Forget any cutesy ledes—we're just going to give this one to you right up front: Republican pollster Gravis Marketing blew a race by 96 points on Tuesday night. Yep, 96 points. It's by far—by far—the biggest polling disaster we've ever seen, so let's talk about it a little.

Back in January, Gravis conducted a poll for former state Del. Michael Smigiel, who was challenging Rep. Andy Harris in the GOP primary in Maryland’s conservative 1st Congressional District. Smigiel's poll gave him an impossible 58-29 lead on Harris, and we knew, just knew, that those numbers had to be total bullshit, but we didn't find out why until Politico's Steve Sheppard discovered that Gravis had conducted a so-called "informed ballot" poll.

In such a poll, respondents are given information about each candidate before asking which they'd prefer in a direct matchup. That's contrasted with an "initial ballot" test, where voters are asked for their preferences without hearing any candidate information; usually you ask both, with the initial ballot, as you'd expect, coming first.

Informed ballots are common practice, but if you're releasing informed ballot numbers, you simply have to explain that that's what they are. Ordinarily, in fact, we wouldn't even bother to assess how well informed ballots stack up against actual election results, since they represent some platonic ideal of a campaign rather than reality. We also usually don't evaluate polls taken four months before Election Day. But we're making an exception here.

That's because Smigiel went out of his way to conceal the fact that this was not an initial ballot test. Indeed, in his press release, he presented his numbers as though they were totally normal. The fact that they weren't only surfaced after Sheppard went digging. This sort of shenanigan would have prompted any legitimate pollster to flip its lid and cut ties to such a shady client, but Gravis did no such thing. So that means this poll is on Gravis, and they have to live with the results.

And the results were beyond brutal. Harris wound up obliterating Smigiel by a 78-11 margin, a net clobbering of 67 points. Remember, though, that Gravis showed Smigiel ahead by 29 points. That means they were off on the final margin by that astounding figure: 96 points. To put this tire fire in context, the worst poll we've ever come across prior to Tuesday was a 52-point miss.

If Gravis wants to try to redeem itself even a little, it can admonish Smigiel for the stunt he pulled (though boy would that be belated), and it can also release the initial ballot numbers that almost certainly were a part of the poll but were just never shared, probably because they looked awful for Smigiel. But really, Gravis's track record is so pitiful, there's really no coming back for this. Just remember, the next time you see any data from them: 96 points.