Recent high-profile terror attacks pose a new challenge for police and intelligence services. All seem to be the work of lone-wolf actors. Yet police and intelligence services, by the nature of their work, target groups. It’s possible to adjust that focus, but that would require Western societies to make an important trade-off.

On Monday, a 17-year-old Afghan asylum seeker gravely wounded four fellow passengers with an axe and a knife on a regional train near Wuerzburg in Germany before police shot him to death. The attack continues a series of terrorist acts by loners: the shooting in Orlando in June, the truck rampage in Nice last week, the cop shootings in Dallas and Baton Rouge. Islamic State claimed responsibility for Orlando, Nice and Wuerzburg, but that’s merely in keeping with its status as the umbrella brand for Islamist terror. The killers weren’t members of any terror group, and they weren’t acting on anyone’s orders.

Yet the Orlando and Nice attacks had so many casualties — a combined 133 people dead and dozens more wounded — that an organized group could hardly have been more effective. The November attacks in Paris, which were actually planned by an Islamic State affiliate, took 130 lives.

A full-scale military operation against Islamic State or active intelligence work against it wouldn’t have prevented the attacks by people who flew under the conventional radar, never traveling to the Syrian war zone or hanging out with known terror operatives. Besides, the lone-wolf phenomenon has little to do with any particular ideology, though the Islamic State banner has seen some heavy use lately (sometimes even literally: The Wuerzburg assailant had a hand-painted IS flag with him). As Jerrold Post wrote in “The Mind of the Terrorist,” “the cause is not the cause.”

“Lone wolf” is a term popularized in the 1990s by white supremacists Tom Metzger and Alex Curtis, who called on like-minded people to commit uncoordinated acts of terror. Some of the deadliest of the predatory species have been white racists: Norway’s deadliest killer Anders Breivik, Austrian letter bomber Franz Fuchs, “London Nail Bomber” David Copeland, Charleston church shooter Dylann Roof, Oklahoma City bomber Timothy McVeigh. It’s likely, though, that their ideologies ran only skin deep, just as radical Islam does for the recent killers of Middle Eastern and North African extraction.

“The conscious belief system,” psychiatrists J. Reid Meloy and Jessica Yakeley wrote in a 2014 psychological analysis of lone-wolf terrorists, “is upon closer examination often quite superficial: a cherry-picked cluster of prescriptive or proscriptive statements that provide a broad rationalization for the homicidal aggression.”

It’s enough for a system to be simple and binary, pitting good against evil, for it to serve as a justification for extreme violence. “Such defensive maneuvers are often part of a pathological narcissism in which the good object is within and the bad objects are all without,” Meloy and Yakeley wrote.

The narcissism is an obstacle to affiliating with an actual terrorist group, the psychiatrists pointed out — but it tends to strengthen the lone terrorists’ dependence on virtual communities, such as those formed on the internet. There, communication is free from the constant trauma of real-world interactions. That trauma is ever-present in lone-wolf terrorists’ stories — rejection by a father, a nonexistent or troubled sex life, professional or academic underachievement. On the internet, none of this matters.

The radicalization of Orlando shooter Omar Mateen and Nice truck driver Mohamed Lahouaiej Bouhlel could have been tracked by their internet activities. The Wuerzburg attacker, who had arrived in Germany as an unaccompanied minor and who lived with a foster family in a German town, probably left a similar trail.

The tracks these people left online are known as “weak signals” because they are hard for intelligence services to pick up. Yet there are techniques for doing that. A team from the Swedish Defense Research Agency documented one in a 2013 paper. It’s based on a number of behavioral markers that help determine whether a person is capable of radical violence.

For example, potential lone-wolf terrorists “leak” — they burn to tell outsiders what they want or even plan to do; they also display “fixation” — a preoccupation with a person or a cause — and “identification,” picking a “warrior” role model. The Web can be trawled for the linguistic attributes of these behaviors. For example, the Swedish researchers wrote, “leaked information of intent is likely to contain auxiliary verbs signaling intent (i.e., ‘… will …,’ ‘… am going to …,’ ‘…should …’) together with words expressing violent action, either overtly or, perhaps more likely, through euphemisms.”

A web crawler program instructed to look for such markers in the social network posts of people who visit radical sites (or, presumably, watch certain YouTube videos) could come up with a short list of people to watch. The Swedish researchers recommend always using a human analyst to check over the results of such trawling:

“Having such a human-in-the-loop makes it possible to tolerate a higher number of false positives than would be acceptable in a fully automated system. Since there is a trade-off between false positives and false negatives, the increase of false positives should decrease the number of false negatives (i.e., classifying weak signals from potential terrorists as non-interesting).”

If this sounds somewhat cynical — a false positive leading to a person being watched without his knowledge as a potential terrorist, or perhaps interrogated because of a computer program’s suspicions — it is. Unlike racial profiling, it is at least color-blind. There is simply no other way to look preemptively for potential lone terrorists, just as, in pre-Web days, there was no way to prevent “Unabomber” Ted Kaczynsky from mailing his first bomb.

It’s up to society in every specific country to decide whether intelligence services should have the power to use methods such as those suggested by the Swedish researchers — and to act on the information collected by such methods. The latter is especially important: The internet is full of hotheads of every description, and their freedom of speech is constitutionally protected in Western countries, though in most of Europe, hate speech is illegal.

My personal preference would be to live with the risk of lone-wolf attacks rather than let law enforcement agencies track citizens’ online activities so they can prevent them. Yet I suspect many who have lost loved ones in the attacks would gladly sanction such an application of modern technology. It’s a debate that can no longer be put off, given the growing number and effectiveness of lone-wolf attacks.

(c) 2016, Bloomberg View · Leonid Bershidsky