New to this? Watch the short narrated walkthrough first. It drives the whole lesson once, start to finish. Then scroll on for the interactive and the argument in full.
On a left/right line, the U.S. Overton window looks like it's barely moved in twenty years. On the grid that "left vs. right" erases, it's moved dramatically — just up, not sideways. Drag the year. Watch the window the line was hiding.
The video dragged the window across two decades. Now drag it yourself and watch it climb the axis the left/right line hides.
This is the same Overton window — but flattened onto the left/right line. Watch this bar as you scrub above. It barely moves. The line says: "the window is stable." The grid says: "the window has moved 30% up the control axis." Both are reading the same window. Only one is showing it.
The blob is the U.S. Overton window — the range of policy positions politically acceptable to the mass public at a given moment. The fuzzy edge is intentional: the window doesn't have a sharp boundary. The dashed outline marks where "acceptable" fades into "fringe." The dot is the window's center of gravity.
The two bars below the grid show the same window projected onto each axis separately. The ownership projection is what the standard left/right line shows you. It's been roughly stable since 2008. The control projection is what the line erases — and it's the axis where most of the actual movement happens.
The years are calibrated keyframes. Window positions interpolate between them. Coordinates are approximations grounded in policy salience, not survey data — this is a frame, not a measurement.
Across every keyframe, the window's horizontal position barely changes. The U.S. mass public has not, since 2008, seriously entertained collective ownership of capital. Single-payer healthcare, public housing at scale, worker ownership, nationalized energy — all of it stays outside the window across every administration.
The window's vertical position has moved a lot. Tea Party pulled the bottom edge down (anti-government, libertarian-capitalist). Trump 1.0 stretched the top edge up (mass executive power, immigration enforcement, tariffs). COVID and 2020 stretched it further. Trump 2.0 has the window centered upper-right, with the entire bottom-left half of the grid effectively unreachable.
This is the move the left/right line cannot represent. Authoritarian capitalism is right-and-up. Libertarian capitalism is right-and-down. They are both "right." The line collapses them. The grid keeps them separate. Twenty years of American politics is mostly a fight on the axis the line erases.
This grid fixes one thing the line got wrong: it restores the control axis, the direction where most of the movement has actually happened. That is real. But it leaves a second thing untouched, and the second thing is the one that decides who wins. The compass locates a position. It has no way to show a capacity, the standing ability to absorb an attack and still be there next week. Two collectives can sit on the exact same coordinate and have opposite fates. Watch the compass fail to tell them apart.
One position. One coordinate. Ownership: collective. Control: low. The compass has now told you everything it knows about this collective. Press the control below and watch it run out of things to say.
Has a legal defense fund. A safe-space network. Food and medical logistics. People who have trained together and will show up the second time. When power moves on it, it absorbs the hit and answers.
Can hold ground.Has an account and a better theory. No fund, no logistics, no one who has done this before. When power moves on it, there is nothing to fall back to.
Dissolves on contact.Same coordinate. The compass plotted one point for both, because capacity is not a direction on this grid. One survives the month. One does not. The map cannot see the difference that decides everything.
The scrubber above concedes one honest thing the line refuses: the boundary moves. But it presents that movement as ambient, a climate, a tide that simply comes in. The missing mechanism is infrastructure. Windows move because some actor has the standing capacity to move them: organized bodies, money, media reach, the ability to put people in a street repeatedly and keep them fed and out of jail. The rightward-and-upward shift of the last twenty years is not weather. It is the visible output of infrastructure-building by actors who understood that persistent capacity, not persuasive positioning, relocates the boundary. The compass shows the effect and hides the cause.
A map that shows positions but not capacity is not neutral. It trains you to experience politics as a taxonomy of opinion, a game of correct placement, a tool for spectators. It answers "what do they believe" and never "what can they do," and teaches that the first question is the important one. Shantz's book is a refusal of that teaching: the point is not to occupy the right coordinate, it is to build the infrastructure that lets people defend themselves and go on the offensive.
The one-line version. The compass measures what people think. Infrastructure is what people can survive. The gap between those two is where the actual politics happens, and it is exactly the gap the compass is built to hide.
Source: Jeff Shantz, Insurrectionary Infrastructures (punctum books, 2018). Open access: punctumbooks.com · DOI 10.1353/book.84171.
Multi-constituency overlay. Three windows instead of one: elite/donor consensus, mass public, online activist base. They don't overlap cleanly, and the gaps are where most of the country's political pain lives. Toggleable layers.
Policy pins. Plot specific policies as fixed points (M4A, ACA, mass deportation, child tax credit, Patriot Act, UBI). Watch which ones are inside the window and which fall outside as the year scrubs.
Country toggle. Same frame, different windows. A U.K. Overton window has a meaningfully different shape — left edge extends further toward collective. Worth contrasting.
Survey-grounded edition. Replace the calibrated keyframes with windows derived from ANES / Pew issue-acceptance data. Less editorial, more empirical. Different deliverable.