With a president refusing to commit to a peaceful transition of power, a number of commentators have been sounding the alarm about a "rickety" U.S. electoral system seen as uniquely vulnerable to a postelection crisis. Legal scholars and a group called the Transition Integrity Project have been examining the ways in which the machinery could fail, and the nightmare scenarios gamed out are hair-raising. But while looking at points of possible legal failure is a useful exercise, it neglects an important question: Who has the power to concede?

Ultimately, all democratic transitions are based on one side being willing to concede power to another. Without a concession at some stage, power must be allocated by force: Either the military must decide, or there is civil war. There is growing concern that the U.S. may be arriving at a moment where a concession is no longer achievable — but if this is the case, this is ultimately a problem with the state of American politics, not its legal machinery.

To the extent that the electoral machinery matters, the American system is in many ways surprisingly robust. In ordinary presidential systems elsewhere, an election commission announces the outcome. Then, the political spotlight shifts immediately to the defeated candidate, who must make the crucial decision: Will they accept the result? It is a democracy's most defining and most perilous moment.

By comparison, America's electoral machinery, for all its oddities and flaws, offers greater systemic safeguards. What elsewhere is a single decision by an individual is in the U.S. spread out over up to two and a half months, set within a labyrinthine array of legal procedures — procedures that vest the power to concede in a vast number of actors across the constitutional system.

Any presidential election in the U.S. takes place in two stages: the vote-counting stage at the state level ahead of the Electoral College vote in mid-December, and a second stage in January when Congress counts the electoral votes.

The first stage involves innumerable local and state officials and courts. If even a small number of these actors break partisan ranks, they can effectively concede the election: a governor that certifies results supporting the opposing party, a judicial ruling that both sides agree to obey.

State legislatures in the U.S. have an untested reserve power that allows them to ignore their state's vote and appoint electors themselves. This has been portrayed as a grave danger to the system, providing yet another way for a presidential election to go off the rails. But it also serves to imbue even more actors within the system with the power to concede. A very small group of state legislators can break partisan ranks and yield an election on behalf of a presidential candidate.

The 2000 election was arguably the closest in history, decided by Florida by a margin of one one-hundredth of 1%, so close that we will never know who the true winner of the state was: Media recounts months later concluded that the results would change depending on which counting method was used. Even so, a concession occurred before the second stage.

That next stage, when Congress convenes in January to count the electoral votes, provides further opportunities for concession. If a state's electoral votes are disputed, the House and Senate meet separately to adjudicate the controversy. A potentially small number of representatives or senators can break rank, conceding the election by agreeing to resolve the dispute in favor of the other party. There is a potentially dangerous legal ambiguity here: If the House and Senate arrive at different decisions, the law governing the proceedings is unclear about how to reconcile them, with the potential for an unresolvable constitutional deadlock. But while legally dangerous, politically this ambiguity — along with the deadline of Inauguration Day — only serves to ratchet up the pressure to fold. The vice president, who presides over the vote count, may have a last opportunity to concede a disputed election by choosing to resolve this ambiguity in favor of the opposing party. The process culminates in an official pronouncement of the next duly elected president by the vice president, with the speaker of the House by his side.

Only once has a disputed electoral vote count reached the second stage. In the 1876 election, Congress invented a procedure to resolve the crisis, punting the disputes to an appointed commission.

The House and Senate could vote on the commission's recommendation — but if the two houses disagreed, the commission's decision would stand. The commission first recommended, on an 8-7 partisan vote, that Florida's disputed electoral votes be allocated to the Republican candidate, Rutherford B. Hayes. But while the Republican Senate endorsed the recommendation, the Democratic House rejected it. The pivotal moment came immediately afterward: With the speaker of the House at his side, the vice president announced that the commission's decision stood. The Democrats accepted the call; they allowed the count to continue.

This was the first crucial part of concession. A few weeks later, the vice president, with the speaker at his side, would oversee the completion of the count, and proclaim Hayes as president. A threat of chaos after Inauguration Day was the Democrats' only remaining weapon. Hayes provided certain reassurances, though the nature of those reassurances — the "Compromise of 1877" — is still disputed by historians today. Among them, however, appears to have been a promise regarding three Southern states where Black Republicans still held political power, protected by federal troops. Following 1877, the troops' protection was withdrawn and white Democrats took control, with lasting consequences.

America's electoral machinery provides many, many opportunities to find concession. But fundamentally, it still must be one of the parties that defers. Candidates must show a willingness to step back from the precipice. The nightmare scenarios necessarily presume that not only might a candidate reject the election result, but that an entire party will do so at every level within the system. If this comes true, the American experiment is probably over in any case.

If the American experiment is finished, the United States' unusually complicated electoral machinery merely guides how it would end: Elsewhere, an aspiring autocrat must corrupt or defy an electoral commission to kill a democracy. Determined Americans could kill theirs through obscurities of constitutional hardball.

Peaceful transitions of power require political will. In the end, people on one side must step back from the brink.

If history is any guide, they will.

Daniel Larsen is a historian and a lecturer in the Department of Politics and International Studies at the University of Cambridge. He wrote this article for the New York Times.