The number that surfaces in the comments under every one of these stories is one in five. One in five kids solicited online. It has been quoted in school assemblies, church bulletins, and city council meetings for twenty-five years, and the researchers who produced it have spent most of that time trying to correct it.
An FBI investigation and arrest in southeast Michigan usually reaches you as a push alert, somewhere between the first cup of coffee and the school drop-off. A man, an age, a township you've driven through, a charge with the word minor in it. The alert is accurate. What most of us assemble in our heads over the next ninety seconds is not.
I started reading the underlying court documents for an unglamorous reason: a case broke two miles from my house, and I wanted to know what the affidavit said versus what the neighborhood had already decided it said. Those turned out to be different documents.
What most people do with the news
The screenshot goes to the group chat. Someone pulls the address and measures it against the nearest elementary school. Someone else posts the booking photo with a caption that assumes the trial has already happened. Within a day, a parent somewhere in the thread announces they've installed monitoring software on a thirteen-year-old's phone.
None of that is unreasonable, exactly. It's just aimed at the wrong threat model, and one part of it actively backfires.
The part that backfires is the vigilante sting. "Predator catcher" channels — adults posing as minors, filming confrontations in parking lots for an audience — have proliferated across Michigan and the Midwest. Prosecutors have declined or dismissed cases these groups generated, because the evidence chain is unauthenticated, the decoy has a financial stake in the confrontation, and the encounter itself can taint an investigation police were already running. A viral video is not a case file. Sometimes it's the reason there isn't one.
What the evidence actually shows
Start with that one-in-five figure. It comes from the first Youth Internet Safety Survey (Finkelhor, Mitchell, and Wolak, 2000), which found that 19 percent of young internet users reported an unwanted sexual solicitation in the previous year. "Solicitation" included crude comments from other teenagers. It did not mean an adult was grooming your child. By the second survey (2005) the figure had fallen to 13 percent; by the third (2010), to 9 percent. The number people still quote is the highest reading of the most inclusive definition, taken during the dial-up era.
The more useful paper is Wolak, Finkelhor, Mitchell, and Ybarra (2008) in American Psychologist, drawing on law enforcement interviews from the National Juvenile Online Victimization Study. Their central finding cuts against the stranger-in-the-van image almost completely: the typical internet-initiated sex crime involved an adult man and a 13-to-15-year-old, the adult rarely lied about being an adult, and the contact usually developed over weeks or months into something the teenager understood as a relationship. Deception about age was the exception, not the method. Seduction was the method.
Meanwhile the raw volume of reported material is enormous and hard to interpret. The National Center for Missing & Exploited Children logged roughly 36 million CyberTipline reports in 2023 and about 20 million in 2024 — a drop NCMEC attributed largely to how a single platform bundles its submissions, not to less abuse. Treat that curve as a measure of reporting infrastructure, not of underlying crime.
One more figure, offered with a caveat. The commonly cited estimates from that same study put arrests arising from undercover operations at several times the number arising from cases with an identified child victim. The ratio is directionally important; the specific numbers are two decades old, and no one has published a comparably rigorous update. That gap matters more than it gets acknowledged.
The order it actually happens
Affidavits read as narrative, but the investigation has a fixed sequence, and knowing it tells you what a given case does and doesn't establish.
First, contact — either a platform tip routed through NCMEC to the appropriate field office, or an agent maintaining a persona in a chat app or dating platform. Second, the conversation, preserved in full, because prosecutors need the defendant's own words establishing what he believed the other party's age to be. Third, identification: subscriber records from the platform, obtained by subpoena or warrant, then corroborated against phone numbers, payment accounts, and vehicle records. Fourth, the act that converts intent into a chargeable attempt — usually travel to a meeting point, sometimes the transfer of an image. Fifth, the arrest, then devices seized under warrant. Only after that does the public document appear.
And here's what that public document is. A criminal complaint is a probable-cause allegation sworn by an agent. It is not a finding. The government must still obtain a grand jury indictment, generally within thirty days. At the detention hearing, several of these charges carry a statutory presumption in favor of holding the defendant — which is why "held without bond" appears so often and means less than readers assume.
| Statute | What the government must show | Exposure |
|---|---|---|
| 18 U.S.C. § 2422(b) | Attempted enticement of a person believed to be under 18 | 10 years to life |
| 18 U.S.C. § 2423(b) | Interstate travel with intent to commit illicit sexual conduct | Up to 30 years |
| 18 U.S.C. § 2251 | Production of child sexual abuse material | 15 to 30 years |
Is it entrapment if the "minor" was an agent?
Almost never, and the reason is narrower than people expect. Entrapment requires that the government implanted a criminal design in an otherwise unwilling person. In Jacobson v. United States (1992), the Supreme Court reversed a conviction after federal agencies spent roughly two and a half years soliciting a Nebraska farmer through fictitious organizations before he ordered anything. The holding: predisposition must exist independently, before the government showed up. A single agent responding to a message and answering honestly when asked doesn't come close.
The second question — how can it be a crime if no child existed? — has an equally settled answer. Federal courts have uniformly held that factual impossibility is no defense to an attempt charge. The offense turns on what the defendant believed and what he did about it.
What I actually do
I read the charging document before I form an opinion, because it's public and it's usually four pages. I hold the word alleged where it belongs, not as a formality but because roughly 97 percent of federal convictions come from guilty pleas, and the small remainder is where wrongful outcomes actually live.
And I've stopped teaching my kid about strangers.
The honest rule of thumb: tonight, don't audit the phone. Ask who they've been talking to for more than a month that you've never met, and make the answer boring to give. The evidence says the danger is duration and secrecy, not anonymity.
What I can't tell you — what nobody can, because the study hasn't been run — is whether proactive undercover operations actually prevent harm to a real child, or whether they mostly surface men who would never have found one. Which is it?
