On Stochastic Terrorism

Moses Ma
10 min readNov 1, 2022

--

Stochastic terrorism is defined as the public demonization of a person or group resulting in the incitement of a violent act, which is statistically probable but whose specifics cannot be conclusively predicted. A perfect example of stochastic terrorism — that attains the level of an attack against our democracy — is the failed assassination of Speaker Nancy Pelosi, who is second in line of succession to the president. At issue is how to analyze prior available information to be alerted to the potential for such an act and how to evaluate the kinds of information that give the strongest indicators of this potential.

This is a proposal for a probabilistic model for understanding and addressing stochastic terrorism, which could be considered a “decentralized” crime.

It is useful to start by noting Justice Oliver Wendell Holmes, Jr.’s opinion in the landmark First Amendment case, Schenck v. the United States, which concluded that while most speech is protected, some speech isn’t. Examples of non-protected speech include intentionally telling lies and false statements and inciting lawlessness. In other words, there is no inherent Constitutional protection for telling lies, and this is part of the argument that it should be illegal to yell “Fire” in a crowded theater… when there is no fire. The case was later partially overturned by Brandenburg v. Ohio in 1969, which limited the scope of banned speech to that which would be directed to and likely to incite imminent lawless action, for example, a riot like what happened on January 6th.

Now, let us consider the following thought experiment — the possibility that stochastic terrorism does in fact turn out to be a crime. Currently, perpetrators are hiding behind laws that were written before the invention of the Internet. I believe that this situation is related to something like file sharing. Think about it, BitTorrent enables people to share files, and just because there isn’t anybody “in charge” — like it was with Napster — and just because the technology for sharing is novel, doesn’t mean that putting this technology to use improperly is not a crime. The end result in this instance was that someone obtained a copy of something copyrighted… without paying for it. It is the end result that determines the legality or illegality of an act.

More broadly, it could be that a new mode of murder is now being invented, just like file sharing is a new mode for the theft of intellectual property. Both are the result of the societally transformative nature of the Internet. And just like file sharing, you can’t just say, “hey, there were lots of people doing it” to avoid facing consequences for your act of stealing something. If you knew that some action you were about to take held a significant probability of killing someone, then taking that action is just like yelling fire in a crowded theater. The only difference is that this theater isn’t just a few hundred people, it’s tens of millions. The actual act of stochastic terrorism is functionally equivalent to hiring a hitman or swapping out someone’s life saving medication or winding up an angry mob, giving them some lynching rope, and directing them at your desired target. This qualifies as malice aforethought.

For the law to catch up with this emerging technology for inciting and directing violence, society will need to decide that simply “decentralizing” a crime doesn’t change its essential illegality. To do so, the DOJ could theoretically charge a perpetrator of stochastic terrorism with the “indirect murder” of a victim, just like if the perpetrator had paid a hit man. What’s more, it doesn’t matter if their act of stochastic terrorism does not wreak the intended damage — it is just like if someone paid a hit, but he shoots and misses, it is the act of paying the hitman that breaks the law.

However, in this case, a more apt analogy is that the perpetrator organized a token offering to raise money from a number of people to pay for a hit man, or used a DAO to operationalize the assassination. The bottom line is that this perpetrator wanted harm to befall a targeted individual and knew that his actions would likely lead to that happening.

Now, for a prosecutor in our thought experiment to succeed, the jury must find the facts sufficient to convince them beyond a reasonable doubt that the perpetrator did something meaningful to effect the murder of the victim. To do so, the jury needs to understand the motive, means, and opportunity, in order to convict the perpetrator.

Let’s start with motive. In this case, we need look no further than the desire of certain politicians to fundraise off hate. When a politician finds that they only need to stoke violence in order to better engage their supporters, they usually find that it’s as easy as shaking a money tree. Hate and polarization are very profitable business models. What’s more, hate works to win elections. In the political satire cum sci-fi superhero TV show called “The Boys”, there’s a Nazi character named Stormfront, who says, “You can’t win the whole country anymore. No one can. So why are you even trying? You don’t need 50 million people to love you. You need 5 million people fucking pissed. Emotion sells, anger sells.”

That pretty much sums up why stochastic terrorists do what they do.

The problem is that the vicious cycle of such fundraising tends to amplify the hate over time. The politician is forced to “ramp up” the hate, in order to extract more money. This is why it has been argued that certain conservative politicians have been overly aggressive in tapping into the nuclear fuel of hate to drive their campaigns, in essence pulling out all the control rods and risking a core meltdown.

The question is… when does such behavior become irresponsible? What line has to be crossed for such behavior to be considered illegal?

I argue that it’s when people start getting killed or hit in the head with hammers.

As for means and opportunity, the FBI and the CDC need to get together and run a “public safety” threat assessment — by developing models based on the likelihood of violence from a specific politician’s followers. The FBI could collect data based on online chatter, active investigations, sentiment analysis… and the CDC could develop models and Monte Carlo simulations to see how much violence would be caused by various levels of rhetoric. Just as the CDC can develop models for spread and death rate due to some novel infectious disease, models need to be developed to chart the spread and death rate of stochastic terrorism, and over time, these models must be improved to increase their accuracy. It’s not that different from estimating vaccine hesitancy and using those estimates to generate the likely number of fatalities.

Let’s consider a couple of initial conditions for our gedankenexperiment. First, if a politician’s base is comprised of small number of vegan pacifists, and the politician is responsible and dissuades violent behavior, then there is probably not a huge risk of generating stochastic violence. However, if a politician’s base consists of militias, biker gangs, and domestic terrorists… and the politician is a demagogue who consistently encourages violence, then the model will likely predict a significant risk of violence. If these models were openly published, making it possible to spot malice aforethought before it becomes a fait accompli, they could be used to warn politicians that they need to re-insert the control rods. The cognizance of open threat assessments and the implementation of serious and significant consequences for such reckless behavior, could help to reduce the risk of another societal black swan event like Jan6 or the Pelosi assassination attempt.

There’s one more point. The theoretical prosecution would likely need to use a variation of the RICO Act, which could be modified to combat the novel decentralization of organized crime. In the case of Paul Pelosi, any of the likely defendants could argue that “a whole lot of people” were vilifying Nancy Pelosi, so why should any single perpetrator be responsible? Using a 21st-century RICO+ Act, all of the perpetrators could be charged collectively. This is just like file sharing, right? You can’t let people hide in a crowd to avoid consequences.

Additionally, it may be possible to pursue civil lawsuits, rather than a criminal prosecution, to thwart stochastic terrorism. For example, if a victim were severely harassed by the perpetration of stochastic terrorism by certain political figures, it should be possible for those victims to sue those perpetrators. Consider victims of stochastic terrorism like Wandrea “Shaye” Moss and her mother Ruby Freeman.

If you don’t recall Wandrea and Ruby, a certain political campaign falsely claimed that video footage showed these women engaging in “surreptitious illegal activity.” The candidate himself singled out Freeman, by name, 18 times in a now-famous call in which he pressed Georgia officials to alter the state’s results. He called the 62-year-old temp worker a “professional vote scammer,” a “hustler” and a “known political operative” who “stuffed the ballot boxes.” This demonization of random rank and file election workers to rile up the voter base resulted in upending these workers’ lives, and they were forced into hiding by the many death threats that were unleashed against them.

Such a lawsuit would essentially be a decentralized version of the Alex Jones litigation, but instead of a single defendant, the litigants would seek punitive damages against an entire network of perpetrators. The parties would need to assess the same metrics as those described above, to determine culpability and damages.

But let’s get back to the possibility of criminal prosecution for stochastic terrorism…

Finally, we need a better name for this crime! In the American legal system, murder is divided into three distinct categories: first-, second-, and third-degree murder. First degree murder requires malice aforethought, premeditation, and criminal intent. For a crime to be classed as murder in the second degree, it must not be premeditated, but still demonstrates a clear intent to harm — i.e., a crime of passion. For a crime to be classed as murder in the third degree, criminal law states that there must be some intent to cause harm to the victim, but not the intent to kill — i.e., accidental homicide.

Perhaps it is time to create a new classification “murder in the fourth degree” based on stochastic terrorism. This would probably require two key factors: (1) cognizance of a significant likelihood of violence, and (2) premeditation that demonstrates intent to cause harm. This should still be felony murder — as it is probably equivalent to depraved indifference murder which is classed as a third-degree crime. (I just learned this term in a recent episode of Law & Order!)

What we are facing in today’s brave new world is that perpetrators can now decentralize the method of murder. What we need to do, as a society, is to start enforcing consequences for people getting killed due to stochastic terrorism. So if a perpetrator held malicious intent, primed his cult followers to be more violent, and there was a scientific process for gauging a significant likelihood of violence … then that perpetrator absolutely needs to face consequences for his 21st century crime. What’s more, stochastic terrorism is essentially a hate crime, and deserves being treated like any other hate crime.

Finally, there is some light at the end of the technological tunnel. The U.S. Army Analytics Group (AAG) recently began looking at ways to address this problem. In 2021, the AAG engaged a quantum computing AI company, Entanglement, Inc., and Groq, Inc, a U.S. semiconductor company, to try out proprietary technologies to detect “anomalies.” The AAG efforts had previously been able to detect 120,000 inferences per second, but this fusion of quantum simulation and AI achieved an anomaly detection rate of 72 million inferences per second, while correlating data simultaneously to arrive at a more accurate picture of where the threat lies. Recently, Entanglement has been to achieve 100 million inferences per second across the key workload — an approach that moves cyber threat detection to the threshold of a “total observability picture.”

This is only one example of the kinds of technological breakthroughs that are coming to address the problem of stochastic terrorism. Humanity will get better and better at this, and eventually bringing to life sci-fi concepts like “pre-crime.” Murder is defined as the act of intentionally causing the death of another human being. Acknowledging the existence of stochastic terrorism and progressively improving our ability to identify it through advances like those of the AAG, provides us with a way to detect such intentions ahead of time, to prevent violence from happening! The key here is that it isn’t enough say “Ralphie, you’ll shoot your eye out” — an admonition from the film “A Christmas Story” to nine year old Ralphie who wanted an air rifle — in the mistaken belief that urging caution by itself will avoid serious consequences. We need to collect data and use scientific principles, because it is impossible to manage what isn’t measured.

The reality is that unless we start treating stochastic terrorism as a public health threat, and ensure that consequences are imposed, it will face no restraints to keep it from growing exponentially until America tears itself apart. Let’s start being smart about this. An honest assessment demands we acknowledge that consequences should have been enforced on January 7th for the MAGA insurrection, but the failure to do so has emboldened the perpetrators of stochastic terrorism.

Even today, after the heinous crime against the Pelosi family, many politicians on the far right are leaning into hate and conspiracy theory, rather than doing the right thing by publicly denouncing these atrocities. Unfortunately, these are the people who are increasingly taking over the reins and setting agenda within the GOP. Their more reasonable partners in legislation either wrap themselves in a shroud of silence or offer “thoughts and prayers”… but don’t do anything to stop stochastic terrorism. Few are willing to do the right thing and vow to change the behavior of the Republican party.

Since Congress was unable to act, and the Department of Justice is slow to move… it is up to all Americans rise to the occasion and defend democracy by exercising the right to vote. The midterm elections are historically a time when thoughtful voters become lax and extreme elements post gains to entrench their position. It is up to us to show we can do better and bring our country back on the right path.

--

--

Moses Ma
Moses Ma

Written by Moses Ma

Moses Ma is the managing partner of FutureLab Ventures, a venture studio, think thank, and innovation consultancy that works with global companies.