How it works

Nothing hidden except the names.

Same Question puts candidates' own words next to each other, so you can judge what they said, not what someone said about them. Here's exactly how it works.

Where the words come from

Every quote comes from a public debate or forum where both candidates got the same questions from the same moderator. Most are debates with both on stage. In Florida there's no debate, so we use a forum where each candidate sat in the same chair, one after the other. We transcribe the full debate, find each candidate's answer to each question, and link to the exact moment in the original broadcast. If you think we've missed something, play the full answer. It's always one tap away.

Quotes are word for word. The only things we take out are filler like "uh" and repeated false starts like "I've, I've." Where we skip part of a sentence to get to the point, you'll see "...". The video always has the complete answer.

We only use questions both candidates were asked. If a candidate wasn't asked something, we say so. We don't fill the gap with something they said somewhere else.

Our fairness rules

Same question, equal footing. Both candidates answer the same question. Who shows up as ◆ and who shows up as ● is a coin flip for each visitor.

The most on-point words, never the worst moment. The short quote is the sentence that most directly answers the question. The video plays the whole answer.

Blind means blind. While names are hidden, we swap out names and party words for [brackets]. Nothing else in the quote changes.

We say when someone didn't answer. If a candidate talked about something else instead, we mark it "Didn't answer," and it doesn't count toward anyone's match. Both sides do this. We report it the same way for both.

No colors for parties. No red, no blue. Candidates are only told apart by ◆ and ●.

Every placement can be questioned. "Did we get this wrong?" sits under every answer. A person reads every note, and we publish what we change.

We never tell you who to vote for. We show you how close you are. What you do with that is up to you.

How we place a candidate

For each question we turn the answer into a plain statement, like "Pause the federal gas tax." Then we place each candidate on the same five-step scale you use, from strongly disagree to strongly agree, based only on what they said in that answer. The quote that supports each placement sits right next to it. Placements are reviewed by people before they go live.

How your match is scored

This is the same proximity method Glassbooth published back in 2007, with one change.

For each question: if you and a candidate are on the same step, that's 100%. One step apart is 75%, two is 50%, three is 25%, and opposite ends is 0%.

Weighting: each question counts 1, plus the points you put on its issue. So an issue you gave 3 points counts 4 times as much as one you gave none.

Your match is the weighted average across the questions you answered. Questions a candidate didn't answer, and questions you skipped, are left out.

The one change: Glassbooth only asked about the issues you gave points to. We ask everything, so you see the whole picture, and the points change how much each question counts.

Separately, we count your blind picks, the times you tapped "closer to me" before you knew who said it. Your match and your blind picks don't always agree. That's part of what's interesting.

Your privacy

No accounts, no sign-up, no ads, no tracking pixels.
Your answers stay in your browser. They are never sent to us.
Each time you tap an answer with the names hidden, we add one to an anonymous count for that answer. If you guess who said it, we add whether the guess was right. That's where "how others picked" and "how many guessed right" come from. Nothing about you goes with it.
Today's six questions are the same for everyone, so results can be compared. Your result and your streak are saved only in your own browser.
If you send us a question, we keep the question and, if you give it, your ZIP code. Nothing else.
Videos play through YouTube's privacy-enhanced player and only load when you press play.

What's coming

This is a preview built for the 2026 midterms. It covers a few close Senate races, using the debates that already happened. Here's what we're building for 2028:

Every race on your ballot. Down to school board and city council, where there's usually the least information.
Questions from voters. The questions people send us become the questions we ask every candidate, on the record.
The whole record, over time. Debates, town halls, interviews and votes, with every claim linked to where it was said, so you can see how someone's position has moved.
Where neighbors agree. How much common ground exists across party lines in your own community.

Who's behind this

Same Question is made by Civic Designers, a nonpartisan civic design group. We don't endorse candidates, and we take no money from campaigns, parties or PACs. The transcripts and matching run on SurfaceThread, a tool we built to help communities turn what people say into knowledge they can see and share, with every line traced back to its source.

It's inspired by Glassbooth, a nonprofit quiz from 2007 that matched voters to candidates using the issues they cared about. It was a good idea that arrived before the technology to do it everywhere. We think that's finally changed.

Questions or corrections? Use "Did we get this wrong?" on any answer.