George Washington, the medical consensus now holds, died from aggressive bloodletting. Feeling sick from what may have been strep throat, he called his doctors at 3 a.m. on December 14, 1799. After losing nearly half of his blood in successive “treatments,” Washington died less than 24 hours later.

Bloodletting is no longer accepted medical practice, but we routinely perform comparable quackery on our political system unawares.¹ Voters seeking a remedy for the problem of “out-of-touch elites” damage the body politic when they weaken the political parties that form the basis of healthy political competition and democratic accountability. Citizens are right to remain alert and engaged, but home remedies can cause great and unintended harm. The shrill cry that American democracy is dying rests on a misdiagnosis. If anything, we have introduced too much democracy in the wrong places.

Consider the Democratic Party’s recent decision to downgrade the role of superdelegates in presidential primaries. “I hope that the grassroots who have felt dismissed and who have lost faith in the party . . . understand that they have had warriors on this commission who are completely in line with their values and that we fought and we won a lot to make this party inclusive.” Thus spoke Nomiki Konst, a Bernie Sanders appointee to the Democratic Unity Commission, after it had just agreed on new rules for selecting the party’s presidential candidate.² Beginning with the 2020 nominating contest, superdelegates—719 party bigwigs and other members of the Democratic establishment who represent about 15 percent of the total—will effectively be frozen out of the candidate selection process.

Superdelegates, who could tip the balance in divisive cases, had been the Democratic Party’s attempt to fix the party leadership’s loss of control over candidate selection following the McGovern-Fraser reforms of 1968.³ In now reverting to a bottom-up system of candidate selection, the Democratic Party has made strategic platform construction and electoral competition even more difficult. This is one instance of a larger pattern in which people call for more democracy to redress their alienation from the political process, yet institute reforms that end up compounding the problem.

Why Disciplined Parties Are Better than Weak Ones

The urge to dismantle strong party leadership often stems from legitimate gripes, but it can result in tragic error because bottom-up decision-making is not the same thing as democracy. Political parties are the core institution of democratic accountability because parties, not the individuals who support or comprise them, can offer competing visions of the public good. Voters lack the time and knowledge to investigate the costs and benefits of every policy, let alone to think about how their own interests must weigh against those of other citizens. Individual politicians may appeal to voters with brilliant ideas, but those ideas can only be implemented by a legislative majority. Lone politicians cannot tell voters in advance of elections what policies they will deliver unless they are members of a party campaigning on a platform on which its long-term reputation rests. Because parties gain or retain an electoral majority by offering widely appealing policies, party discipline—a party’s ability to command the votes of its legislative members—is key to democratic accountability. Only a disciplined party can credibly promise to deliver proposed policies if elected.

Parties are able to help citizens achieve what they could not on their own by considering the long-term consequences of each policy for every other goal or interest. Formulating and campaigning on a party platform is the antithesis of bottom-up democracy, such as primaries that allow voters to pick over the party platform, or referendums that give voters a say on one issue at a time.

Take primaries. More competition generally being preferable to less, a McKinsey consultant might say, primaries would appear to be a made-for-order cure for top-heavy, unresponsive party government. Primaries have, in fact, been peddled the world over as a cure-all, including most recently to imbue new life into India’s declining Congress Party. But primaries do no such thing. By creating bottom-up membership selection, primaries undercut the leadership’s ability to punish “cheating” on the party platform. The result is individually strong politicians who are collectively weaker than they otherwise could be. Or, to use a market analogy, giving locally based customers the ability to shape the product for sale undermines the firm’s ability to create the most desirable product for the market as a whole.

Consider another bottom-up measure, the referendum. Voters in California overwhelmingly passed Proposition 13 to set a ceiling on property taxes in 1978, with the unintended consequence that the Proposition undermined their ability to provide quality education for their children. Referendums, by slicing decisions around one issue at a time, undercut the unique ability of legislative parties to consider the relative costs and benefits of relevant issues in a bundle.

Brexit is a more recent instance of piecemeal politics gone wrong. The growing popularity in 2015 of Nigel Farage’s chauvinistic United Kingdom Independence Party (UKIP) made some members of Prime Minister David Cameron’s Tory Party nervous about sticking to their party’s official pro-Europe policy as parliamentary elections approached. Thanks to Britain’s plurality electoral rules, UKIP remained a minuscule parliamentary presence. But, particularly in some tight electoral districts, Tories joined the chest-thumping about protecting British jobs for the British and began humming the right wing’s anti-immigration tune.

Cameron did not have to hold a national referendum on Brexit. A parliamentary majority, and in fact majorities in both major parties, favored Britain’s continued EU membership. By the numbers, there was little doubt that Britain’s economy was better off in the European Union, given its enormous exports of financial services, and the lower labor costs from liberal EU immigration rules. But Cameron decided to put Britain’s EU membership to a popular vote in the summer of 2016, in response to widespread pressures to be “more democratic;” and no doubt also because he was confident that the Brexit gambit would fail. After all, in poll after poll majorities of British citizens had acknowledged the overwhelming benefits to the British economy from EU membership; and they had elected representatives who had made those judgments on their behalf in Parliament. Grabbing the immigration piece of the issue, the intense Brexiteers turned out in higher numbers than those who favored remaining in the European Union—robbing the British electorate of a parliamentary discussion and vote that would have elevated the broader, long-term interests of the British people.

These examples illustrate how strong party government is like a marriage, which enables its members to invest and plan for the whole family and for the long term. By contrast, in the colorful language of Argentinian commentator Eduardo Fidanza, weak parties are prone to patronage deals that are more like sex for pay. In a succession of electoral “reforms” across Latin America in the 1990s, voters gave themselves power to choose among individual legislators rather than, as formerly, to cast votes between party platforms. What seemed like a surge of democratic engagement in that period proved disastrous. Party discipline in Latin America collapsed into what Fidanza called “polygamous alliances” among opportunistic politicians, followed by presidential deal-making “orgies” of political corruption.

Fidanza’s orgy imagery calls to mind Lula’s Brazil. In such a weak party system, the President employs financial and regulatory resources to secure votes from free-agent legislators to pass bills. The result is a cornucopia of favors in exchange for legislative votes. Calibrated by the number of bills passed, the regime can look like a great success, as some scholars have argued. Measured instead by misspent resources and lower prospects for long-term prosperity, the voters get a raw deal.

Barriers to Party Discipline

Given the clear public benefits of electoral competition between disciplined parties with identifiable party platforms, why is there not a universal trend towards this model? If anything, entropy appears to work powerfully in the other direction.

One reason is that assigning credit and blame for policies is inherently hard. Reaching for the closest lever, voters seek to disempower leaders who appear to have erred. Error being hard to ascertain, incumbents do well when the economy hums along and lose votes when the economy stumbles.

Another reason to resist strong party discipline is that voters’ interests may diverge widely, particularly in geographically and demographically complex societies like ours. A majority in one electoral district may be at odds with those in other districts, which will disincline voters in that district to submit to party platforms that aim at the overall best interests of the country. One example is the Southern Democrats, who withdrew from the Democratic Party in the 1960s over differences over civil rights. India is another instructive case. As in the United States, India’s legislature is chosen in single-member districts, a system that tends to produce two parties since only large parties can form the legislative majority needed to shape policy.⁴ Despite the enormous advantages that accrue to majority parties with a credible chance at national rule, vast differences across Indian states along ethnic, linguistic, and religious lines have produced party fragmentation. This fragmentation may be diminishing as India’s regional differences diminish with the result that India will develop a system of two-party competition, but that has not happened yet.⁵

Without question, it is easier to be a responsible party of the kind we advocate when electoral districts are internally diverse and when districts are relatively similar to each other in their diversity. David Hume expressed this intuition when he argued that geographically large districts “enlarge and refine” public responsibility, an insight that James Madison echoed in The Federalist (No. 10).⁶ Ideally, politicians elected from more or less similar electoral districts would represent nationally representative interests. Single-member districts would push these politicians into one of two parties, both aimed at a widely appealing vision of the public good. Because the districts would resemble each other in major respects, representatives from each need not fear for their local reputation and re-election chances were they to delegate to party leaders the authority to implement policy for the whole party.

Ideal conditions are rare, and the exigencies of state formation, as in America in the 1770s-1780s, often require locking in long-term concessions to powerful local elites. Willing to combine against the British threat but keen to protect existing privileges, the colonies reserved for themselves substantial powers as future states. As a result, the Senate, with two seats from every state irrespective of population, continues to vastly over-represent states with small populations, forcing U.S. political parties to aim at a rurally weighted public interest rather than one representing the average American voter. The Italian Senate, which Prime Minister Renzi tried but failed to eliminate in 2017, produces similar distortions in that country.

Does Any Country Get Things Right?

As a historical matter, electoral competition between strong parties emerged gradually and by accident, first in Britain as the elimination of rotten boroughs and multi-member districts in the 19^th century shifted the center of gravity to large, competitive single-member districts.⁷ Electoral campaigning on the basis of policy favors for some and an open beer tap for all gave way to policy competition between two large and disciplined parties.⁸

As we argue in our new book Responsible Parties (Yale University Press, 2018), the policy-based competition that emerged in Victorian England continues to benefit British voters to this day for a set of interrelated reasons. Two-party competition forces parties to formulate and promote broadly appealing policies and creates a simpler set of choices for voters to absorb and from which to choose. The shift to single-member districts from multi-member districts had the effect of eliminating the incentives for politicians to stake out their own claims outside of the party platform, which can confuse voters and undermine the party’s ability to deliver on a promised course of action. In turn, shifting the basis of campaigning from personal loyalty to party platform made campaigns far cheaper, reducing the role of money in politics. Finally, the two-party system ensures that there is always one large party in opposition with powerful incentives to discover and advertise corruption and broken promises of the incumbent party.⁹ No one had the foresight to set up this system, but in retrospect it has remarkably appealing properties.

Yet Britain has not been immune from decentralizing reforms that have eroded the accountability of its parties. The parliamentary parties used to select their own leaders, but in recent years the memberships have played a larger role.  In 1998, the Tories finally acceded to pressure to democratize the leadership selection process, instituting a system whereby MPs vote among the contenders until only two remain, at which point the entire party membership chooses between them. This means that the party can end up with a leader who is not its preferred candidate, as happened in 2001 when the Thatcherite Iain Duncan-Smith won the at-large contest with 60 percent of the vote even though two-thirds of the parliamentary party preferred his centrist rival Kenneth Clarke. It was not a happy marriage. Duncan-Smith was forced out in a no-confidence vote two years later, at which point the Tories were so demoralized that Michael Howard, their fourth leader in six years, was elected unopposed.

Labour has had its own troubles with direct election of leaders. In 2015, following rule changes to allow party members rather than the members of the parliamentary party to select their leader, long-time backbencher Jeremy Corbyn, from the radical left wing of his party, was elected Leader. He was so far from the median Labour voter that most Labour MPs could not support his policies. Following streams of resignations from his shadow cabinet, in June 2016 the parliamentary party passed a motion of no confidence in his leadership by 172 to 40. Yet three months later the at-large membership, heavily dominated by activists, re-elected him with 62 percent of the vote. It is hard to imagine how Corbyn could govern in the event that Labour wins an election.

Proposals for Stronger U.S. Political Parties

Can political parties in the United States strengthen internal discipline? The U.S. system is often paired with Britain’s because they share single-member district electoral rules, but it lacks a constitutional feature that favors disciplined parties: parliamentary democracy.

In parliamentary systems, common to Britain and much of Western Europe, voters choose among parties, which in turn choose a Prime Minister on the basis of a parliamentary majority. Parliamentary parties have strong incentives to pull together as a team, standing behind a common electoral platform, because failure to do so can result in successful no-confidence votes that trigger new elections. Valuing their political lives, members of parliamentary parties work hard to solve disagreements internally before opposition parties have a chance to pry them apart in full view of a disappointed public. Parliamentary parties rise or fall together, whereas presidential systems, impervious to this version of sudden death, permit far greater levels of internal party dissension. While public disagreements may sound appealing for the same reasons given for primaries and referendums—that they return power to the people—their effects are the opposite. They undermine democratic accountability by stunting the ability of parties to create, compete over, and implement encompassing sets of public-minded policies. Parliamentary systems therefore have an advantage in nurturing responsible, disciplined parties that weigh competing claims to construct a platform that has the best chance of serving the most voters.

Voters in presidential systems labor under a structural disadvantage from this point of view. Still, it is possible to push political parties in presidential systems toward party cohesion and, by extension, greater political accountability. In 1950 a commission of the American Political Science Association chaired by E.E. Schattschneider recommended that American political parties establish strong party councils that would fuse leadership of the executive, legislative, and state and local parties, an idea that was never implemented but would have far exceeded the superdelegates innovation of the 1970s.¹⁰

The obstacles to party centralization are political, since the Constitution is silent on parties altogether—and, for that matter, silent on candidate selection. Starting in 1808, when James Monroe challenged James Madison for the Republican nomination for President, members of the congressional party, known as the congressional caucus, made the decision. This system prevailed until 1824, when it became a casualty of the fracturing Republican Party. Four years later both the incumbent John Quincy Adams and his challenger Andrew Jackson attacked the caucus system, so that no caucus was held in 1828. From 1831 onwards, congressional nominating caucuses were replaced by national presidential nominating conventions. The use of primaries grew rapidly from 17 in 1968 to 35 in 1980. Today every state has either a caucus or primary, so that the voters who participate in them have become the gatekeepers in presidential contests. Similar reforms among Republicans have had similar results, as was dramatically underscored in 2016 when the party establishment proved powerless to stop Donald Trump’s populist stampede to the presidency via a primary system in which superdelegates had already been stripped of their power unless no candidate won a majority before the Convention.

The emergence of stronger British parties in the 19^th century helps explain why self-strengthening measures have been few and fragile in the United States. American geographic diversity, amplified and exacerbated by gerrymandered districts, produces members of Congress who represent narrow and distinct slices of the American public. This electoral set-up cannot help but produce political parties that are internally divided and therefore disinclined to empower leaders to formulate and implement policies, however favorable to the average American citizen, which might be out of line with their own districts. One reform that would go a long way toward establishing conditions for party strengthening—what political scientists call endogenous institutional change—would be to draw electoral districts that approximate the political diversity of the nation. It is not easy to see how this could be done, but in our view it is far preferable to other ideas on offer including jungle primaries (which also weaken parties) or a move to proportional representation (which would fragment the party system in ways that Sweden and Germany struggle with now) or eliminating the Electoral College, which, by legitimating the President at the expense of the legislature, would further weaken parties.¹¹

Constructive change is hard, but political reformers should at least obey Hippocrates’s injunction not to do harm. Partisan gerrymandering and majority-minority districts promote the intraparty competition that breeds clientelism and support for sectoral interests. Healthy political competition is between parties, not within them. The best way to achieve that is to gerrymander for competitiveness among the parties.¹² This will be much more likely to occur if redistricting is taken out of the hands of state legislatures and given to independent commissions, as states like California have now begun to do. Another possibility is the British practice where an independent commission redraws constituencies following the Census and then Parliament votes to accept or reject, with no option to revise the commission’s plan.

The British also have their work cut out for them. Just as blue-state-red-state sorting makes it harder to obey Hume’s dictum to maintain large districts that mirror the country’s diversity as much as possible in the United States, regional variation in Britain undermines it there too. Welsh, Scottish, and Northern Irish voters are decreasingly like one another or like their English counterparts, a phenomenon exacerbated by the fact that London’s prosperity has not been matched elsewhere in the United Kingdom. The British would be best served by reducing the number of their constituencies (which are notably smaller than their counterparts in France, Germany, and the United States), and perhaps including a slice of London in each of them.

Real, Not Bogus, Accountability

One of the great paradoxes of electoral accountability is that less is more. Disciplined, centralized, vertically organized political parties solve a multitude of problems of information, judgment, and competitive strategy that voters as individuals or groups of activists dabble in at their peril. Nibbling at their power in hopes of increasing representativeness instead undermines it.

The McGovern-Fraser reforms, which opened primaries and caucuses to wider participation, solved one problem by creating another. Voters in primaries and caucuses tend to be activists who are well to the left of Democratic Party voters, creating the risk that they will pick someone who will not do well in the general election. This quickly became evident in 1972 when Senator McGovern won the nomination under the new rules, defeating the establishment candidates Ed Muskie and Hubert Humphrey, but then Richard Nixon beat him in a rout, winning every state except for Massachusetts and the District of Columbia. In 1980 Jimmy Carter was defeated for re-election by Ronald Reagan following a brutal primary challenge from Senator Edward Kennedy that many establishment Democrats blamed for fatally weakening Carter. Subsequent soul searching led to a Commission chaired by North Carolina Governor Jim Hunt to recommend the introduction of superdelegates to counteract grassroots influence. But after 2016 the grassroots successfully fought back.

Activists see superdelegates as an anti-democratic throwback to the smoke-filled rooms that had prevailed until 1968. The process had indeed been opaque, with backroom deals among state and congressional party officials determining the choice of candidates. But the turn to bottom-up nomination processes has further weakened already weak congressional parties, reducing their capacity to govern in ways that alienate the voters who demand decentralizing reforms. The supposed cure fails to address the underlying problem: Political parties cannot develop and run on coherent platforms that their members can get behind when they run, and implement if they win.

Before 1968 primaries were seen as ways for potential candidates to prove their viability to party leaders. Now they are seen as the font of a candidate’s legitimacy, compromised by the superdelegates. What is missing from this picture is any attention to the relations between the congressional parties and their presidential candidates. They are supposed to be on the same team, fighting for the same agenda, but the grassroots selection of presidential candidates drives a wedge between them because they have to worry about different electorates. Presidents must craft messages that can win among primary voters nationwide, while members of Congress must win in their individual constituencies, which might have a very different political makeup. Superdelegates were always a band-aid on this problem, because although members of the congressional party select most of them, they have never amounted to more than a fifth of the total number of Democratic delegates, and less than half of that for Republicans. In 2016 they were in any case powerless to stop Trump being selected as their party’s candidate by the less than five percent of the U.S. electorate that had voted in primaries and state caucuses.

Getting rid of primaries is likely impossible, but a rule that declared Convention delegates unbound by primary results in which turnout fell below, say, 75 percent of the party’s vote share in the previous general election would blunt the power of activists at party extremes. Still better would be reforms that allowed the party’s sitting representatives and senators to select the candidate if the 75 percent threshold was not met. Rule changes of this kind might not be such a difficult sell politically, because proposing them would highlight the exceedingly low turnout in most primary races.

The vices of presidential primaries are replicated at the congressional level, where activists turn out in disproportionate numbers and pull parties toward extremes, fueling gridlock in Congress and concomitant disaffection among the electorate. One might contemplate comparable reforms there, empowering House and Senate incumbent leaderships to ignore the primary result and select the candidate if the 75 percent turnout threshold was not met. A more robust option still would be to place the presumptive choice of candidates in the party’s congressional hands, to be overturned only when the turnout threshold was reached.

The great virtue of the early-19^th century congressional caucus was that it gave everyone the incentive to be on the same team. Members of Congress had to select the candidate who could best articulate a platform that they could run on and candidates for President had to win and retain the confidence of the congressional party—at least if they hoped to be successful in office and re-nominated. This came as close as to approximating a parliamentary system as was possible within the strictures of the American constitutional separation of powers system. It is a system that could be restored without a constitutional amendment, one that would strengthen the capacity of parties to deliver for their voters, attenuating the alienation that grows out of the gridlock that feeds voter alienation but which is worsened by decentralizing reforms.