Free speech has limits, as a famous Supreme Court example illustrates. “Falsely shouting fire in a theater” is not constitutionally protected speech, justice Oliver Wendell Holmes wrote in 1919.

Nearly eight decades later, the first criminal prosecution of threats on the Internet again tested the boundary of free speech. As an FBI agent, I was a player in that 1995 landmark case.

The defendant was twenty-year-old U-M student Abraham Jacob Alkhabaz, known as Jake Baker. He was described as quiet and nice and wrote stories with innocent titles like “Going for a Walk.”

But he harbored demons. The stories were lurid, graphic tales of kidnapping, raping, torturing, and killing young women–so-called snuff stories. Baker posted them at alt.sex.stories, a Usenet chat group. The Internet was in its infancy, and his case raised issues law enforcement and courts had never imagined.

Almost twenty-five years later, we still face the tricky, high-stakes questions: Where does freedom of speech end, and when do words become a crime? How do you predict when hateful or misogynistic speech will morph into violence?

To examine the issue, it’s worth looking back at United States v. Alkhabaz, a touchstone in the history of cyber law.

Baker’s writings were discovered by a Michigan alumnus who stumbled across one of Baker’s stories and knew from the IP address that he had some U-M affiliation.

The story used the name of a real Michigan student as a victim. (In court papers she was referred to as Jane Doe.) The student was not aware of the story or that she was about to be a player in a First Amendment controversy.

The alum contacted university officials, who notified the campus Department of Public Safety. Detectives talked to Baker and obtained a warrant to search the computer in his East Quad dorm room. The search revealed several more snuff stories by Baker. Two used Jane Doe’s name, and one had her address and phone number.

As an introduction to his stories, Baker wrote: “Torture is foreplay. Rape is romance. Snuff is climax.”

This is an excerpt from one:

Then Jerry and I tie her by her long brown hair to the ceiling fan, so that she is dangling midair. Her feet don’t touch the ground. She kicks trying to hit me, Jerry or the gorund [sic]. The sight of her wiggling in mid-air, hands rudely taped behind her back, turns me on. Jerry takes a big spiky hair-brush and starts beating her small breasts with it, coloring them with nice red marks. She screams and struggles harder.

At this point the story goes from R-rated to X-rated. It ends with Baker’s protagonist lighting Jane Doe on fire.

A search of Baker’s email account found numerous exchanges with an online acquaintance who apparently lived in Ontario. They discussed getting together to actually commit the crimes Baker had imagined. In one, Baker wrote:

I’ve been trying to think of secluded spots, but my area knowledge of Ann Arbor is limited to the campus. I don’t want any blood in my room, though I’ve come upon an excellent method to abduct a bitch–As I said before, my room is right across from the girl’s bathroom. Wiat [sic] until late at night, grab her when she goes to unlock the door. Knock her unconscious, and put her into one of those portable lockers (forgot the word for it), or even a duffle bag. Then hurry her out to the car and take her away …what do you think?

The acquaintance replied:

Hi Jake. I have been out tonight and I can tell you that I am thinking more and more about “doing” a girl. I can picture it so well … and I can think of no better use of their flesh. I HAVE to make a bitch suffer!

Baker’s response, in part:

I know how you feel. I’ve been masturbating like the devil recently. Just thinking about it anymore doesn’t do the trick … I need TO DO IT.

The U-M police contacted the Washtenaw County prosecutor, who concluded the posts didn’t violate any state statute. They then turned to the local FBI office.

After reading Baker’s stories and emails, I concluded that in context with the stories, the emails constituted a threat as defined by federal statute 18 USC 875(c). It forbids transmitting “in interstate or foreign commerce any communication containing any threat to kidnap any person or to injure any person.”

The statute was written long before the Internet, but it clearly was an instrument of interstate commerce.

I presented the case to the Detroit U.S. attorney’s office, which agreed with my conclusion. Our contention was that Baker had threatened not only Jane Doe but every woman in East Quad.

Baker was arrested and arraigned before a federal magistrate in Detroit. The case was assigned to U.S. District Court Judge Avern Cohn.

Judge Cohn wasn’t a fan of the government’s case. He made it clear that Baker’s stories were protected by the First Amendment’s free speech clause and couldn’t be part of the prosecution. So when Baker was indicted, all references to the stories were eliminated, leaving only the emails.

Cohn then dismissed the indictment, saying the emails were nothing more than a private conversation discussing fantasies. He criticized the government and its “overzealous agent,” referring to me.

The government appealed to the Sixth Circuit Court of Appeals. In a split decision, it ruled that the emails did not constitute a threat because they were “not conveyed to effect some change or achieve some goal through intimidation.”

I don’t know where Baker is now, and I have no reason to believe he ever tried to bring his horrific fantasies to life. But he might have if we hadn’t interceded. (He spent thirty days in custody before Cohn tossed the indictment.)

In an age of terrorism, both domestic and international, law enforcement is left with the conundrum of how to address Internet communications that could be preparation for criminal acts.

Most mass shootings have been preceded by Internet postings. The man charged with killing twenty-two people at an El Paso Walmart in August allegedly posted a white nationalist screed describing the attack as “a response to the Hispanic invasion of Texas.”

The murders occurred minutes later. But if no attack had occurred, would the post have constituted a criminal threat that could be prosecuted?

Misogyny–the professed hatred of women and/or violent acts against women–is a characteristic of shooters in more than half of U.S. mass shootings from 2009 to 2017.

The FBI now pays more attention to social media–including chat rooms and sites where users share white nationalist and misogynistic messages–to develop profiles of potential mass shooters. Such profiles, coupled with stricter background checks and a federal “red flag” law, can’t eliminate all mass shootings–but could stop some.

In June 2019, FBI information led to the arrest of Concord, California, resident Ross Farca. Farca, twenty-three, allegedly authored online posts threatening a mass shooting at a synagogue.

They read, in part: “I would probably get a body count of like 30 kikes and then like five police officers because I would also decide to fight to the death.”

No date, time, or specific synagogue was identified.

Based on the posts, local police obtained a search warrant for Farca’s residence. They found an illegally modified AR-15, thirty high-capacity magazines, and Nazi literature. Farca is charged with making criminal threats and possessing an illegal assault rifle.

Perhaps this was a mass shooting that didn’t occur because of better vigilance of the Internet–and a better understanding of what constitutes a threat.

The writer, an FBI agent for thirty-one years, retired as resident agent in charge of the Ann Arbor office in 2006. A version of this article previously appeared on www.DeadlineDetroit.com.