Cook Moved Alaska to Toss Up. Here Is What a Toss Up Actually Means.

First, a short introduction

My name is Jameel Gibson, and this is my first column for US Political Rank. I cover Congress here, which in practice means I cover the datasets that describe Congress: race ratings, roll call records, effectiveness scores, election returns. I maintain a spreadsheet of every 2026 Senate rating change since spring. It has conditional formatting. I am, as you can imagine, tremendous fun at parties. The deal I will make with you in this space is simple. I will show you where every number comes from, I will tell you when the data runs out and judgment takes over, and I will not pretend a coin flip is a forecast.

What actually happened in Alaska this week

On July 1, the Cook Political Report moved the Alaska Senate race from Lean Republican to Toss Up (Cook Political Report, July 1, 2026). The immediate cause was not a poll, a scandal, or a fundraising report. It was a ballot. Two days earlier, on June 29, the Alaska Supreme Court ruled that Dan J. Sullivan, a retired teacher from Petersburg, must appear on the August 18 primary ballot alongside the man he happens to share a name with, incumbent Republican Sen. Dan S. Sullivan (Alaska Beacon, June 29, 2026; Anchorage Daily News, June 29, 2026). The Division of Elections had kept the challenger off the ballot. A superior court judge, Thomas A. Matthews, ordered him restored on June 26, and the state's highest court affirmed that order roughly three hours after hearing oral argument (Alaska Beacon, June 29, 2026). The division's sample ballot now lists the challenger as "Sullivan, Daniel J. Jr." with no party affiliation, while the senator appears as "Sullivan, Dan S." with the notation registered Republican and incumbent (Ballotpedia News, June 30, 2026).

Why does a name on a primary ballot move a Senate rating? Because Alaska uses a top-four primary feeding a ranked choice general election, and both parties are now preparing for the possibility that the second Sullivan advances to November and siphons at least a few points from the first one (Cook Political Report, July 1, 2026). In a state where Democrats recruited former Rep. Mary Peltola, a candidate Cook describes as a proven electoral overperformer, a few points is the whole ballgame. Alaska has not elected a Democratic senator since Mark Begich in 2008 (Ballotpedia, 2026). The most consequential political actor in America this week may be a retired teacher from Petersburg, and I say that as someone who logs these things professionally.

What a rating change is, and what it is not

Here is the part that headlines routinely get wrong. A move to Toss Up is not a prediction that the incumbent will lose. It is not even a prediction that the race is exactly fifty-fifty. It is an analyst's declaration that the available evidence no longer supports giving either side an edge. The information moved; the votes have not been cast. Ratings are built brick by brick from candidate quality, fundraising, state partisanship, polling where it exists, and, as Alaska just demonstrated, the mechanical facts of the ballot itself. I will concede the honest caveat up front: the internal weighting is judgment, not formula, and Cook does not publish a model we can audit. What we can audit is the track record, because forty years of ratings have been graded against forty years of results.

The track record: what forty years of graded ratings show

Cook's own accuracy review, covering 1984 through 2022, found that races rated Solid went the rated way 99.9 percent of the time, Likely races 97.4 percent, and Lean races 95.3 percent (Cook Political Report, ratings accuracy review, 1984-2022). Read that ladder carefully, because it is the whole epistemology of the enterprise. Each step down trades certainty for competitiveness, and the drop from Lean to Toss Up is the step where the ladder simply ends. A Toss Up carries no directional claim to be right or wrong about.

The independent evidence says the label is honest. Political scientist James E. Campbell of the University at Buffalo studied Cook's pre-Labor Day ratings across eleven election cycles and found that Democratic-held seats rated Toss Up stayed Democratic 49.2 percent of the time, while Republican-held Toss Ups stayed Republican 55.0 percent of the time (Campbell, The Accuracy of the Cook Report's Pre-Labor Day Ratings, University at Buffalo). That is remarkably close to an actual coin flip, with a modest thumb on the scale for the party already holding the seat. And the label has been getting sharper, not looser. In 2022, nearly 98 percent of races Cook rated Toss Up were decided by under 10 points of two-party vote share (Cook Political Report, ratings accuracy review). In other words, when the raters say a race could go either way, it almost always turns out to be the kind of race that could have gone either way. That sounds circular until you consider how rare calibrated humility is in this business.

One more caveat, freely admitted: Campbell measured pre-Labor Day ratings, and it is July. Ratings sharpen as elections approach, and a July Toss Up has more time to drift back out of the column than an October one. Alaska could be Lean Republican again by September if the second Sullivan fails to advance on August 18. The rating is a snapshot of uncertainty, and uncertainty has a shelf life.

The math this one rating touches

Now place the snapshot in the frame. Republicans defend a 53 to 47 Senate majority, counting the independents who caucus with Democrats, which means Democrats need a net gain of four seats (CNN, July 2, 2026). Alaska's move follows a June 11 shift in North Carolina, where Cook moved the open-seat race between former Gov. Roy Cooper and former RNC chair Michael Whatley from Toss Up to Lean Democrat (Carolina Journal, June 2026). Alaska now joins Maine, Ohio, and the open Michigan seat in Cook's Toss Up column (Cook Political Report ratings, July 2026). Our own consensus ranking, which aggregates Cook, Sabato's Crystal Ball, and Inside Elections, places Alaska fourth among the fifteen most flippable Senate seats of 2026, with a flip-likelihood score of 6.9 out of 10, behind only North Carolina, Maine, and Ohio (US Political Rank, July 2026).

Apply the historical base rates to that map and the arithmetic gets interesting without anyone needing to shout. If Lean seats hold about 95 percent of the time and Toss Ups split near even, Democrats reaching four net flips requires essentially running the table on the coin flips while defending Michigan, their own most exposed seat. That is possible. Toss Ups do sometimes break together in a strong national environment. It is not, on the July evidence, likely, and the spreadsheet does not care which outcome anyone is rooting for.

The precise conclusion

So here is what July 1 actually established, stated as narrowly as the data allows. A court ruling changed the composition of a ballot. That ruling erased the informational basis for a Republican edge in one state, so the professional raters filed the race where the evidence now sits, in the column that historically splits about 49 to 55 percent depending on which party is defending. It moved the Democrats' path to a majority from implausible toward merely difficult. That is all it established, and that is plenty. When Alaska moves again, in either direction, the movement itself will be the news, and I will be here with the conditional formatting to log it.