There’s a story famous among pollsters involving George Bishop, a University of Cincinnati political scientist. Almost four decades ago, Bishop, an expert in public opinion research who passed away earlier this year, devised an experiment to unleash on the American electorate. In a series of surveys, he asked a sample of people their opinion on whether the Public Affairs Act of 1975 should be repealed. Roughly a third of respondents offered a firm opinion one way or another. This was suspicious enough. A whole third of Americans knew enough about the Public Affairs Act to have an opinion on it? But there was another reason to doubt the results: There was no such thing as the Public Affairs Act of 1975. It was an invention, only for the purposes of the poll, but people felt compelled to weigh in anyway.

Bishop called responses like these “pseudo-opinions”—and today, almost 40 years later, our political culture is full of them. He was emphasizing the reality-distorting power of social pressure in polling. People are generally not well-informed, but they want to give the impression that they are, whether they’re responding to a pollster, chatting about politics at the office, or frantically scanning Wikipedia before posting a scorching hot take on Twitter or Facebook. “The simple fact is that on a lot of big policy issues, there really isn’t any informed public opinion,” Bishop would say years later. Pollsters today routinely ask people to give their opinions on issues—health care, immigration, the environment, the economy—and how each party or president is handling those broad topics. These questions are asked without any real explanation or context, and yet the results of these polls end up steering consequential decisions made in campaign offices, newsrooms, lobbying shops, and the White House.

Today’s loud and combative political atmosphere has added an ugly layer to the problem of pseudo-opinions. Our current gatekeepers of knowledge are, sadly, politicians, partisan media outlets, and social media performers who discovered politics three years ago. Plenty of research shows that conservatives are more likely than liberals to blindly trust their tribal chieftains, but there’s broad agreement that voters on all sides of the political spectrum generally don’t develop their opinions based on deep research, universities, or think tanks. Instead, when it comes to expressing a public thought about an issue, voters take their cues from whatever their preferred elites are saying—and they reason backward from there. If you like Bernie Sanders, Medicare for All is good. If you like Joe Biden, Medicare for All is bad. Never mind the details. Witness how Republicans have changed their opinions on Russia to align with whatever nice thing Donald Trump is saying about Vladimir Putin: In 2014, only 22% of Republicans called Russia an ally or friendly, according to Gallup. In 2018, that number had grown to 40%. With social media and the smartphone, you can form an “opinion” with a few swipes of the thumb and a glance at your favorite Twitter account.

The academic terms used to explain the Trump moment—pseudo-opinions, elites cues, motivated reasoning, negative partisanship—all surf a similar undercurrent: It’s much easier to pick a political team than it is to spend time digging into the issues. This weekend in Texas, though, a pair of Stanford professors are performing one of the largest in-person social science experiments in the history of U.S. politics, to test a quaint idea that might make the 2020 presidential campaign a bit more sane: What if voters actually knew what they were talking about?

Beginning Thursday, 500 carefully selected Americans from around the country will arrive at the Gaylord Texan hotel in Dallas for the country’s largest-ever experiment in Deliberative Polling. Culled from a pool of registered voters identified by random stratified sampling, they’re coming to participate in a project called “America in One Room,” with the optimistic goal of testing what would happen if Americans of every demographic and partisan shade sat down in person to ponder the country’s biggest issues. It sounds boring—people in hotel conference rooms reading briefing papers and sitting through guided discussions for a long weekend—but organizers say it’s never been more important for citizens of opposing views and backgrounds to talk deeply about American politics, insulated from the toxicity of everyday political combat.

“We are not assuming a civil conversation, but we are proposing that it can happen under good conditions,” said Larry Diamond, a senior fellow at Stanford’s Hoover Institution. Diamond and his Stanford colleague James Fishkin are the brains behind the Dallas experiment, a project led by Helena, a nonpartisan institute based in Los Angeles that develops projects aimed at tackling critical societal problems. “It can happen when you get out of a Fox News or MSNBC studio, when you get away from people shouting at each other from the sidelines, out of the heat of a political rally for Trump or Bernie Sanders,” Diamond said. “It can happen if you get out of the pitched battles and put people together in a room who have all been exposed to the same balanced briefing papers, who all have to introduce themselves and sit face-to-face over three days.” Their hope is to add a layer of nuance and deeper understanding to the traditional conventions of presidential campaigns: polls, debates, media appearances, and fund-raising appeals.

Helena, led by Henry Elkus and Sam Feinburg, a pair of brainy 20-something Yale dropouts, funds and runs its projects alongside a membership committee of boldface names that includes Stanley McChrystal, investor Ken Griffin, philanthropist Ray Chambers, and Nobel Peace Prize winner Beatrice Fihn. “The problem we saw is that the will of the people is not fully ascertained in American society,” Elkus told me. “The debate system is informative, but it doesn’t provide the ability for the American public to understand in depth what the candidates actually believe about critical issues facing our country. We have a character limit on Twitter, and we have a low attention span in the mass media. What Jim and Larry are doing with Deliberative Polling is a very small single tool that helps fight against that. It’s one step in the right direction of really understanding what the American public believes when they’re truly thinking about the issues and the candidates.”

The Deliberative Polling process pioneered by Fishkin and Diamond—once described as “intense democracy” by the New York Times—uses a conventional poll to sample a large audience and then isolate a smaller group of attendees reflecting the country’s demographics. Those delegates engage in three days of paper-reading and face-to-face guided discussions exploring certain topics—in this case, health care, immigration, the economy, the environment, and foreign policy. The weekend of debate and deep learning is concluded with a confidential questionnaire to measure whether opinions changed after several days of conversation under “good conditions.” Those results will be compared to a previously polled control group that was not exposed to the same Deliberative Polling process, allowing organizers to see how the opinions in Dallas changed compared to those who process politics like the rest of us, on television and through social media. The presumption heading into the weekend is that America’s political conversation sucks. Is there a better way for voters and candidates to make sense of 2020? “It’s a reform of the polling process, because democracy is supposed to make some connection between the will of the people and what’s actually done or who is selected,” said Fishkin. “But sometimes the polls are very superficial. They just represent an impression of sound bites, or they may not even represent an opinion at all.”

The experiment will provide a rich set of qualitative data about the 2020 campaign that, according to Diamond and Fishkin, will more accurately reflect what voters believe than any conventional poll or focus group. “The process will incentivize evidence-based thoughtful discussion that’s based upon real information, not made-up facts,” Fishkin told me. “One of the big problems driving us apart is not only the polarization, but the fact that everybody’s in their own filter bubble, partly because of the social media and the increased liberty they have to just find information they find congenial. If we can put people in one room, and if they can talk to each other and understand each other, then they may be able to deliberate and weigh arguments from competing sides, from different points of view, different political viewpoints, different demographics. Part of the experiment is: Will they be able to listen, as well as talk to each other in a civil manner, and not just offer insults?”

Given the Illuminati-like nature of Helena, America in One Room might seem like a post-Trump civility prayer cooked up by power elites who worship at the altar of high centrism. But Deliberative Polling has been used 108 times in 28 different countries since 1994 to foster healthier political dialogue. Fishkin and Diamond, both Helena members, have conducted Deliberative Polls in societies far more riven by racial and ethnic divides—places like Northern Ireland, Uganda, and Bulgaria. In Northern Ireland, in 2007, a representative group of Protestants and Catholics convened for a Deliberative Poll in the town of Omagh on the subject of school integration. After the poll ended, the percentage of attendees believing Protestants were “open to reason” increased from 36% to 52% and the percentage believing Catholics were “open to reason” jumped from 40% to 56%, according to a Financial Times story at the time. That same year, more than 200 Bulgarians met in Sofia to participate in a Deliberative Poll on how to improve the lives of the country’s impoverished Roma population. When the meetings finished and participants were surveyed again, those who thought “the Roma should live in separate Roma neighborhoods” dropped from 43% to 21%. Bulgarians who said “Roma schools should be closed and all the children should be transported by buses to their new school” went from 42% to 66%.

These stories unfolded in small European societies to impressive results. Minds were clearly changed when people engaged with one another in real life. But in the United States, in the Trump era, no serious organization has bothered to put Americans of different political persuasions in a single room for a large-scale attempt at civic discussion, away from hosts or moderators trying to stoke conflict. The political conversations most Americans witness today usually involve Facebook comments, paid partisans yelling at each other on cable news, or televised primary debates that are promoted like pro wrestling matches. The American political class emerged from 2016 without much pause, running straight into the next election without taking a breath to ask if there might be a different way of doing business. America in One Room will not command massive prime-time ratings or generate thousands of retweets, but it is worthy of our attention, if only to see if voters are as divided and dug in as they appear through the media looking glass. “We get that there will be people out there who say that this is not how the game works,” said Feinburg, Helena’s executive director. “But we think there are critical problems with the game. One of the tenets of the status quo, the game, is that there is not a substantive, actual discussion of the issues as there should be. Is the will of the people currently being reflected or acted upon? I think the answer is a resounding no.”

Peter Hamby is the host of Snapchat’s Good Luck America.