History is at least as much about the structures of power as it is about the personalities of “great men”—or terrible ones. Of course, a president’s idiosyncrasies matter, and the outsized and outrageous personality of the current president of the United States has riveted the public and press. But most condemnations of President Donald Trump are also, if only implicitly, accusations that his administration has broken with precedent and violated the norms of presidential government. We are not so sure. Trump’s manner of exercising power certainly has a precedent. In terms of the structure of politics—a term associated with the Cambridge historian Lewis Namier—Trump’s presidency is not so new, even if his personality is sui generis.
The most useful data to support this claim may come, paradoxically, from seemingly unreliable sources. Michael Wolff’s book Fire and Fury aroused a media frenzy with its hyperbolic accounts of Trump’s tantrums and the machinations of those around him. But Wolff is not the first journalist to write a breathless account of a new presidency. Bob Woodward did it for Bill Clinton, as well as for George W. Bush and Barack Obama, while Rowland Evans and Robert Novak did it for Richard Nixon. These accounts are each imperfect in different ways. They nevertheless offer valuable data about the pattern and frequency of interactions between officials. Even if some dialogue may be invented, the simple occurrence of meetings or conversations is less likely to be fabricated.
Network diagrams can represent almost any structure of relations as a set of nodes and ties (also called vertices and edges). We have used a straightforward approach to create network models for three modern American presidencies. Social ties are the sum of two individuals’ interpersonal interactions; the more they interact, the stronger their relationship.
To study those interactions, we use the labor spent by publishers in compiling an index of the pages in which an actor plays a significant role. In this way, we aim to avoid false positives—when an actor is merely mentioned in passing but has no relevance to an interaction—and false negatives—when an interaction takes place but the actor’s name does not appear, such as when they are referred to by pronouns or titles (e.g. “the president”). We represent each indexed actor or theme as a node. If two actors appear on the same page, we infer that there existed a tie between them, either in the mind of the author or in reality or both.
The sum of these co-occurrences is the strength of the tie linking two individuals. One limitation of this approach is that we ignore the reality that ties are neither symmetrical nor neutral: an official might care a great more deal about the president than the president cares about him. We should also keep in mind that the interactions we analyze are not the only ones to have occurred, just the ones that took place during events that, according to the chronicler, “mattered.”
As a simple example of this approach, we drew a network map of Trump’s first year in office by graphing the ties between the 25 individuals (represented here as squares) and the 25 themes (shown as circles) most frequently mentioned in the index of Wolff’s book.
The graph encapsulates Wolff’s narrative, and perhaps also his sources. Trump himself, his son-in-law Jared Kushner, his daughter Ivanka, and his chief strategist Steve Bannon are the most mentioned actors. By comparison, White House Chief of Staff Reince Priebus is barely visible. Russia is by far the dominant theme. The result is far from an objective documentary of White House intrigue. Rather, it is the formal reconstitution of Wolff’s narrative reconstruction of a group’s subjective interpretation of events and their internal dynamics.
We compared the structure of politics in the first years of the presidencies of Trump, Clinton, and Nixon, as recorded by their respective chroniclers. A few commonalities emerged: All three presidents were more connected than their officials, and each appointed a core group composed of campaign operatives and Washington insiders. But there were also a few key differences. Many commentators have drawn parallels between Trump and Nixon, perhaps reflecting liberal hopes that Trump may ultimately share Nixon’s fate. In terms of the network structure of their administrations, however, Clinton seems a better match.
Network science provides two concepts that capture different facets of power: centrality and equivalence. Degree centrality describes the number and strength of ties an actor has with adjacent nodes, or peers. High degree centrality can often increase an official’s quantity and quality of alternatives. A man with many close friends has many avenues for help in times of need, more so than a man with fewer close friends or one with many acquaintances.
Trump and Clinton had a weighted degree of centrality (that is, the sum of the strength of their ties) more than twice that of Nixon. Like Clinton, Trump does not shy away from meetings with officials, often losing his temper when dissatisfied with their performance. (Presidential tantrums are a leitmotif of both Wolff’s and Woodward’s books.)
Nixon, by contrast, had an almost reclusive streak, and “detested confrontation” of any kind. He was happiest alone in his office, scribbling notes on a yellow legal pad. His isolation, according to Evans and Novak, “was more pronounced than any President’s since Hoover—isolation from the press, from most of his aides, from his Cabinet. [He was] reclusive by nature.”
Or take closeness centrality, which measures how close an actor stands in relation to everyone else in a network, and is frequently used as a proxy for influence. An official participating in many committees and meetings has greater closeness centrality in an administration’s social network than one who is a member of just a single committee. Such “jacks of all trades” need to go through fewer intermediaries to reach any other official in the administration.
By reshuffling the nodes so that those with stronger, thicker ties are nearer to each other, we see how Jared and Ivanka Kushner and Bannon not only heavily influenced each other, but also had the highest relative influence in the entire administration after the president himself.
Clinton’s core advisers were his wife Hillary, Treasury Secretary Lloyd Bentsen, campaign adviser Paul Begala, counselor to the president David Gergen, and Vice President Al Gore.* As in the Trump White House, there was a turbulent contest for influence between those, like Begala, who had been close to Clinton on the campaign trail the previous year, and those, like the former Republican Gergen, who were brought in to give the administration political experience midway through the first year in office.
However, in Clinton’s case, those with Washington experience fought the struggle from a position of greater relative strength than Reince Priebus under Trump. Moreover, Hillary Clinton would appear to have exerted less influence than the combination of Jared and Ivanka Kushner.
A similar graphic captures the early importance of Robert Finch, Bryce Harlow, and John Mitchell in Nixon’s White House. Finch was (in the words of Evans and Novak) “Nixon’s protégé and alter ego,” whose decision to accept the cumbersome post of secretary of health, education and welfare instead of vice president or attorney general condemned him to gradual marginalization. Bryce Harlow, a veteran of the Eisenhower administration, was the key Washington insider.
“No one in the inner Nixon circle,” write Evans and Novak, “was known to anyone outside that circle. This meant that Republicans based in Washington … had only one entrée into the new administration. That was Harlow.” But his role was confined to dealing with Congress, and his “Hill Staff” often felt undercut by the “Downtown Staff” in the White House.
Nixon’s lawyer colleague John Mitchell, who had run his 1968 campaign, emerged as the most powerful of this trio as attorney general, often to be found lurking in the antechamber to the Oval Office, always ready to take Nixon’s late-night phone calls.
Although not yet apparent in the graph of Nixon’s first year, access to the president would soon become strictly controlled by the former campaign “advance men” H.R. “Bob” Haldeman and John Ehrlichman, the so-called “Germans,” who succeeded in building a “Berlin Wall” around the reclusive president.
It is a general feature of modern presidential politics that political outsiders enter the White House by dint of their role in the previous year’s election campaign. Because of their earlier proximity to the president, they gain a degree of influence over information and coordination out of proportion to their experience of government. This was true of all three of these administrations, but the concentration of power in the hands of a clique close to the president was significantly greater under Trump and Clinton than under Nixon.
More importantly, Trump and Clinton, despite the importance of their former campaign operatives, eventually rebalanced their administrations towards those more experienced in government. Just as David Gergen and Roger Altman grew more influential under Clinton in the course of 1993, so Trump yielded power to John Kelly, who took over as chief of staff in July 2017 and secured Bannon’s departure the following month.
In Nixon’s case, the balance of influence shifted the other way, with the ascendancy of Mitchell, Ehrlichman and Haldeman, the campaign operatives, and the marginalizing of Harlow (as well as Arthur Burns, who moved with relief to the Federal Reserve chairmanship with relief in 1970, after a frustrating time as counselor to the president). The consequences of this political structure were ultimately disastrous for Nixon.
Structurally, the Trump and Clinton administrations resembled each other—and differed from Nixon’s—in other ways, too. Both Trump and Clinton functioned as the key influence and information brokers in the White House. Additionally, Trump’s and Clinton’s network ties were very similar to those of their core advisors, potentially magnifying the power of the president, though increasing the risk of informational and policy confusion. Nixon’s role, in contrast, might have left him isolated and granted his advisors too much autonomy, but it also made him unique.
Betweenness centrality describes the extent to which an actor stands between others. Officials with high betweenness are the ones whose roles and decisions link their more isolated colleagues; they are able to act as influence brokers or information choke-points. Even in relatively small networks such as these, it is often difficult for the naked eye to discern the betweenness of a specific node, as this involves counting the number of pathways on which a node resides.
Nonetheless, the centrality scores for the three presidencies, standardized for comparison, reveal a stark difference. Unlike Trump and Clinton, the most between-central actors in their administrations, Nixon came in third behind Harlow and Finch, followed by two officials only about half as central, Ehrlichman and Mitchell. Clinton’s core advisors were all almost as between-central as he was: Hillary, Bentsen, Gore, and Begala.
For Trump, the most between-central advisors are Bannon, Kushner, Ivanka, and Hope Hicks. As Wolff describes the scene in the Oval Office: “Bannon invariably found some reason to study papers in the corner and then to have a last word; Priebus kept his eye on Bannon; Kushner kept constant tabs on the whereabouts of the others.” Such proximity to the president allowed these officials to increase their betweenness—that is, their ability to broker information and affect policy.
Clinton’s White House graph similarly depicts an inner court formed by Clinton and his wife, along with Bentsen, Begala, and Gore. It was characteristic of Clinton’s first year that Hillary refused to accept the traditional First Lady’s office in the East Wing of the White House, insisting on a presence in the West Wing.
Advisors in the Trump and Clinton administrations replicated their boss’s pattern of network ties, which partly explains why they have similarly high betweenness centrality. This similarity is also evident in the structural equivalence between these two presidents and their core advisors. (Structural equivalence measures the overlap between two actors’ sets of peers, indicating sometimes reinforcing, and sometimes conflicting, responsibilities and jurisdictions.) In the early Trump White house, for example, it appears that Bannon, Jared, and Ivanka Kushner had a similar set of ties to that of the president, likely reflecting that they often attended the same meetings and spoke with the same officials as Trump.
In marked contrast, Nixon seems more removed from his administration with thinner and sparser ties. This implies that a good number of important policy meetings and interactions did not involve the president himself, but rather his core advisors. As far as Nixon was concerned, the president’s principal concern should be to make foreign policy. “I’ve always thought this country could run itself domestically without a president,” he had told an interviewer in November 1967. “All you need is a competent Cabinet to run the country at home.” As Evans and Novak note, once installed in the White House, “Nixon did not care for structured meetings any more than unstructured meetings. He wanted written rather than oral communications, much preferring a thorough reading of reams of memoranda to face-to-face contact with his advisers.”
By turning over patronage decisions to Cabinet members, they argue, Nixon effectively killed his own chances of imposing Republican control over an increasingly interventionist bureaucracy. Instead of making a systematic attempt to roll back the Great Society programs, Nixon was persuaded to retain much of his predecessor’s legacy; the only constraint—enforced by Mitchell—was the political imperative to show gratitude to Southern states for having backed him sufficiently in the 1968 election.
The events of 2017 highlighted the difficulties that arise when power in the White House is concentrated in an inner circle that lacks not only experience of government but also a clear distribution of responsibility. Yet this is not a unique story. Most presidencies in the first year operate much like a royal court: the president is the focal point, and access to him is power. In his first 12 months, however, he is a powerful novice, and his inexperience magnifies the importance of those he appoints to key positions. He needs to work with the other branches of government—Congress, the Supreme Court, the Federal Reserve—which operate under different rules.
Meanwhile, the press exists in a symbiotic relationship to the government, needing the news it generates, communicating its actions to the voters who elected it, but also seeking to shape those actions by the stories it publishes. To meet these challenges, the president comes to see that he needs experienced insiders, not his campaign sidekicks.
The Trump and Clinton administrations had the following five traits in common during their first years:
- There was a painful transition in personnel from campaign operatives to Beltway professionals;
- Because of poor cooperation with Congress, both administrations failed to reform health care, but they both narrowly passed tax bills (hikes for Clinton, cuts for Trump);
- Each administration had a fixation on a particular financial market as a metric of success (the bond market for Clinton, the stock market for Trump);
- In each case, there was an excessive involvement of family members in policy-making
- Both administrations were constantly dissatisfied with the press coverage they received.
The problems of the Nixon administration in its first year were not wholly dissimilar—Nixon’s war on the press in many ways foreshadowed Trump’s. However, the differences outnumber the resemblances:
- The campaign operatives Ehrlichman, Haldeman, and Mitchell were able to prevail over the Washington insiders such as Harlow and Arthur Burns;
- Despite (unlike Clinton and Trump) facing antagonistic majorities in Congress, the Nixon administration secured the passage of a broad selection of legislation—a tribute, perhaps, to the skill of Harlow’s Hill staff. It suffered its worst defeats in its unsuccessful campaign to nominate a Southern judge to the Supreme Court;
- The Nixon administration—perhaps because Burns had been sidelined as a counselor—was insufficiently attentive to financial indicators, and was taken by surprise by the attack on the stagflationary effects of “Nixonomics” in 1970;
- Nixon’s family played no part in policy-making.
Moreover, unlike Trump and Clinton, Nixon prioritized foreign policy in his first year, and was able successfully to delegate key strategic initiatives to Henry Kissinger on the National Security Council, Mel Laird at the Department of Defense, and William Rogers and Joe Sisco at the State Department.
Namier’s primary goal in his prosopographical study, The Structure of Politics at the Accession of George III, was to show that ideology mattered much less in the Hanoverian House of Commons than previous scholars, accustomed to 19th- and 20th-century parties, had appreciated. “Men went there to ‘make a figure’, and no more dreamt of a seat in the House in order to benefit humanity than a child dreams of a birthday cake that others may eat it … The seat in the House was not their ultimate goal but a means to ulterior aims.”
Men and women say they wish to be president in order to benefit humanity, or at least its American branch. Once elected, however, they only seldom deliver the benefits that they have promised. Whatever their motives, presidents cannot rule alone. The White House serves as the presidential “court,” within which power is delegated and distributed. Journalists report excitably on the president’s conduct and speculate who is “in” and who is soon to be “out.” Too rarely is any attempt made to delineate the structure of political activity.
Our network analysis shows that Donald Trump has repeated the early error of the Clinton administration. He has been too much engaged with his inner circle, including family members, and allowed their responsibilities to overlap confusingly with those of other advisers and himself. In sum, in the Trump administration as in the Clinton administration, the president is more prominent than in the case of Nixon, but there are too many influential advisers working too closely with Trump, just as was the case in Clinton’s first year.
Richard Nixon kept his distance from most officials and legislators and established distinct domains for his advisors in foreign and domestic policy. The problems he encountered in his first year stemmed more from his own indecisiveness or miscalculation than from in-fighting or duplication of responsibilities. The dangers of his aloofness, and his readiness to delegate domestic politics to Mitchell, Ehrlichman, and Haldeman, did not become apparent until much later with the Watergate investigation.
If nothing changes in the Trump White House, the president will remain the center of all attention, and he will retain an inner circle of close advisors drawn from his past associates and family. It is possible that the increasing influence of more politically experienced operatives will reduce the disorienting gyrations in personnel and policy that attracted so much media attention last year. However, that is not the only direction in which this administration could evolve (especially if John Kelly does not survive the storm over his deputy Rob Porter’s private life).
The case of Richard Nixon shows that the structure of politics can be configured differently if the president’s mood or personality prompts him to withdraw from constant interaction with senior officials.
This is not to suggest that Nixon should be Trump’s model, rather than Clinton. It is to suggest that the structure of politics is as deserving of our attention as the personality of the president, even if the latter must substantially shape the former.
* This article originally stated that Paul Begala had served as counselor to the president. He didn't formally receive the title until 1997. We regret the error.