How Diplomats See the World

A few years into my Foreign Service career, a colleague leaked a falsehood to the press. I couldn’t figure out why they’d done it.

A more experienced colleague explained the situation to me: The leaker was trying to break into partisan Washington. Their false claim supported a narrative favored by certain political leaders — and saying what powerful people want to hear opens doors. That colleague was competing to enter offices on the Hill, a world with entirely different rules.

I’d never operated in that world. I was deep inside what some call the “deep state,” the bureaucracy composed of long-tenured career civil servants who outlast individual administrations. Since leaving the government, I’ve tried to explain that culture. The clearest framing I’ve found is this: At its best, the national security apparatus is a distributed fact-checking machine.

Like journalists and advocacy groups, diplomats deal in information. But each one has a different relationship to what that information needs to do.

Journalism is often optimized for public interest. Selection pressure is toward the dramatic over the incremental, and the negative over the positive. Advocacy groups are optimized for creating change. The information advocates and journalists provide is often excellent. They often have ground-level detail that’s hard to get anywhere else. However, it’s rarely the full picture.

Diplomats optimize for small, measured, and accurate updates. The diplomat’s goal is to build a correct, if boring, model of the world, continuously update it, and prepare decision-makers to hold to that model with the right level of confidence. Here’s how diplomats see the world.

When I was posted to Egypt as an economic officer, I didn’t typically work on  crises or breakthroughs. My goal was to build a granular, quarter-by-quarter picture of the Egyptian economy.

One of my portfolio items was tracking poverty. Egypt is primarily a cash economy, which makes estimates difficult. Government poverty statistics are the best available data, but they’re released infrequently, sometimes only every two years. Governments, especially non-democracies, also have incentives to make conditions look better than they are.

To get around this, I tried to find proxies, such as the sales of commodities like electric kettles or soccer balls. Most families only need one, and it’s something you buy when you have a little money above subsistence; therefore, if imports are up, conditions are probably improving.

I spent much of my time finding small bits of useful data, sometimes tracking the economy or interviewing professionals operating in specific sectors. My assignment was usually open-ended. “Go talk to investment bankers and see what their assessment is.” “What are the charities helping poor families seeing?” “What is the head of that multilateral organization expecting and why did they decide that?” Sometimes, I wouldn’t find out much more than the organizations’ official stance. Sometimes, I’d find out a lot.

I’d form as many contacts as possible via conferences, mixers, introductions, and business dinners. This way, I could ask for clarity quickly when news broke. For my sources, I’d aim to have a mix of practitioners, local economists or analysts, non-profit and advocacy group leaders, government employees or officials, and businesspeople. Investment bankers, who often tracked the economy with a similar level of minutiae, often had the deepest insights, perhaps because there was money on the line behind their decisions.

All of these sources had high context and had no incentive to go on the record. Practitioners don’t want their assessments publicly attached to their names because this could bring unwanted media attention or reveal data meant to be private. Similarly, government officials usually prefer to stay private so that people don’t inundate them with cold outreach. This means that people with the most accurate information are also the least likely to share it publicly, but are sometimes willing to share it for a report that will stay very private. The people sharing information for articles or reports with wide public reach are sometimes interested in something other than accuracy, such as building reputation, advancing a policy agenda, or exposing wrongdoing.

The output of my research took the form of diplomatic cables: two-to-five-page reports summarizing what people on the ground were saying about a given subject, written on the assumption that most readers would only get through the first paragraph summarizing the piece and the last paragraph, the diplomat’s comments. Seen only by government agencies, diplomatic cables are similar to academic journal articles, except they are “small” updates with not-particularly-exciting headlines.

A typical finding in a diplomatic cable might be: Poverty in Egypt appears to be modestly improving at a rate roughly consistent with regional peers. Or: A mid-level finance official signaled openness to IMF program conditions, consistent with a moderate shift in the ministry’s position.

“Poverty in Egypt appears marginally better than last quarter” would not be news outside of some very niche journals. The goal is not a story, but a well-calibrated model of the world.

The closest civilian equivalent I’ve found to operating in this diplomatic mode is turning to practitioner communities. Today, when I want a reality check and I don’t personally know someone in the sector, I’ll find the relevant Reddit subthread to learn truckers’ views on freight demand or watch cyber practitioners debate the ramifications of a new attack. Reddit attracts people actually in the field and its anonymity means they can say what they think, even if it conflicts with media or official reports. Private Discords and Signal groups work the same way, telling the truth because their audiences are not general ones.

The accurate version of a story is usually the complicated one. The rules of the job force diplomats to capture complexity. A diplomatic cable has to get sign-off from different embassy offices — including the political or economic sections who see the same host-country issues from a different angle — as well as the ambassador, who might see people in the sector you wrote about at a reception later that week. Back in DC, many more offices will read it: labor and population desks, the economic growth teams, human rights offices. If your account is one-sided, you look out of touch. You must write the version that holds up from every angle.

Journalists and advocacy groups don’t always aim to write that way. I noticed this during coverage of one of my favorite companies, Samasource, previously known as Sama. The late social entrepreneur Leila Janah built Samasource to fight poverty by hiring people in low-income countries to do data work like annotation, validation, and model evaluation. Samasource inverted the usual hiring apparatus. Rather than screening out people because they were poor, unemployed, or lacked formal work experience, Samasource specifically sought to recruit from those communities. Janah was inspired by call centers in India, which mostly hired people who were already middle class, because they already had the English and computer skills. She wanted the workers those jobs skipped: people earning under $2 a day in the informal economy, picking up odd jobs and sometimes dangerous ones.

Samasource workers in Kenya made around $9 a day, rates set by the local labor market and more than four times Kenya’s minimum wage for an unskilled worker in 2018, though still below what a Kenyan nurse or attorney makes. By American standards $9 is almost nothing, and some coverage implied that these wages, coupled with allegedly poor working conditions, exploited workers.

But this skips over local conditions: If you pay data annotators well above what nurses and teachers earn, you pull people out of hospitals and classrooms to do data entry. Janah was trying to create sustainable economic opportunity without distorting the labor market she was working to improve. She didn’t want Samasource’s arrival to hollow out higher-paid employment.

A journalist might write about Samasource mostly in light of the trope of tech companies behaving badly, rather than in a way that adds context, for example that of the local medical field which might struggle to retain staff. An advocate might use it to push a campaign, perhaps for more aid or tighter rules on multinationals, a warning about Big Tech, or a push for donations. A diplomat would flag an unusual model, note the several offices and agencies that would want to know about it, and not highlight any particular verdict.

The culture that produces these habits of mind is built deliberately through hiring and internal structuring. The Foreign Service was explicitly designed to displace the 19th-century spoils system, in which jobs went to loyal partisans.

The Foreign Service selects people through a process where neither connections nor subject matter expertise are necessary or helpful to advance. Entry is decided almost entirely through exams graded by separate teams at different stages. The written exam is offered globally, including at embassies abroad. The oral exam is an eight-hour ordeal that involves simulations of embassy work and an interview panel. You negotiate with other candidates over which of your fictional projects moves forward, then brief a fictional ambassador. Most people fail. Those who pass go on a ranked list ordered by score. The person at the top of the list is the next person to get hired if a diplomat needs to be replaced. A candidate’s position on the list expires after 18 months.

The process doesn’t reward growing up on the coasts, majoring in international affairs, or knowing the right people. My colleagues came from all 50 states and included former teachers, software engineers, lawyers, and small-business owners. In this way, the Foreign Service, and national security circles at large, hire people who are less similar to one another than a think tank or a newspaper might.

Once you’re in, a rotation system prevents entrenchment. Diplomatic assignments change every two to three years. You might spend two years covering migration issues in Southeast Asia, then move to economic affairs in Latin America, then manage embassy procurement and operations in Namibia.

A diplomat abroad has a truth-seeking edge that is impossible to manufacture from home: distance. American journalists and advocates usually lean left or right, because American politics runs on two parties and most issues get sorted into one bin or the other. A diplomat posted overseas typically has much less of that reflex. The local politics are not yours. You also are often tasked with reading and understanding every newspaper in the country, and you always know that you may need to change regions soon.

The promotion system for U.S. diplomats requires that you serve in at least two different regions. Every officer has to be “worldwide available,” ready to serve what are known as the “needs of the service.” Early in your career, you go where the Department sends you. Your first two assignments are directed: If officers are needed in China, Gambia, or Sri Lanka, you might be the one to go. Officers who want to stay on a single issue or a single country usually leave the service.

Instead of becoming attached to one region or cause, you develop the skills required to quickly get up to speed on complex issues and efficiently determine which experts, practitioners, and data sources to trust.

Over time, this selects for people who keep their identities small, at least around foreign policy questions. You have to be willing to spend months learning the language of a country that wasn’t on your radar before your assignment. You may have to carry out policies you personally disagree with, and represent positions you find flawed. This compartmentalization around foreign policy does not mean that the people in the foreign service are apolitical in their personal lives.

The State Department does tend to lean left, primarily because the college-educated professional class leans left. Still, this ideological lean is not the same as political polarization outside of the Foreign Service. Most of what the State Department handles doesn’t have a partisan solution. Additionally, career officials have little professional incentive to become partisan operatives. Promotions are decided by committee and based on performance in key areas including communication, management, and technical expertise.

Political appointees, who might help you get more high-visibility postings, will leave after four years. And across the entire federal government of two million employees, only around 4,000 positions are available for political appointees. Those appointees hold powerful positions and set priorities, but the fact-checking apparatus runs without them. In 2020, there were roughly 14,000 Foreign Service personnel, including generalists and specialists, and about 13,000 civil service employees in the State Department. And though there have been cuts to the Foreign Service under the Trump administration, the scale of the career foreign and civil service is still significantly larger than the scale of the appointee roles.

Embassies continue their analysis, their citizen services’ work, and their relationship-building whether or not there’s an ambassador in post. Even without an Assistant Secretary of State for Consular Affairs, consular sections worldwide continue reporting on country conditions and immigration information. The work is mostly too technical and too unglamorous to attract political interference. Even within the current U.S. administration, the structural factors haven’t changed. Though there is a new job category of “Schedule Policy/Career” that can make employees easier to fire, it hasn’t appeared to be applied so far, and it’s possible the hiring structure and requirements of the Foreign Service mean it may be exempt.

Embassies and bureaus will hold off on making large decisions until their political appointee is in place. In a crisis, career servants will work with the next-highest-level of political appointee in place.

None of this is to say the foreign service apparatus is perfect nor immune to bias. Much of foreign policy analysis doesn’t often use quantitative data and is slow to adopt new technologies or data analysis tools. Information silos can still form when people up the chain become convinced of something and begin selectively amplifying the reporting that confirms it.

The most famous such example is the 2002 Iraq WMD assessment, where most of the national security and intelligence community concluded Saddam Hussein had an active nuclear program. He didn’t. The one exception was the State Department’s Bureau of Intelligence and Research, which got it right, in part because INR analysts were structurally more insulated from pressure to reach a particular conclusion. INR analysts are longtime experts who typically spend over 14 years on a single portfolio, and they write individual assessments rather than negotiating consensus documents. But the broader State position held that Iraq was armed.

The Senate’s postmortem of the intelligence community blamed systemic weaknesses, including a failure to convey intelligence uncertainties and groupthink that made ambiguous evidence always read one way. A similar blind spot hit the department in 1979. Diplomats had ample accurate reporting on the demonstrations and riots preceding the Iranian Revolution, but the Shah’s survival seemed so certain that leadership didn’t take the alternatives seriously. The current system doesn’t always work, but on the whole, it does reasonably well. When WikiLeaks published thousands of purported U.S. diplomatic cables in 2010, diplomats from other countries remarked, sometimes with surprise, on how carefully written and analytically sound the American cables were.

What makes the system work is less who diplomats are than how the job is structured. Much of what reaches the public about the world passes through the lenses of journalism or advocacy. Both are useful. Both also have their own hierarchies, selection pressures, and incentives that shape what gets reported and how. We account for this when the lens is obvious. We tend to forget it when it isn’t.