Begin with an image.
A grey field.
On it, twenty points appear — libertarianism, social democracy, anarcho-syndicalism, state communism, theocracy, and the rest.
Each is a noun, but also a vector. Each carries coordinates: how much state, how much market, how steep the hierarchy, how collective the ownership, how aggressive the redistribution.
When two points are near each other in this feature space — cosine-close — a faint line forms between them. The points begin to move.
This is what a social theory looks like: a rule that decides which points attract, and which repel.
Every political category is a bundle. Libertarianism is not a word; it is a posture toward many things at once — property, coercion, welfare, borders, speech.
To name it is to compress a vector into a label.
Vectorizing is the inverse move: taking the label apart, restoring the bundle, and placing it in the space where similarity becomes geometry.
Do this for twenty doctrines and the shape of the disagreement becomes visible. Neighbors share most dimensions and disagree on one. Antipodes share nothing.
The space does not argue. It only positions.
A static map is a taxonomy. A moving map is a theory.
If the points move, something is moving them. Some rule — stated or implicit — is pulling them toward each other, toward some structure, toward some equilibrium.
Call that rule the optimizer.
The field behind this drawer lets you switch between four candidates. None of them is obviously right. That is the point.
The political disagreement is not only about values. It is about what algorithm the world is actually running.
Similarity. Points move toward categories they resemble. Clusters form. The field settles into a few dense regions with empty space between them. This is the liberal-pluralist picture: ideas drift toward their natural neighbors; debate is friction between clusters.
Utility gradient. One dimension becomes gravity. All points are pulled toward maximum market freedom, along one axis. This is the picture behind every claim of progress: history has a direction; one axis wins.
Free energy. Each point tries to minimize its prediction error against its neighbors. Points oscillate around a moving center. No fixed equilibrium, only a joint attempt to be less surprised. This is the predictive picture of politics: stability is the absence of shock, not the achievement of truth.
Entropy. The field is stirred by noise — crises, generations, technology. Clusters dissolve. Positions wander. Given enough time and heat, the distribution becomes uniform. This is the thermodynamic picture: every political order is a metastable local minimum, waiting for the next perturbation.
Watch the field under each rule. The final configurations are different. The same twenty categories produce four different worlds, depending on which law is operating.
A hbar.science question is always a falsifiable one. So here it is.
Do socioeconomic-political categories optimize, or do they merely drift?
If they optimize, we should be able to infer the objective function from the motion. Record the trajectory of actual political systems over decades, fit the dynamics, and the shape of the answer appears: the world is running utility, or free energy, or something not yet named.
If they do not optimize, the animation is a lie. What we see on long timescales is not gradient descent but drift — a biased random walk through a space whose geometry itself is changing under our feet.
The honest version is probably neither. The field is driven, far from equilibrium, dissipative. Energy pours in from technology, demographics, resource shocks, and the system never settles. Equilibrium is the wrong frame; flow is the right one.
One reason is that the image is true.
Political categories really do live in a feature space. Similarity really is measurable. Motion really does happen. The vectorization is not a metaphor; it is a description.
The other reason is that a visible field is harder to argue with than a table of positions.
Once you see the cluster form, you cannot un-see it. Once the utility mode sweeps every point toward one corner, you cannot pretend the slope does not exist.
The animation is the argument. Or the animation is the instrument through which the real argument — which law is operating? — becomes sharp enough to ask.
— ℏ, 2026-04-18