Much as we’re understanding drug addiction, as a stand alone state, to be a health problem, the need to get the next fix causes people to do terrible things. Patrick Baker, 43, did something terrible.

Posing as a United States marshal, Mr. Baker killed Donald L. Mills Jr. during a home invasion in May 2014, the U.S. attorney’s office said. Mr. Mills’s wife and children were held at gunpoint while Mr. Baker ransacked the home for oxycodone pills, the office said.

In 2017, Baker was convicted in Kentucky state court of reckless homicide and robbery, and sentenced to 19 years in prison. His incarceration didn’t last long.

In December 2019, Mr. Baker received a pardon from Mr. Bevin, Kentucky’s governor at the time. The pardon was one of hundreds that Mr. Bevin, a Republican, issued in his final days in office after losing a re-election bid.

Regardless of the problems with the drug war, the treatment of addicts and the push against incarceration, this pardon was not taken with equanimity.

After the pardons were made public, lawmakers called for investigations and prosecutors complained that neither they nor the victims’ families had been notified. While a majority of the people freed by Mr. Bevin were serving time for low-level drug offenses, at least a few had been convicted of violent crimes.

Among them was Mr. Baker, whose family hosted a fund-raiser for Mr. Bevin that raised more than $21,000 in 2018, according to The Courier-Journal, which won a Pulitzer Prize in 2020 for its reporting on the pardons.

The upshot was that the feds decided to jump in. Double jeopardy, you ask? Nah. The doctrine of dual sovereignty, observed mostly in the breach, kicked in, Baker was prosecuted, and later pardoned, for his crime under state law. But the feds are a different sovereign, with its own laws and independent interest in prosecution. While Kentucky couldn’t prosecute Baker again, the feds could. So the feds did, with astounding alacrity.

A federal grand jury indicted Mr. Baker in May 2021, and a jury convicted him in August on a federal charge of murder committed during a drug trafficking crime, finding that he had unlawfully caused the death of Mr. Mills by using a firearm during a robbery or kidnapping.

While federal prosecutions of people already tried by the state are, in themselves, highly unusual, this was doubly so in that it was not the product of a state acquittal or conviction, but of a state pardon. When a publicly hated defendant is acquitted, there are often calls for a second bite of the apple, which would be offensive under the notion, if not the rule, of double jeopardy, but for the public’s vacillating support for principle based on how much it hates a particular defendant. When it happens after conviction of a hated defendant, it’s largely shooting fish in a barrel to milk it for whatever political benefit they can grab.

But a pardon? To a significant extent, there’s a federal crime that can be prosecuted as an analogue to most state crimes. Murders are generally prosecuted by states, although they have an obvious federal interstate commerce hook from drugs or guns. And if nothing else, there’s always the civil rights aspect. It violates a person’s civil rights to murder them, unless qualified immunity kicks in, but that’s a different matter inapplicable here.

The point is that if the feds want to prosecute someone, an imaginative prosecutor can almost invariably find a crime to charge, a law broken. And they did.

“The Justice Department believed that in order to fully punish the acts that they believe Mr. Baker caused, they should pursue the case federally, even after the pardon,” he said.

The only limitation of the federal exercise of power under the dual sovereignty doctrine is the Petite Policy.

This policy precludes the initiation or continuation of a federal prosecution, following a prior state or federal prosecution based on substantially the same act(s) or transaction(s) unless three substantive prerequisites are satisfied: first, the matter must involve a substantial federal interest; second, the prior prosecution must have left that interest demonstrably unvindicated; and third, applying the same test that is applicable to all federal prosecutions, the government must believe that the defendant’s conduct constitutes a federal offense, and that the admissible evidence probably will be sufficient to obtain and sustain a conviction by an unbiased trier of fact. In addition, there is a procedural prerequisite to be satisfied, that is, the prosecution must be approved by the appropriate Assistant Attorney General.

Notably missing from this policy is the key distinction here, that the defendant was convicted in Kentucky and sentenced to 19 years’ imprisonment, but subsequently pardoned. Does the case itself involve a “substantial federal interest”? Did the prior prosecution leave that interest “demonstrably unvindicated”? Not really, as this was a relatively mundane drug murder and the defendant was convicted and sentenced to a lengthy term of imprisonment. As far as the crime and prosecution went, there was nothing to see here.

It was the pardon, rather, that changed everything. State officials and their federal friends took serious issue with the pardon, both that it happened and it emitted the unpleasant odor of being bought with some campaign funds to the pardoning governor. And yet, a pardon is a pardon, an act of executive mercy for which a governor need offer no explanation. It is a power that resides in the executive alone, unreviewable and absolute. If the governor sold his power, then he could be prosecuted for his malfeasance, possible, but the pardon was still a pardon.

But the feds decided that this was an unpardonable offense, and prosecuted, raising a new avenue of prosecution and turning state acts of mercy into federal prosecutorial opportunity. Ironically, Baker would almost certainly have been left to do his state time, 19 years, but for the pardon. Instead, he’ll do 42 years, with credit for time served. The math worked against him. The pardon worked against him.

And if this is sufficient cause to overcome the Petite Policy’s limits against dual prosecution, no state pardon of an unpopular defendant, a hated crime or tainted pardon (I’m looking at you, Crazy Joe) is safe from the jeopardy of going from state court “bad” to federal court “worse.” If there is a purpose to a pardon, that it can’t be undone, then that purpose is hardly clear and absolute going forward if this is what the Petite Policy permits.