Party color is not the position

The dots are colored by party so the label is visible. They are placed by a record so the label is not the only fact. Those two choices fight each other, and they should. A chart that colors Republicans red and then also forces them onto the right half has said the same thing twice. A chart that ignores party pretends the label does not matter to anyone in the room. Uncanted draws both, and lets them disagree when the file disagrees.

On the presidential chart the disagreement is historical and ordinary. Theodore Roosevelt is colored as a Republican and sits left of Grover Cleveland, a Democrat, on economics. Cleveland’s note is limited government, the gold standard, and vetoes. Roosevelt’s is trust regulation, conservation, and a strong executive. A reader who “knows” that Republicans are the market party and Democrats are the public party will misread both of those presidencies. The color keeps the reader from forgetting which ticket they ran on. The place keeps the reader from thinking the ticket is the program.

Early presidents make the color problem worse. Federalist, Democratic-Republican, Whig, and Democrat are not the 2026 parties wearing old clothes. The chart does not give them today’s names. Parties before the Civil War use a different color so the eye does not file John Quincy Adams with a modern Democrat or a modern Republican. The placement note still has to do the work. Adams’s national improvements, and his later fight against the gag rule, are the record. The color is only a warning that the word “party” has changed underneath the word.

The current Congress is the case where color and position mostly agree. In the October 2026 Voteview snapshot, every Democrat’s first-dimension score is left of every Republican’s. Uncanted does not invent a crossover to make the picture more interesting. The independents are the split that is actually in the file. Sanders and King vote with the Democrats and are not painted as Democrats. A member who leaves a caucus keeps the votes they already cast. Kevin Kiley’s break with the Republicans does not drag his old roll calls across the chart. The color can change. The past votes do not get rewritten to match the new press release.

Inside a party, color is useless and position is the whole story. Two red dots on top of each other would mean the conference is a point. It is not. The member pages rank each score against the chamber and against the party median. A Republican to the left of their own median is still a Republican on the color. They are also, on the votes, less conservative than the middle of their conference. Both sentences fit on one page because they answer different questions. “What do they call themselves?” and “How have they voted?” are not synonyms, even in a year when the synonyms are close.

Former members are colored by the party they sat with, and placed by the editorial record. Barry Goldwater and Ron Paul can share a party color and sit far apart on authority. Edward Kennedy and a later Democrat can share a color and sit far apart on economics if the records do. The former chart is where a reader can see a party change shape across decades without the site quietly moving the old names to match the new platform. The old names stay where their careers were.

The 2026 chart is the one place where color and position are tied on purpose. A candidate with no voting record is placed on the party program. The color and the dot say the same thing, plus a small display offset so two nominees of one party are not one pixel. That agreement is a limitation, not a discovery. The page says the dot is the party line. A nominee who is running against their party’s program will be mis-drawn until there is a record, or until the file is edited from a public platform that is specific to them. Until then, pretending to know the difference would be a second invention.

Read color as the name tag and position as the notes. If they match, say they match. If they do not, the chart has found the only thing it is really for: a label that does not describe the record. People argue about those cases because the cases are the argument. A compass that irons them out has chosen a side. This one would rather be slightly harder to glance at.

All articles · Methodology