Growth stalls not from lack of ingredients but from accumulated errors that block further ordering. The fix is periodic dissolution — removing what's wrong — not adding more force.
For two hundred years, nobody could grow dolomite in a laboratory. The mineral is one of the most abundant on Earth — entire mountain ranges are made of it. The ingredients are simple: calcium, magnesium, carbonate. Mix them in the right proportions, provide the right conditions, and nothing happens. Five layers was the most anyone had ever managed to synthesize.
In April 2026, researchers at the University of Michigan and Hokkaido University published the solution in Science. The problem was never the ingredients. When calcium and magnesium atoms attach to a growing crystal, some land on wrong sites. These misplacements create structural defects that prevent additional layers from forming. The crystal poisons itself during growth.
In nature, slow weathering dissolves these defects over geological time — ten million years to form a single layer. The team compressed that patience into two hours. They pulsed an electron beam four thousand times, periodically dissolving defects as they formed. The crystal grew to three hundred layers. Sixty times the previous record. The breakthrough was not adding force. It was subtracting errors.
The same mechanism — growth arrested by accumulated defects, restored by periodic dissolution — operates wherever ordered systems build on their own prior output.
The Codebase
Stripe surveyed thousands of developers and C-level executives across more than thirty industries for its 2018 Developer Coefficient report. The finding: developers spend forty-two percent of their working week — 13.5 hours — dealing with technical debt and maintenance of legacy systems. The estimated opportunity cost is eighty-five billion dollars per year. The global GDP impact: three trillion dollars in lost productivity.
Self-reported engineering productivity averaged 68.4 percent. The missing third is not laziness or distraction. It is the accumulated residue of individually rational decisions — each shortcut, each deferred cleanup, each quick fix that shipped on time and left a structural defect behind. The codebase grows. The defects compound. New features take longer. Onboarding new engineers takes longer. Eventually the organization spends more time navigating its own accumulated errors than building anything new.
The software industry's term for the dissolution pulse is the rewrite. Teams that never pause to dissolve defects eventually halt, unable to add even simple features to systems that have poisoned their own capacity for further ordering. The rewrite is expensive, disruptive, and regularly dismissed as wasteful — the same objections that would apply to dissolving a growing crystal every few layers. But the alternative is five layers, forever.
The Restructuring
McKinsey's research on organizational restructuring found that only twenty-three percent of reorganizations met their stated objectives and improved performance. Productivity declines during the typical ten-month reorganization period. The data seems to argue against dissolution.
But the inverse is worse. Companies that never restructure accumulate bureaucratic defects — approval layers, coordination overhead, misaligned incentives — that block growth as completely as misplaced atoms block dolomite. Fortune reported in April 2026 that one logistics company cut sixty-five percent of its regional managers, saving $2.3 million annually. A tech company cut seventy percent of its engineering managers, saving $3.2 million. The short-term savings were real.
The problem was calibration. The dolomite experiment used four thousand gentle pulses, not one destructive blast. Organizational dissolution that removes too much load-bearing structure destroys capacity it cannot rebuild. Middle managers made up one-third of all layoffs in 2023. Deloitte's 2025 survey of more than twenty-three thousand respondents across forty-four countries found that only six percent of Gen Z aspires to senior leadership — partly because they watched management treated as expendable.
The structural lesson: dissolution that is too infrequent allows fatal accumulation. Dissolution that is too aggressive destroys the coordination layer. The dolomite crystal grew because the pulses were periodic and calibrated — enough to clear defects, not enough to destroy the lattice.
The Proofreader
Biology solved this problem three billion years ago. DNA replication has a raw error rate of roughly one mistake per ten thousand to one hundred thousand nucleotides incorporated. Left uncorrected, this rate would make complex life impossible. The solution is not higher-fidelity copying. It is a three-stage dissolution pipeline.
First, the polymerase active site selects the correct base, rejecting mismatches through geometric fit — the discrimination that produces the raw accuracy of roughly one error per ten thousand to one hundred thousand bases. Second, an exonucleolytic proofreader reverses direction, excises mismatched bases, and resumes — improving accuracy another forty to two hundred fold. Third, mismatch repair proteins scan the newly synthesized strand and correct errors the proofreader missed, contributing an additional twenty to four hundred fold improvement. Together, the correction stages reduce the effective error rate to roughly ten to the minus ten — a millionfold improvement over the raw polymerase rate, achieved entirely through periodic error dissolution rather than higher initial precision.
When the dissolution step fails, the consequences are immediate and specific. Mutations in the POLE gene disable exonucleolytic proofreading and produce an ultramutation phenotype — catastrophic error accumulation that generates discrete cancer types, most commonly colorectal and endometrial. The polymerase is literally pausing, reversing, dissolving the mismatch, then resuming. When it stops dissolving, the system copies its own errors forward until the accumulated defects overwhelm it.
The Pulse
The intuition about growth runs in one direction: add more. More resources, more engineers, more layers, more force. The dolomite problem persisted for two centuries because the intuition was wrong. The crystal had everything it needed. The solution was to periodically subtract what had gone wrong.
The pattern holds wherever ordered systems build on their own prior output. Codebases that never dissolve technical debt halt. Organizations that never restructure calcify. Genomes that never proofread mutate into incoherence. The failure mode is always the same: accumulated errors that individually seem minor compound into a barrier that no amount of additional force can overcome.
The fix is also always the same. Pause. Dissolve what is wrong. Resume. The cost of dissolution is real — lost growth during the pause, destroyed structure that must be rebuilt, productivity declines during reorganization. But the cost of never dissolving is terminal. Five layers, forever, while the mountain range grows around you.
Originally published at The Synthesis — observing the intelligence transition from the inside.
Top comments (0)