In December 2011, a man fleeing from a drug robbery shot and killed New York City police officer Peter Figoski. New York reacted with understandable outrage, particularly when newspapers there revealed that the officer’s killer, Lamont Pride, should have been in jail at the time.

The police in Greensboro, N.C. were already after Pride on charges that he had shot another man during an argument. But when Pride turned up in New York City — where he grew up — they opted to let him go instead of making the trip to pick him up. A little more than a month later, Pride dashed out the door of a dingy Brooklyn apartment and shot Figoski in the face. New York’s police commissioner at the time, Ray Kelly, said Pride “should not have been out on the streets.”

Putting a fugitive – especially one wanted for something as serious as a shooting – back on the street struck me as both somewhat remarkable, and something the government might track. But in months of looking for a reliable data source, I found that’s largely not the case. Jails usually don’t keep a record when they release someone with an outstanding warrant. Neither do police departments. Or sheriff’s offices. Every agency I called had anecdotes; none had reliable data.

Eventually, I turned to a vast, and ordinarily confidential, FBI database of wanted fugitives. Police agencies across the country use the database – part of the National Crime Information Center (NCIC) – as a sort of clearinghouse for information on fugitives. That database is one of the main ways police capture fugitives in the United States. Whenever a state trooper pulls you over for speeding, he can check your name against a list of more than a million people with outstanding felony warrants. The system can tell them within seconds whether a person is wanted. And it provides one other critical bit of information: How far the agency that wants the person is willing to travel to retrieve him.

These data aren’t a perfect measure. A lot of law enforcement agencies don’t list the majority of their fugitives in the FBI’s database; many list only those fugitives they actually intend to pursue into other states. And other agencies indicate in their entries that they will travel anywhere in the country to retrieve their suspects even though they actually won’t. Still, these were by far the most complete and reliable data sets for looking at police departments’ extradition practices, and it let us focus on fugitives police and prosecutors had affirmatively decided they would not pursue across a state border.

Getting the data from the FBI took two Freedom of Information Act requests. Eventually, the bureau agreed to release a very limited portion of its NCIC database, at no charge: For each fugitive, it would give us the charges they were facing, the date the warrant was entered, a code indicating how far the police would travel to pick him up, and another code identifying the department that had entered the warrant into the system. The FBI said (and the Justice Department agreed) that giving us the fugitives’ names or anything else that might identify them (including their last known states of residence) would be an unreasonable invasion of their privacy.

The data were a mess – which maybe shouldn’t be a surprise since the database that produced them was built in 1967. The FBI’s database uses the wrong postal code for Nebraska, and lists every federal immigration warrant as originating in Vermont. The first batch of records showed up with 200mb of data and 300mb of ASCII null characters. The second was in a much cleaner pair of Excel spreadsheets.

The FBI also wouldn’t tell us the names of the agencies that entered the warrants. Instead, it gave us nine-character ORI codes, then declined to tell us what those codes meant. We were apparently in good company, though – a researcher in another arm of the Justice Department told us they wouldn’t tell her, either. So my colleagues and I built our own list with data scraped from the Justice Department’s asset forfeiture section, the New York State Police and a handful of other law enforcement agencies. I also built a simple web application that let us manually review every agency in the database to make sure we had the correct name and location, and to weed out federal warrants (because the federal government doesn’t have to bother with extradition within the United States). In the end, we were able to put agency names with about 93 percent of the warrants in the FBI database.

That left us with a data set that told us where we’d find a story: Philadelphia.

Police opt not to go after fugitives in almost every city. But Philadelphia immediately stood out in the data. It’s a big city with a significant crime problem. Its courthouse is less than two miles from the state border, and its suburbs sprawl into three other states. Yet according to the FBI data, police and prosecutors were unwilling to retrieve 93% of their felony suspects from outside the state.

But we still didn’t know who the fugitives were, what they had done, or where they were. To figure that out, I obtained a similar extract of warrant data from the Pennsylvania State Police. The State Police provided the records at no charge in a fairly clean text file. They, too, declined to give us names, but an official provided us with records that included police departments’ incident numbers. (I followed similar steps in other states to track down their fugitives, too.)

I used those numbers to scrape the Pennsylvania court system’s online case lookup. Rather than reverse engineer the site’s JavaScript, I used a Perl module called WWW::Mechanize::Firefox, which automates a Firefox browser. Within a few days, our database contained court dockets – totaling about 100,000 pages of PDF files – for more than 20,000 fugitives. Those dockets gave us the fugitives’ names and birthdates. (A second scrape, for records outside of Philadelphia, pulled in so many records that I’m still blocked from using the state court website.)

Even those were a mess. Court clerks had more than 25 different ways of spelling “Philadelphia.” Sometimes they put it in the wrong state. Hundreds of records referred to towns that don’t appear to exist anywhere in the United States. The python script we used to clean up just the city and state names so we could plot them on a map ended up requiring 500 lines of code and a copy editor with an atlas.

The scrape left us with a huge pool of cases to look at. We narrowed it by using a few Perl and Python scripts to automatically query each name in a variety of public records databases and used the results to assign a rough score of how interesting the fugitive was likely to be based on whether his last known address was out of state, whether the charges were especially serious, and whether there was some indication that he had a criminal record elsewhere. Those checks narrowed the pool to 572 Philadelphia fugitives.

Armed with those names, we used records from police departments, courts and prisons from across the United States to conduct more detailed background checks on each of those fugitives (plus more than 200 others from different states). We also reviewed Philadelphia court files for most of those fugitives – though in some instances, the courts had either lost the file or refused to allow me to inspect them because they contained the names of sex crime victims. We used a custom web application to track the results, and to help tag the fugitives whose experiences would become the backbone of our stories.

Those records led us, for example, to Thomas Terlecky, who is on the run from charges that he sexually assaulted a 14-year-old girl in a Philadelphia hotel room. We found him with no difficulty in Miami, where he is a registered sex offender, having been convicted of having sex with two other underage girls in the years that followed. And we found that, although police in Miami had tried repeatedly to send him back to Philadelphia, authorities there repeatedly refused to travel to Florida to pick him up. Armed with that evidence, Terlecky – and most of the other fugitives whose cases we documented – talked on the record about their experience.

Brad Heath is an investigative reporter for USA Today. Contact him at bheath@usatoday.com.

 


Want more?

Behind the Story: Learn how Brad Heath tracked down fugitives and got them to talk on the record. Heath also discusses the challenges of interviewing crime victims and law enforcement officials.