The Information Crisis
Every time a human cell divides, it copies about 6.4 billion base pairs of DNA, two sets of 3.2 billion, with extraordinary accuracy. A single change in the wrong gene can contribute to cancer. The raw error rate of DNA polymerase, the molecule that does the copying, is approximately 1 mistake per 100,000 base pairs.
That rate, if uncorrected, would produce roughly 64,000 errors per cell division. A human being has approximately 37 trillion cells. The math is unambiguous: uncorrected replication is incompatible with biological complexity.
In 1971, German physicist and chemist Manfred Eigen (b. 1927) published the mathematical framework that made this problem precise. He proved that there exists a sharp threshold, the error threshold, above which genetic information loses stable maintenance. Cross the threshold from below, and the genome dissolves into random noise: what he called error catastrophe.
"There is a maximum genome length compatible with stable inheritance, determined by the product of genome length and error rate per base. Above this threshold, the master sequence is lost."
Three Correction Systems
The cell operates three sequential fidelity layers: base selection by the polymerase, 3'→5' exonuclease proofreading, and mismatch repair. Each acts in turn on every copied base. All three are required; weakening any one causes a measurable increase in mutation rate and, in humans, cancer predisposition syndromes. Base excision and nucleotide excision repair add genome maintenance between replications.
The polymerase active site reads each incoming base by geometry, accepting only nucleotides that pair correctly with the template. Base selection alone holds the error rate to about 1 in 100,000 bases (10⁻⁵).
The replicative polymerase has a built-in 3'→5' exonuclease domain. After inserting each base, it pauses to verify geometry. Incorrect bases stall the active site: the exonuclease then excises and re-inserts. Reduces error rate by about 100-fold, from 10⁻⁵ to 10⁻⁷.
After replication, MutS scans for geometric distortions in the double helix: bulges from mispaired bases. MutL recruits MutH, which cleaves the newly synthesized strand (identified by delayed GATC methylation). The error is excised and re-synthesized. Reduces rate by about 100-fold, from 10⁻⁷ to 10⁻⁹.
Together the layers deliver about a 10⁴-fold gain beyond base selection (10² from proofreading, 10² from mismatch repair) and a net rate near 10⁻⁹ per base per replication. That leaves about 3 new changes per 3.2 billion base set, about 6 per diploid division.
The astonishing detail: these layers are mechanistically independent. They use different proteins, different recognition mechanisms, different chemistry. They form a load-bearing stack rather than a redundant system where one could evolve apart from the others. Loss of mismatch repair causes Lynch syndrome, a hereditary cancer predisposition. The genome maintenance pathways that run between replications carry their own syndromes: xeroderma pigmentosum (NER deficiency) and MUTYH-associated polyposis (BER deficiency).
III. The Goldilocks ExplorerThe Error Threshold Window
Eigen's theorem specifies a precise mathematical relationship between mutation rate, genome length, and information maintenance. Move any slider below and explore how narrow the viable zone is. The parameters are taken from Eigen's 1971 paper and subsequent empirical measurement.
Survival Probability
What This Requires
The fidelity layers had to appear together in evolutionary history: each requires the others to function at all. Base selection sets the raw rate. Proofreading brings that rate down to where MMR can handle it. MMR brings the residual rate down to where a large genome can survive it. Base excision repair, running between replications, clears the chemical damage that would otherwise swamp the system.
Eigen's error threshold theorem sets the constraint: the larger the genome, the lower the error rate it must hold to keep its information. With base selection and proofreading alone, the achievable rate is ~10⁻⁷. All three layers, working in sequence, hold the measured rate near 10⁻⁹, about one error per billion copied bases.
This is a case of irreducible minimum specification rather than redundant backup: the information storage capacity of the genome exists only with all three layers of correction in place, acting in sequence on every copied base. The question of how the information arose before the storage system required to maintain it stood in place is precisely Eigen's Paradox, and it remains unresolved.