I can't believe this really happened.

A long-running problem in theoretical physics isn’t a lack of imagination—it’s a failure to learn. Sabine Hossenfelder argues that large parts of high-energy theory and related “foundations” work function like mathematical fiction: researchers repeatedly generate elegant-looking models (new particles, forces, modified gravities, extra dimensions, multiverse scenarios) without evidence that the underlying ideas correspond to nature, and then keep publishing despite decades of nonperformance.

The core complaint is methodological. In her view, the scientific method is not just “make hypotheses and test them.” What matters is the quality control that determines which hypotheses are worth testing in the first place, and that quality standards must evolve based on past failures. She contrasts this with what she says has happened since roughly the 1970s: theory development increasingly relies on guessing mathematics that “looks nice,” producing predictions that are later falsified, while the field does not adjust its standards to prevent the next round of low-value speculation. She uses the end-of-the-world date Dec. 31, 2025 as an analogy: it is falsifiable, but it still wouldn’t qualify as a scientific hypothesis because the field has learned that random, evidence-free guesses waste resources.

Hossenfelder also points to a structural reason the cycle persists: publication incentives. Papers that add complexity—often unnecessary assumptions layered onto existing explanations—are easier to generate in bulk, and journals benefit from the resulting citation and subscription dynamics. She argues that this resembles problems seen elsewhere in science, such as psychology’s p-value hacking, where researchers eventually updated standards after widespread irreproducibility. In theoretical physics, she claims, the opposite has happened: instead of refining what counts as a promising hypothesis, the literature piles more guesses on top of earlier ones.

To illustrate the “empty world-building” critique, she highlights inflation cosmology and the pattern of model tweaking. She cites Will Kinney’s assessment that inflation research is highly productive but largely “useless,” driven by endless variations of prior models with no realistic expectation of correctness. She connects this to a broader sociological diagnosis from Jesper Grimstrup’s book *The Ant Mill*, describing mainstream pressures that encourage tribalism and group think, discouraging independent lines of inquiry.

A secondary thread is personal and institutional pressure. Hossenfelder says a physicist contacted her demanding she remove her criticism, and she reports losing affiliation with the Munich Center for Mathematical Philosophy after what she describes as attempts to exert pressure. She frames this as part of why few physicists speak publicly: reputational and career risks make it costly to challenge the prevailing belief system.

Her proposed fix is concrete: journals should stop publishing mathematical fiction by enforcing reviewer guidelines—rejecting new theories that rely on unnecessary assumptions or fail to solve a real problem or advance mathematical consistency. Without such editorial gatekeeping, she argues, the incentives that reward paper production will keep the cycle going, regardless of how obvious the evidence gap appears to outsiders.