Why "lone actors" are so hard to find, and how we might get closer

Context: Despite initial concerns that the New Orleans attack suspect — Shamsud-Din Jabbar, who died in a shootout with police — may have had accomplices, the FBI quickly revised its statements after investigators began to believe Jabbar acted alone.

• "Lone actors or small cells of individuals typically radicalized to violence online and who primarily use easily accessible weapons have posed the greatest terrorism threat to our homeland," said FBI deputy assistant director Christopher Raia on Feb. 5.
• Raia said that's because lone actors are "difficult to identify, investigate and disrupt, especially when radicalization and communication with other like-minded individuals happens online."

What they're saying: Jabbar was "inspired from afar by ISIS," outgoing FBI director Christopher Wray said on "60 Minutes" earlier this week.

• "Guys like this," Wray said, "who radicalize not in years but in weeks, and whose method of attack is still very deadly but fairly crude … there are not a lot of dots out there to connect, and there's very little time in which to connect them."
• Put another way, with lone actors, "there are no informers, no infiltration, no ongoing surveillance … no territory so you can monitor who moves in and out, [and] you don't know what their capability is until they strike," says Ramón Spaaij, an Australia-based researcher with expertise in the sociology of terrorism.

Yes, but: Everyone leaves breadcrumbs, and there are still signs that someone is planning an attack, as long as you know what you're looking for and you have the resources to even be looking in the first place, Spaaij tells Axios New Orleans.

• So far, those identifiers — what Spaaij calls "pre-crime behaviors" — are most easily spotted after an attack. Hindsight is 20/20, as they say.
• Those behaviors, Spaaij says, include things like surveilling a potential target area, gathering weapons, researching other attacks, training with weapons and traveling suspiciously.

Zoom in: The New Orleans attacker did each of those things, FBI investigators have said.

• "What is difficult about online activity is [separating] extremist talk and radical messaging from those [that] might be a precursor to violent action," Spaaij says.
• And that's where machine learning might one day be helpful.

What we're watching: Defense and intelligence communities are investing heavily in predictive language models that can pick up on when people consistently use words indicating they might be planning an attack, Spaaij says.

• In his research, a sense of urgency and the "broadcasting of intent" seem to be clear indicators that could help warn law enforcement of planned violence, he says.
• In New Orleans, for example, Jabbar posted several videos on Facebook detailing his plans shortly before carrying them out, the FBI said early in its investigation. But it appears no one saw them quickly enough, let alone raised enough of an alarm, to stop what happened.
• "How do you create systems to detect these kinds of things … very efficiently and quickly?" Spaaij says. "There's a role there for machine learning and AI."

Reality check: "Human behavior is really complex and notoriously difficult to predict. The Holy Grail in anything to do with human societies is to predict behavior," Spaaij says.

• "It's a complete illusion to think we'd be able to prevent all terrorist attacks. That's the stuff of movies."

Go deeper

• The victims of the Bourbon Street attack
• How to help after the Bourbon Street attack
• How the New Year's Day attacker was able to get through Bourbon Street security barriers
• City, state leaders seek answers after Bourbon Street attack
• NOPD releases bodycam footage in New Year's Day attack