These are perilous times for understatement and modest expectations. In the age of Trump, even the smallest of things are transmogrified into epoch-defining events. These are the days of mountains out of molehills, “a new low” almost daily, and more proof (as if more were needed) that your political opponents are every bit as debased as—no, even more debased than—you rightly concluded long ago.

In keeping with the times, many now detect a strong whiff of impeachment in the fetid Washington air. And it is here that I would like to apply a little critical political realism to the question, to set aside personal views and analyze it as coolly and dispassionately as possible. I’m sure there’s still an audience for that sort of thing. But just in case there isn’t, let me begin by saying that the idea Donald Trump is going to be removed from office is about the most farfetched fantasy in the rich history of Washington partisan delusion.

To return to measured understatement, the likelihood of such an outcome is not zero. But if you examine the hypothetical chain of events that would produce Trump’s removal, you will find not a president barely maintaining his balance atop a house of cards sure to collapse at any moment, but rather a confluence of constitutional procedures and political calculations that will work to keep him in the office to which he was elected certainly for the next two years, and very likely for the rest of his term.

Most Democrats and their boosters, as well as the more ideologically diverse Never-Trumpers, regard the 45th president as singularly unqualified and unsuited for the office, if not worse. This has been true since he emerged as the likely GOP nominee for president, and indeed, as Jonathan Allen and Amie Parnes’s Shattered: Inside Hillary Clinton’s Doomed Campaign makes abundantly clear, this judgment about his unacceptability led to the conclusion that he was therefore unelectable, which in turn drove the Clinton campaign’s message to voters. This might have been a perfectly good strategy in an election set to produce a 60-40 landslide à la LBJ-Goldwater, but anti-Trump fantasies to the contrary notwithstanding, 2016 was never that election. Democrats assumed their conclusion—unacceptable equals unelectable—in the most disastrous political wishcasting in living memory.

Trump won, but Democrats and Never-Trumpers have not changed their view: He remains every bit as unqualified and unsuited to office as he has always been. Besides which, they rationalize, he did not win the popular vote, and his margin of victory in the three states that put him over the top in the Electoral College was minuscule, and the Russians interfered to help him and discredit Clinton, and FBI director James Comey unfairly announced the reopening of the probe into Clinton’s emails two Fridays before the election, and so on. The conclusion of unfairness and therefore of illegitimacy was an understandable response to the dislocation caused by Trump’s victory in the face of the certain conviction of his opponents that he would lose.

When Trump took office and continued tweeting and being, in general, himself, taking presidential politics into a new era, he gave his opposition no reason to view him any differently—not that they were at all inclined to in their own new era of “Resistance.” And Trump’s decision to fire Comey, the appointment of Robert Mueller as special counsel investigating the Russia connection in the election, and a steady drip of leaks and revelations have combined to create, in the minds of many Democrats, a uniquely vulnerable occupant of the Oval Office. One more something, they believe—Trump firing Mueller? an indictment of family members or friends and associates?—and he’s a goner.

But how, exactly, is that supposed to happen? I have seen no plausible scenario put forth in answer to that question. Even if Democrats are correct in every judgment they make about Trump’s unsuitability for the office he won, they seem stuck in the “Have you no honor, sir?” phase of opposition. They are calling on Republicans to denounce Trump and all his works, and when Republicans decline to do so, Democrats lambaste them for “normalizing” or enabling Trump or worse. Democrats have also been warning Republicans that standing by or at least in proximity to Trump will be political suicide, as soon as 2018. Numerous articles, meanwhile, have pondered at length the supposed delicacy of the position of Vice President Mike Pence, whose fate, in this vision, is to become the 46th president of the United States under the worst possible circumstances. With the release of the “I love it” Donald Trump Jr. email in response to an offer of Russian dirt on Hillary, Democrats have also been able to recruit a few more GOP-leaning commentators to the condemnation.

All of this, however, is sound and fury signifying not very much. Democrats are having a conversation with themselves, mutually reinforcing the conclusions they have already reached and validating their far-flung speculation by repetition. What really matters is where the power to affect events lies and the likely calculations of those wielding that power. So we need to be thinking not so much about what Democrats are saying, but about what Republicans are likely to do—specifically the elected Republican officials who control the White House, the House of Representatives, and the Senate.

There are five ways a president can manage not to finish a term: death; resignation; uncontested removal by the vice president and the majority of the cabinet for the inability “to discharge the powers and duties of .  .  . office” (the 25th Amendment); presidentially contested removal by the vice president and cabinet majority under the 25th Amendment, which then goes to Congress and requires agreement by two-thirds majorities in both the House and Senate; and Congress’s impeachment process, which entails House majority approval of one or more articles of impeachment, a trial in the Senate over which Chief Justice John Roberts would preside, and a two-thirds Senate verdict in favor of removal.

I will leave consideration of death in office to the Kathy Griffins of the Resistance. The 25th Amendment scenarios require wholesale mutiny of a cabinet the president himself appointed, which is unlikely, to put it mildly, at least in the absence of some wildly mad act, such as attempted misuse of a nuclear weapon. In any case, the relevant 25th Amendment scenario is the one in which the president contests the decision of his cabinet. That yields an even higher threshold for removal than the impeachment process requires: two-thirds of both legislative chambers. If impeachment and removal is highly implausible, removal via the 25th Amendment is more so. In the only case of presidential resignation in U.S. history, that of Richard Nixon, his decision was intimately connected to the impeachment proceedings underway in Congress in 1974. So to impeachment we shall go.

Why would the GOP majority in the House vote to impeach Trump? The constitutional standard is “high crimes and misdemeanors,” and while there may be little doubt in the minds of many Democrats that Trump is guilty of conduct that could be so classified, no GOP member of the House of whom I am aware has taken that view. Moreover, while in a functional sense “high crimes and misdemeanors” means whatever the House decides it means, in the two previous impeachments, those of Andrew Johnson in 1868 and Bill Clinton in 1998, the House acted on the basis of presidential conduct that a majority believed was actually illegal. In the Johnson case, it was the president’s violation of the Tenure of Office Act by removing Secretary of War Edwin M. Stanton without the consent of the Senate (a power the Supreme Court would later rule was indeed the president’s alone). A previous attempt to impeach Johnson, in December 1867, failed by a vote of 57 in favor, 108 against, with 66 Johnson-hating Republicans refraining from voting to impeach him because no genuine violation of the law was at issue. For Clinton, the crimes alleged were “process” crimes: perjury and obstruction. But criminal acts they were and are.

Now, even top House Democrat Nancy Pelosi has said that she has not yet seen enough evidence to warrant Trump’s impeachment (though she seems to be hoping). But let us recall that she is the minority leader. The majorities in the House voting to impeach Johnson and Clinton were of the opposition party. So was the House majority when its Judiciary Committee voted to approve articles of impeachment against Nixon. Why would anyone think a Republican House majority would vote to impeach a Republican president because Democrats have been leading the charge on his unfitness for office for more than a year? It would likely take more, even, than a “process” crime such as obstruction of justice for the House GOP to turn against him. The threshold for action is simply higher when it’s R on R.

But perhaps enough GOP defectors could join Democrats to produce a majority for impeachment. Here, however, we have to consider the course impeachment proceedings would take in the House. They would start in the Judiciary Committee, which would have to decide to begin an investigation, conduct it, and then approve articles of impeachment to send to the House floor. The Judiciary Committee has 24 Republican members and 17 Democrats. Again, it’s a bit of a mystery why Republicans on the committee would agree to begin an inquiry. Chairman Bob Goodlatte has been staunchly opposed to having the Judiciary Committee look into the Comey firing (the committee has oversight jurisdiction for the FBI). He has cited the ongoing special counsel investigation as reason not to proceed with a committee “fishing expedition,” as he called it. Given the quite conservative composition of the Judiciary Committee, it’s also difficult to imagine Democrats recruiting four Republicans to the cause for a majority, to say nothing of the formidable procedural hurdles of moving against the wishes of the chairman.

No Judiciary Committee investigation, no impeachment. But suppose Special Counsel Mueller concludes his investigation by dropping a bombshell report on the committee, as Whitewater independent counsel Kenneth Starr did on Bill Clinton in 1998. Well, first of all, we are assuming Mueller finds criminal activity on the part of the president, which is a tall assumption. Clinton got snared when the Supreme Court ruled he had to testify in Paula Jones’s civil suit against him alleging sexual harassment; he lied under oath denying a sexual relationship with Monica Lewinsky. Trump can avoid that snare either by answering questions truthfully or (probably a safer legal strategy) declining to answer at all.

And about that report: Note that Starr was appointed pursuant to the now-defunct Independent Counsel Act, which explicitly charged him to provide the House with any “substantial and credible information .  .  . that may constitute grounds for an impeachment.” Mueller has no such authority under his appointment. His investigation must comport with the standards applicable to all U.S. attorneys, which don’t allow for public reports. The provisions of his appointment also hold: “At the conclusion of the Special Counsel’s work, he or she shall provide the Attorney General with a confidential report explaining the prosecution or declination decisions reached by the Special Counsel.”

If Mueller concludes that Trump himself has committed a crime (for example, obstruction of justice), what will he do? The view of most legal scholars and of the Office of Legal Counsel in the Justice Department is that he would not have the constitutional authority to indict a sitting president, which would be without precedent and a proposition the Supreme Court has never directly considered. Starr’s office, however, concluded differently, on the basis of a lengthy if tendentious memorandum by legal scholar Ronald Rotunda, which recently became public through a New York Times Freedom of Information Act request. Nevertheless, Starr declined to pursue this avenue, opting to provide Congress with his report on impeachable offenses instead. The Watergate special counsel had similarly asserted his office’s authority to indict Nixon, but declined to do so.

Because in these hyperpartisan times, restraint is outré and bad behavior tends to beget worse behavior, I wouldn’t rule out a Mueller decision to go where no prosecutor has gone before, in which case we will have an interesting matter for the Supreme Court, which will not welcome the opportunity to clarify matters. But Mueller might also conclude (in a genuine or tactical show of restraint) that he doesn’t have the authority to indict, and explain as much in his confidential memo on prosecution decisions. He would do so in the knowledge that the odds of his report remaining confidential are, what? Zero percent? Zero point one? So when it leaks, a report explaining that he has declined to prosecute the president only because Trump is president would surely be explosively received. Wouldn’t that likely galvanize the same sort of response as the Starr report?

Close, but no cigar. The House Judiciary Committee of 1998 had no stomach for its own impeachment investigation of Clinton, but it was a willing and even eager recipient of Starr’s report. The committee in 2017 or 2018 would be neither an official recipient nor eager. Goodlatte and Republicans would face tremendous pressure, but not in response to an official act of the special counsel. And it’s still entirely in the GOP-controlled committee’s power not to proceed.

There would also likely be strong skepticism should the special counsel choose to prosecute others on conspiracy charges and name the president “an unindicted co-conspirator,” as Nixon was in Watergate criminal proceedings. The U.S. Attorneys’ Manual advises against the practice: “In the absence of some significant justification, federal prosecutors generally should not identify unindicted co-conspirators in conspiracy indictments,” preferring reference to them as “person or persons known.” The Supreme Court hasn’t spoken on the question, but a 1975 Fifth Circuit ruling, U.S. v. Briggs, severely criticized the practice as a violation of due process and ordered that the names be struck from the indictment at issue. Once again, I find it hard to imagine a special counsel exercising restraint: Viewed a certain way, the “public interest” would provide “significant justification” in a case involving the president. But again, his defenders would have a plausible complaint to press about an abusive process.

So it is that from an institutional and constitutional perspective, the route to the floor of the House for impeachment charges is a very difficult one. Now let’s add the GOP political perspective.

Democrats seem to have in mind the bottom falling out from Trump’s political support. If rank-and-file Republicans turn on him, that will enable, even encourage, defections on a potentially large scale among elected Republican leaders. In the Democrats’ perfect world, congressional Republicans will conclude that continued support for Trump, or even a studied neutrality, will drag them under, and so they will cut the president loose.

The model is more or less Nixon: As the Judiciary Committee voted articles of impeachment in 1974, with full House approval a foregone conclusion, GOP emissaries advised Nixon that he would lose in the Senate. He resigned. That was a happy ending for Democrats.

The difficulties here start with the hypothesized collapse in GOP support for Trump. So far, he has maintained a strong base of Republican support, and it is not for want of efforts by Democrats and their allies to undo it over the past year or more. The partisan environment and the media environment are very different from those of 1974. Political scientists say the parties have now fully “sorted” themselves ideologically; liberal Republicans have all but ceased to exist, and “conservative,” even “moderate” (as opposed to “liberal” or “progressive”) Democrats are a species spotted only during campaign season. And I think there is a wee twinge of nostalgia at the major mainstream news outlets these days for the time when they could gang up on a president and bring him down (with plenty of help from Nixon himself, to be sure). They don’t seem to be able to do that now (one of the major lessons of 2016, as it happens), notwithstanding the view of some commentators that Trump has already proved himself to be worse than Nixon. For a significant segment of the Trump-supporting population, Democratic (and media) opposition to Trump is all that’s necessary to kindle reinvigorated support for their man. I would only add that by now, politicians have had ample experience operating in an environment of abysmal approval for themselves and the institutions in which they serve.

Perhaps we are only one more bombshell revelation away from a collapse. Or one more seemingly suicidal move by the president—a decision, say, to defenestrate Mueller and along with him the leadership of the Justice Department. But would it be a permanent collapse, or would Trump think he could rally his base back? Things looked surpassingly bleak for Bill Clinton the day the Monica Lewinsky story appeared on the front page of the Washington Post. Clinton kept his focus, however. He told his staff in no uncertain terms that there was no possibility of resignation. He denied the story, lying first to his staff, then the American people, then under oath. He deployed the women in his cabinet to vouch for him. And by the time the DNA evidence on the blue dress (whoops) proved conclusively he was lying, he had prepared an effective counterattack on the investigators’ abusive process. Let’s just say there is a precedent for toughing it out.

And about those Republicans on Capitol Hill: They’re supposed to be worried about their continued support for Trump dragging them to oblivion. But compared to what? They will surely weigh the risks of dumping Trump as well, something Democrats tend to forget. And here Nixon looms large once again, though to quite opposite effect.

Republicans lost 48 seats in the House and 5 in the Senate in the congressional elections of 1974, three months after Nixon’s resignation. This is a data point of some importance. Could it have been worse if they stuck by him? Who knows? But the point is that a 48-seat loss is no advertisement for the advantage of turning against your president; doing so might just produce the result you seek to avert.

Ultimately, to remove the president, you need two-thirds of the Senate. For all the reasons above, I don’t think the question comes up. But this is the bedrock reason for predicting that Trump remains in office: You need 67 votes in the Senate to remove him. Democrats have 48 votes, and let’s make the assumption they are unanimously in favor of removal—in other words, that the 10 Democratic senators up for reelection in 2018 in states Trump won don’t think they have anything serious to worry about in voting to oust him or are prepared to do so anyway. So you need 19 Republicans, a little more than a third of the caucus, to defect. Voting to remove a president of your own party is a grave act. And it bears noting that 49 of the 52 GOP senators are from states Trump carried. Clinton the Democrat was never in danger of removal in a Senate in which Republicans held 55 seats. All Senate Democrats voted against removing him.

In the case of Clinton, however, the certainty of the outcome in the Senate could not derail the impeachment process in a GOP-controlled House that had whipped itself up into a moral fury. With a Republican president and House majority, the unlikelihood of getting that two-thirds majority in the Senate surely would give pause. The circumstances favor inaction.

I have tried to show why the impeachment and removal from office of Donald Trump is better described as a Democratic political fantasy than a plausible sequence of events. But, you say, what if Democrats win the House next year? I’m afraid I have reached my limit for speculating about unknowns. If Democrats win, we can talk about it on November 7, 2018.