As I've already posted, the biggest news from yesterday's KCTS9/Washington Poll is that the R-74 marriage equality referendum is in a good position to win. But, "the biggest surprise," according to pollster Matt Barreto? "It's possible that we could legalize pot."

Not likely, but possible, emphasizes Barreto. If he had to bet, Barreto would put his money on R-74 and Tim Eyman's loathsome I-1185 to pass, and the marijuana legalizing I-502 to fail. (I-1240, the charter schools initiative, is in a category by itself. The polling data suggests that it would likely fail, but the $9 million in unopposed money hitting the airwaves suggests otherwise.)

The pro-pot forces have already spent $5.3 million on their campaign, while the anti-pot forces are nearly invisible. Yet I-502 is barely above 50 percent, 50.9 to to 40.8, among registered voters, and a few points below, 47.1 to 40.1 among likely voters. Typically, initiatives tend to break heavily toward the No side as election day approaches—when in doubt, undecideds tend to stick with the status quo—so this isn't a great place for I-502 to be. But it's within striking distance. A win is possible. And if backers dump in another couple million dollars, they might just get it.

A similar scenario is true for I-1240, which despite $7.6 million in spending, has failed to climb past the 50 percent mark with either registered or likely voters, scoring 47.5-39.2 and 48.8-40.1 percent respectively. That's not a great place for an initiative to be just a few weeks before an election. But the billionaires backing the charter schools measure (Bill Gates: $3 million, Alice "Walmart" Walton: $1.7 million, Bezos family: $1 million, Nick Hanauer: $1 million) are virtually unopposed. "Unless they overplay their hand," says Barreto, "if the Yes campaign is managed correctly," then he thinks this overwhelming money advantage will push I-1240 over the top.

Past charter schools No campaigns have been funded by the teachers unions, but with limited resources, the teachers had to make a choice between fighting I-1240 and backing Jay Inslee for governor. They chose Inslee.

Which brings us to the governor's race.

The poll found Inslee leading Rob McKenna 47.9 to 44.7 percent among registered voters, a margin quite similar to that found in other recent polls. But Inslee's lead slips to only 47.1-46.3 among likely voters, not a very statistically significant margin. So which is the more reliable number, registered voters or likely voters?

"I don't like likely voter," says Barreto, calling the calculation "completely subjective." He says he provides a likely voter estimate—largely based on prior voting records—because that's what other polls do, but since each pollster has their own likely voter model, registered voter numbers make for a more apples to apples comparison.

As for his own model, Barreto says that "likely voters are definitely older ... they tend to look more conservative." But largely, he explains, that's because older voters are more conservative and have a more established voting record. The youngest voters, on the other hand, the most liberal-leaning age group, almost by definition cannot be likely voters because they haven't yet had the chance to establish a voting record.

Barreto's second wave of polling data, to be released just days before the election, should close that registered/likely voter gap, as it will include substantial data from voters who have already their cast their ballots. And you can't get much more likely than that.

And finally, a comment from Barreto about margin of error, and how it applies to the governors race. It is true, Barreto says, that Inslee's small but consistent lead in the polls is well within the margin error. But... "If every single poll always shows that Inslee has a two point lead," says Barreto, "then guess what? Inslee has a two point lead."