In 1990 Danny Hillis let a population of test cases evolve against the sorting networks they were testing. Seven parts follow: that experiment, running live; four more co-evolutionary systems to run and break; the history of the idea to 2026; and its counterpart in the generator and discriminator of a GAN.
Hillis evolved 16-input sorting networks on a Connection Machine with 65,536 processors. Each host carries 15 pairs of chromosomes and each chromosome encodes 4 exchanges. Where the two copies agree the network gets one exchange, where they differ it gets both, so a network has between 60 and 120 exchanges. Hosts live on a wrap-around grid, the better-scoring half survives each generation, and neighbors mate.
Scored on a sample of test inputs, his best network had 65 exchanges and the population stopped improving after a few hundred generations. He then put a second population on the same grid: parasites, each a set of 10 to 20 test inputs, scored by how many of their tests the host in the same cell fails. With parasites the best network had 61 exchanges, one more than Green's.
Pale patches in the left map are groups of related hosts with few doubled exchanges. With parasites on, the right map shows where tests are currently beating their hosts; the patches move as hosts and parasites replace each other.
The smallest verified sorter will be drawn here.
Everything in the panel is computed on this device. "Verified" means the network sorted all 2n inputs of zeros and ones, which by the zero-one principle means it sorts any input. With 16 inputs and co-evolving parasites, expect the first verified sorter after 500 to 850 generations, as in the reference runs; smaller networks get there in seconds.
Each paused or replaced run is added here, next to reference runs of 8,000 generations.
| Show | Run | Tests come from | Inputs | Hosts | Generations | Smallest verified sorter | Unused removed | Mean size at end |
|---|
No method has produced a 16-input network with fewer than 60 exchanges. Green's network remains the smallest known, and the lower bound in Bert Dobbelaere's list stands at 57. Evolutionary search did set records elsewhere: Juillé found a 13-input network with 45 exchanges in 1995, and Valsalam and Miikkulainen's symmetry-based method improved the best known sizes for several input counts above 16. Minimum depth for 11 to 16 inputs was settled by SAT solving (Bundala and Závodný, 2014), and minimum size is proven only up to 12 inputs.
Hillis's result was not guaranteed. Two populations that are scored only against each other can stop improving in several distinct ways, and the scores they see give no warning. The smallest setting that shows all of them is a number game in the style of Watson and Pollack: each individual is a pair of non-negative numbers, two populations A and B play each other, and "skill" is simply how large the numbers are. Four rules for who wins a contest are enough.
Population A. Population B. Up and to the right is more skill.
A's champion wins. B's champion wins. Rows are A's champions, oldest at the top; columns are B's, oldest at the left. Solid blue below the diagonal and solid orange above it means every champion beats all older opponents.
The grid on the right is the "current individual against ancestral opponents" plot that Cliff and Miller proposed in 1995 for telling progress from running in place.
Summing an individual's wins hides which opponents it beat. Pareto co-evolution keeps that information: every test is treated as a separate objective, and learner a dominates learner b if a passes every test b passes and at least one more. The learners that nobody dominates form the Pareto front. Tests are judged in the same spirit, by the distinctions they make between learners, not by how many learners they defeat.
The game below is De Jong and Pollack's compare-on-one. Learners and tests are pairs of numbers. A test examines only its own larger dimension, and a learner passes if it is at least as large there. Each test is drawn as a threshold line: a learner passes a vertical line by being on or to the right of it and a horizontal line by being on or above it. The plot zooms to wherever the populations currently are.
Learner on the front (filled) or dominated (hollow). Test threshold. Select a learner to see what it dominates.
One row per learner, front first. One column per test, ordered by the dimension it examines and then by difficulty. A filled cell is a pass. One row dominates another if its filled cells include all of the other's.
With summed scores the tests crowd onto one dimension, the learners follow, and the other dimension is abandoned: in 30 test runs of 300 generations only 1 kept the weaker dimension above half the stronger one. With Pareto selection of learners and tests kept for the distinctions they add, 16 of 30 did, and the weaker dimension averaged 33 against 5. Pareto selection is not a guarantee. Once no surviving test examines a dimension, that dimension is as invisible as before, which is why archive methods such as IPCA keep tests explicitly.
Van Valen's Red Queen runs to stay in the same place. In co-evolution that looks like a score that barely moves while both sides change underneath it. Camouflage is the textbook case, and Craig Reynolds's 2023 simulation of evolving prey textures against evolving predator vision is its modern form. The model here is much smaller. A patch of habitat is six texture features. Prey carry six numbers of their own. A predator carries an expectation of what habitat looks like and a limited budget of attention across the six features; shown one prey among several habitat patches, it attacks whichever patch looks most out of place.
In units of the habitat's own patch-to-patch variation. A value near 1 means prey are as similar to the habitat as habitat patches are to each other.
Habitat range. Average prey. Predator attention, which always sums to 100%.
A typical run has three phases. Prey start with random textures and predators catch more than 80% of them once they have learned the habitat. Prey then converge on the habitat and the catch rate falls to roughly twice chance. After that the contemporary catch rate hardly moves, and the two comparison lines say why. Predators frozen at the reference generation catch far fewer of today's prey than they caught of their own, so prey have made lasting progress. Today's predators, however, do worse against the reference prey than the reference predators did: they have tuned their attention to the small flaws of current prey and lost some of their general ability. Ten test runs averaged a 34% contemporary catch rate, 74% for current predators against generation-8 prey, and 42% for generation-8 predators against current prey.
Changing the habitat restarts the race. For a few generations the catch rate drops below chance, because the predators' picture of the habitat is out of date and real habitat looks stranger to them than the prey do. Then they relearn it, prey are exposed, and the convergence repeats.
The two intertwined spirals, 194 points to be separated into two classes, were a standard hard case for neural networks and for Koza's genetic programming. In 1996 Juillé and Pollack turned it into a co-evolutionary problem. Programs are expression trees over x and y. Instead of counting correct points, a program is scored in contests against other programs, earning one unit for every point it classifies correctly that its opponent gets wrong. A program that only repeats what its opponents already solve earns nothing. The solutions reported in 1996 were modular: they divide the plane into regions and solve each one.
Shaded regions are the best program's two answers. Dots are the 194 spiral points in their true class color; a ring marks a point the program gets wrong.
The 1990 paper already contains the parts later work had to make precise: two populations with opposed fitness, a rule for who meets whom, and the observation that an adapting opponent supplies both diversity and a cheap, informative test set. Select an entry to read what it added.
Per Kristian Lehre's GECCO 2026 tutorial frames the field as adversarial optimization: both strategy spaces are too large to enumerate, the payoff can only be queried, and the goal is a Nash equilibrium or a maximin solution. Sorting networks against test cases are its standard example of a test-based game, where the payoff is 1 if design x passes test y and 0 otherwise.
A generative adversarial network trains a generator against a discriminator; Goodfellow and colleagues described it in 2014 as a two-player minimax game. Hillis's hosts and parasites have the same structure. In both, one side is the product and the other exists to find its current weaknesses, and in both the adversary is discarded at the end.
| Hillis, 1990 | GAN, 2014 | Lehre's framework, 2026 | |
|---|---|---|---|
| Product | Host: a sorting network | Generator G | Design x in X |
| Adversary | Parasite: 10 to 20 test inputs | Discriminator D | Test y in Y |
| Payoff | Tests failed. Host and parasite scores are complementary | V(D, G), minimized by G and maximized by D | g(x, y), zero-sum |
| Fixed referee | "Is the output sorted?" Always correct | The real data. D must learn the boundary and can be wrong | The payoff oracle |
| What the adversary buys | Tests concentrate on what current hosts get wrong, so few are needed | A learned loss that concentrates on what current samples get wrong | Informative interactions without evaluating all of Y |
| Update | Selection, recombination and mutation over two populations. A host receives a score only | Gradient steps on one pair. G receives a gradient through D | Black-box: payoff queries only |
| Solved when | The host passes every possible test | Generated and real data are indistinguishable, D outputs 1/2 | Maximin or Nash equilibrium |
| Who meets whom | Same cell of a grid | The single current opponent | The interaction scheme: all pairs, k random opponents, or spatial |
| Co-evolution | What happens | GAN training | Remedies on each side |
|---|---|---|---|
| Cycling | The players chase each other around the equilibrium and revisit old states | Oscillation and non-convergence of simultaneous gradient steps | Worst-case scoring, archives. Gradient penalties, instance noise |
| Disengagement | One side wins every encounter, so scores stop distinguishing anything | Vanishing generator gradient when D separates real from fake perfectly | Keep opponents "worth beating" (Rosin and Belew). Non-saturating and Wasserstein losses |
| Overspecialization | Skill against the current opponents only | Mode collapse: G produces the few outputs that fool the current D | Fitness sharing, spatial demes. Several generators or discriminators |
| Forgetting | Traits that beat past opponents are lost once those opponents are gone | D forgets earlier generator behavior, G returns to it | Hall of Fame. Replay of past generated samples |
The correspondence is close enough that the two fields merged in practice. Lipizzaner (2018) places generator and discriminator pairs on a wrap-around grid, trains each with gradient steps against its neighbors, and applies selection between adjacent cells. Its authors cite Hillis, and Lehre's tutorial lists Hillis's sorting networks and spatial GAN training side by side as examples of the same interaction scheme.
Both panels play g(u, v) = (u − 0.6)(v − 0.4), whose only equilibrium is the marked point. On the left are gradient steps, as in GAN training; with a point-mass generator, a linear discriminator and a linear loss this is the Dirac-GAN of Mescheder, Geiger and Nowozin (2018), shifted. On the right is the (1,λ) co-evolutionary algorithm from the tutorial on the Bilinear game, which is the same payoff over bit strings of length 200, with u and v the fractions of one-bits.
With the default settings both orbit the equilibrium without settling: simultaneous gradient steps spiral outward, and average-payoff co-evolution wanders around the point. Each side has a fix that works in this game. A gradient penalty on one player damps the rotation. Scoring by worst case pulls each population toward the strategy that is safe against every opponent present, and the run settles on the equilibrium.