.

Before Musk officially took over Twitter, the tech oligarch at least feigned an interest in running the company with an eye toward actual governance. “For Twitter to deserve public trust, it must be politically neutral, which effectively means upsetting the far right and the far left equally,” he tweeted in 2022. Trump, however, has made no effort to disguise the vindictive goals of his next administration and how he plans, in the words of the New York Times columnist Jamelle Bouie, to “merge the office of the presidency with himself” and “rebuild it as an instrument of his will, wielded for his friends and against his enemies.” In other words, he plans to run the Elon Musk Twitter playbook on the entire country.