Donald Trump has repeatedly referred to the news media as the “enemy of the American people.” In May, Montana House rep Greg Gianforte body-slammed Guardian reporter Ben Jacobs for asking a question. In July, reporters were ordered to delete photos and videos they had taken of Trumpcare protestors being arrested in a Senate hallway.

It feels as if press freedom has declined in the six months since Trump was inaugurated, but a new site, U.S. Press Freedom Tracker, will show whether things actually are more dire than they used to be. The site, launched Wednesday, “aims to be the first to provide reliable, easy-to-access information on the number of press freedom violations in the United States — from journalists facing charges to reporters stopped at the U.S. border or asked to hand over their electronics.”

The site is run out of the Freedom of the Press Foundation and has 20 partners, including Committee to Protect Journalists, Reporters Without Borders, Knight First Amendment Institute at Columbia University, and others. (Greg Gianforte pledged $50,000 to Committee to Protect Journalists as part of a legal settlement — money that went to help fund the U.S. Press Freedom Tracker; going forward, CPJ the site will be funded through its partners and via fundraising.)

“We wanted to see what we could track comprehensively and quantitatively,” said Peter Sterne, managing editor of the site and a former media reporter at Politico. “Arrests, border stops, and equipment seizures are data points that we can start collecting now and keep collecting in the future, to look and see if things have gotten worse or — hopefully — better.”

At launch, the site catalogs press freedom incidents back to January 1, 2017. For that backlog, Sterne collected and independently verified incidents primarily from published reports; going forward, he expects that most tips will come from partner organizations and individual reporters.

The site categorizes incidents as “Arrest/Criminal Charge,” “Border Stop,” “Subpoena/Legal Order,” “Leak Case,” “Equipment Search, Seizure, or Damage,” “Physical Attack,” “Denial of Access,” “Chilling Statement,” or “Other.” Visitors can filter incidents by category or view all of them; they can also search incidents by location, outlet, reporter’s name, and a variety of other filters, or download all the data to do their own analysis. The filters are handy if, for instance, someone is trying to figure out the U.S. citizenship status of reporters being stopped at the border, Sterne said.

So far in 2017, U.S. Press Freedom Tracker says, there have been 19 arrests of journalists, 12 equipment searches and seizures, 11 physical attacks on journalists, and 4 border stops of journalists.

There’s also space on the site for more qualitative incidents — denial of access, chilling statements, things that “we can’t really count comprehensively because it’s so subjective,” Sterne said. “We’re trying to be representative and say there are some incidents that can give you a sense of what the tone has become. It’s not something we’re looking to count quantitatively and compare over time. But whenever there is an incident that, in my editorial judgment, says something about the state of press freedom in the U.S., it’s something we want to include on the site.” A blog also pulls in reporting, some of which is more qualitative, from partner organizations.

While U.S. Press Freedom Tracker will issue annual reports, its main goal is to be “a constantly updated site and resource and database that people can refer to at any time,” Sterne said. “I’m a journalist. My hope is that, whenever something happens, I’ll be able to get it up on the site relatively quickly.” That includes developments in cases the site is already tracking, like indictments or charges being dropped: “The hope is that journalists, attorneys, and members of the public can come look at what the current status of a case is.”

People want to know if U.S. press freedom is worse under the Trump administration than it was in previous years. But it’s too early to say, according to Sterne. Data before 2017 is not being included in the tracker because it’s difficult to do so comprehensively and accurately.

“My concern was that if we tried to collect data from 2016, 2015, 2014, it might seem artificially low and would appear to have gotten much worse in 2017 when that was actually only an artifact of the data collection,” Sterne said. “That’s why I’m not willing to draw any conclusions yet. In a year — especially in two years — we’ll be able to look back and say, okay, there were this many arrests in 2017. How many were there in 2018?”