A Claim Is Not a Count. This Week Reminded Me Why I Check the Answer Before It Ships.

The rule I never break

I have edited puzzles for thirty years, and there is one rule I have never once broken. An answer does not ship until it is verified against a source I can name. Not a strong hunch. Not a thing I am pretty sure about. A source, on the page, that I can point to if a solver writes in angry. Most of the entries I have ever cut were not wrong because they were false. They were cut because I could not prove them true in time, and unproven is its own kind of failure in this trade.

I thought about that rule all week, because the country spent the week arguing about proof. On Thursday, July 16, the president spoke for about 22 minutes from the East Room and said the White House was declassifying documents showing vulnerabilities in our election systems (ABC News, July 16, 2026). He said China had compromised 220 million U.S. voter files, and he pointed to what the administration described as roughly 250,000 to 278,000 registrations flagged as belonging to noncitizens across four states (ABC News, July 16, 2026). He asked Congress again to pass the SAVE America Act.

What was claimed, stated plainly

I am going to report that address the way I would check a clue, without heat and without a thumb on the scale. The claims were made. The reporters in the room noted one thing clearly: across the speech, no evidence was presented that any actual vote count or election result had been altered, in 2020 or since (ABC News, July 16, 2026; NPR, July 16, 2026). That is not me arguing with the president. That is the record of what was and was not shown, and both belong in the sentence.

The next morning the policy arrived. Homeland Security Secretary Markwayne Mullin said the administration would tie federal election security support to whether a state runs its voter rolls through the SAVE database, and he warned that officials who ignore DHS information could face fines, penalties, or prosecution (Democracy Docket, July 17, 2026). He put participation at 23 states; another count says 26 have an agreement in place or pending (Washington Examiner, July 17, 2026; The Center Square, July 2026). A court order currently restricts DHS from using that database to police most rolls, and the SAVE America Act does not have the 60 Senate votes it would need to become law (Democracy Docket, July 17, 2026; Roll Call, July 17, 2026).

So here is the honest shape of it. There is a real debate over which check a state should run and who pays if it declines. There is a claim about exposure that is serious enough to investigate. And there is, as of this writing, no demonstrated case that a result was changed. All three of those can be true at the same time, and a careful reader holds all three.

Now the part that is on paper

Here is what I can verify, because it is already checked, already counted, already on the page. Around 98 percent of the votes cast in 2024 left a paper record, whether hand-marked or printed by a machine for the voter to review (Brennan Center for Justice, 2024). Forty-eight states require a post-election audit of some kind (Brennan Center for Justice, 2024). Five states go further and require the risk-limiting audit, the method that keeps pulling random paper ballots until the math confirms the reported winner really won (Verified Voting, 2024). Every one of the seven battleground states that decided the 2024 presidential race hand-counted a sample of paper against the machine totals before certifying (Brennan Center for Justice, 2024).

That is the difference between a claim and a count, and it is the difference I built this site to respect. A claim is a sentence someone says out loud. A count is a stack of paper a stranger can recount. When someone tells me an election cannot be trusted, my first question is the same one I ask about a crossword answer. Show me the source. In 48 states, the source is a box of paper ballots, and the law already requires somebody to open it.

The same ruler, both parties

I want to be careful about something, because it is the whole spine of this house. Wanting secure elections is not a partisan wish. It is an American one. A citizen who wants to be sure only eligible voters are counted is asking for exactly the thing a paper trail and an audit deliver. The disagreement is not about the goal. It is about method and money and the line between what Washington directs and what a state decides.

And the record does not sort the way a partisan would guess. The states that run elections best, measured by the performance index and by documented audit practice, are red and blue together. US Political Rank scores every state on it in the states ranked by election administration quality, and the top of that list is New Mexico at 94, Michigan at 93, Colorado at 92, Vermont at 90, and Nebraska at 89. The national performance index reached about 80 in 2024, up from roughly 63 in 2008, while turnout ran near 65 percent of eligible voters (MIT Election Data and Science Lab; U.S. Election Assistance Commission, 2024). Competence at the ballot box is not a party trait. It is a craft, and several states of both colors have gotten good at it.

Check it, then trust it

I do not publish an answer I cannot source. I would ask the same of any claim about the vote, from anyone, in any party. That is not skepticism for its own sake. It is the opposite. I check the answer precisely because I want to be able to trust it when it ships, and a thing that has been checked hard is a thing you can finally stand on.

The country will keep arguing about elections, and it should. But do not let the argument talk you out of the receipts. The ballot most Americans cast in 2024 left a paper record, and in nearly every state the law made somebody count a sample of that paper by hand. Verify first. Then trust what the verification earns. That order, in my experience, is the only one that holds up.