On Tuesday night, millions of people on Twitter were talking about the recall election in Wisconsin, in which Republican Governor Scott Walker beat his Democratic challenger Tom Barrett. Some were gleeful; some were despondent. And some were violent: "KILL SCOTT WALKER KILL SCOTT WALKER KILL SCOTT WALKER KILL SCOTT WALKER KILL SCOTT WALKER KILL SCOTT WALKER! Ole Bitch Ass Pig Ass Nigga!!!!" tweeted a fellow named @__SupaMcNASTY__.

@__SupaMcNASTY__ has since deleted his Twitter account; I have no clue if he was contacted by the police. But at least one other Twitter user says he did — @SeXXX_Symbol_, who wrote that "Somebody need to Abe Lincoln Scott Walker cave frog lookin ass."

@SeXXX_Symbol_, also known as "The Last Prophecy," wrote a few days ago:

But where @SeXXX_Symbol_ garnered attention, and apparently a visit, from the police, @DomoSoloDolo, or "FatBitchess&GoodWeed," author of the tweet "I wanna kill scott walker so fucking baddd!!!!!" seems to have been left alone. So was @Prototypeisgame, who wrote "Please somebody kill Scott Walker." (All have deleted their offending tweets. Neither @SeXXX_Symbol_ nor @DomoSoloDolo responded to my questions about the police.)

So what does it take to be counted as a legitimate threat on Twitter?

Because, look: there is an enormous amount of shit being talked on the internet. Over a billion people maintain an account on at least one social media platform. No one can figure out what "public" actually means online. Most people think of tweeting as somewhere between sending a letter to all their friends and talking loudly in a public bar. But everything that they write is accessible to anyone by default, and being stored, effectively forever, on a server belonging to (probably) some northern California corporation, and indexed for easy search.

Our friend Mr. McNASTY might think he's talking to a circle of people at a party. The Milwaukee police department might think he's making a sandwich board — "KILL SCOTT WALKER KILL SCOTT WALKER" — and walking around Cathedral Square ringing a bell. (And taking photos of the sandwich board and sending them to San Francisco.) And the weird and hard thing about Twitter is that he's basically doing both things at once, plus a whole range of modes of communication in between. @Prototypeisgame says he "expected a few people" — non-friends or followers — to see his tweet. "I was surprised by how many actually did."

So how do police decide when a tweet is real and scary? When the only threats are direct — in postcards written in shaky script or angry phone calls — or public — sandwich boards or letters to the editor — it seems fairly easy to treat them all seriously. But when threats are ambient, indirect, just as easily barroom shit-talk as legitimate warning, what are police supposed to do? Especially since in many states, the very act of threatening public officials is a crime. (It's also a federal crime to threaten U.S. Government officials.)

As you might expect, the Milwaukee Police Department will only say that it's "evaluating the threats made to the Governor via social media" and that "All threats are handled seriously and can represent a criminal act" and that "People cannot assume anonymity via social media while issuing a threat to another's safety or life." Other law-enforcement agencies we reached to didn't even get back to us.

I tend to imagine that police really only look into Twitter and Facebook threats when there's a significant amount of media attention or popular demand. On Tuesday, Twitchy, the low-rent conservative Buzzfeed operated by pundit Michelle Malkin, collected a few dozen of the most violent tweets directed at Walker and published them; the collection was linked to by the Drudge Report. Soon after that the MPD announced it was looking into the threats

Or maybe they investigate when notable or powerful people complain: a few weeks ago, Los Angeles Lakers player Steve Blake missed a final-seconds three-pointer during a playoff game against the Nuggets; later that evening, he and his wife Kristen received dozens of threatening tweets, which were later, apparently, investigated by the LAPD. Yesterday, The View's Sherri Shepherd filed a police report over a Twitter user who threatened to rape her.

In late February, a federal judge required Twitter to hand over the name of a user who'd threatened to sexually assault Michele Bachmann with a "Vietnam-era machete," even though the judge didn't believe the user to be a threat, ruling that the identity of a user needs to be known so that authorities can accurately assess the threat's legitimacy.

And of course I have a sense that the Secret Service is constantly monitoring Twitter and Facebook on some kind of enormous The Dark Knight-style array of screens. I wonder if they saw this guy yesterday:

As far as I can tell that wasn't a legitimate threat.