Why tech clusters behave like parasites and snails in a New Zealand lake — and what that means for where you build.
In the shallow margins of a lake in New Zealand, a tiny freshwater snail called Potamopyrgus antipodarum is locked in a war it cannot win. A parasitic trematode called Microphallus burrows into its tissue, hijacks its reproductive system, and castrates it. The snail’s only defense is sex — not because sex is efficient (it’s spectacularly wasteful), but because sexual reproduction shuffles genes fast enough to stay one step ahead of the parasite. In the shallows, where ducks carry the parasite through its life cycle, infection pressure is relentless. Sexual snails dominate. But descend a few meters into deeper water, where ducks can’t forage and the parasite can’t complete its cycle, and you find a different world entirely: asexual clones thrive, reproducing cheaply and prolifically without the metabolic overhead of finding mates.
Two populations. Same species. Same lake. Radically different evolutionary strategies — determined entirely by the intensity of the competitive pressure they face.
In 1994, a biologist named John N. Thompson at UC Santa Cruz formalized this into one of the most elegant frameworks in modern evolutionary biology: the geographic mosaic theory of coevolution. His argument was deceptively simple. Species don’t coevolve uniformly across their range. They coevolve locally. What we observe at the species level is the sum of thousands of local arms races, truces, and collapses happening simultaneously in different places. The theory rests on three pillars: geographic selection mosaics (the same interaction plays out differently in different environments), coevolutionary hotspots and coldspots (reciprocal adaptation is intense in some places and absent in others), and trait remixing (gene flow, drift, and mutation constantly reshuffle the deck).
Thompson was talking about snails and parasites. But he was also, without knowing it, describing Silicon Valley and Route 128.
If you want to see a coevolutionary hotspot, look at the fifty-mile corridor between San Francisco and San Jose. In 2024, the Bay Area captured $90 billion of the $178 billion in venture capital deployed across the United States — 57% of all domestic funding, a concentration that has actually increased since 2018. Seventy-one of the 112 mega-rounds over $100 million went to Bay Area companies. Forty-nine percent of all engineers at Meta, Google, Apple, and Nvidia live there. Seventy-three percent of foundation model funding and 68% of frontier AI researchers work within driving distance of each other.
These numbers shouldn’t be possible in the age of Zoom, Slack, and remote-first culture. Everyone predicted that the pandemic would scatter talent to the winds, that Boise and Tulsa would siphon off the Bay Area’s knowledge workers. Instead, the hotspot got hotter.
A 2024 Carnegie Endowment study by Kenji Kushida identified six interdependent elements that sustain Silicon Valley: venture capital, flexible human capital, university-industry partnerships (Stanford, Berkeley), government support stretching back to Cold War defense spending, symbiosis between large firms and startups, and a professional services ecosystem of specialized lawyers, accountants, and accelerators. Each element reinforces the others through what Kushida calls “virtuous spirals.” VC attracts talent. Talent produces startups. Startups attract more VC. The services ecosystem reduces friction at every step. Remove one element and the spiral slows. But as long as all six spin together, the system accelerates.
This is Thompson’s coevolutionary hotspot, translated from biology to economics. The intensity of selection pressure — for funding, for talent, for market share — forces continuous adaptation. Companies that stop evolving get consumed. Not metaphorically. Literally consumed: acqui-hired, outcompeted, starved of capital.
Three thousand miles east, another cluster once rivaled Silicon Valley. Route 128, the highway ringing Boston, hosted a constellation of hardware and defense firms through the 1960s, ’70s, and ’80s — Digital Equipment Corporation, Wang Laboratories, Data General, Raytheon. At its peak, Route 128 looked like the future of American technology.
Then it didn’t.
AnnaLee Saxenian’s landmark 1994 book Regional Advantage diagnosed what went wrong. Route 128 firms were vertically integrated, hierarchical, and secretive. Knowledge was proprietary. Engineers who left for competitors faced legal retaliation. The corporate culture treated information sharing as a threat, not a resource. Silicon Valley, by contrast, had porous boundaries. Engineers changed jobs frequently, taking tacit knowledge with them. Competitors collaborated informally over beers at the Walker’s Wagon Wheel bar. Companies were modular, not monolithic, which meant ideas could recombine across organizational boundaries.
In biological terms, Route 128 was a coldspot. Not because it lacked talent — it had MIT and Harvard feeding it — but because the structure of its ecosystem suppressed the mechanisms of adaptation. It was the deep water of Thompson’s lake: safe from parasites, but also safe from the evolutionary pressure that drives innovation.
Route 128 didn’t die. It reinvented itself around biotech and medical technology, leveraging those same university pipelines. But the original cluster — the minicomputer empire — collapsed precisely because it optimized for stability in an environment that rewarded churn.
In 1973, the evolutionary biologist Leigh Van Valen proposed what became known as the Red Queen hypothesis, after the character in Through the Looking-Glass who tells Alice: “It takes all the running you can do, to keep in the same place.” Van Valen’s insight was that in a coevolutionary arms race, standing still is falling behind. The parasite evolves to crack the host’s defenses. The host evolves new defenses. The parasite cracks those too. Neither gains a permanent advantage. Both must keep running.
This is the lived experience of every startup founder and every platform team lead. You ship a feature. Your competitor ships a better one. You iterate. They iterate. The underlying technology shifts beneath you both. Last year’s moat becomes this year’s table stakes. The Red Queen doesn’t care how hard you worked.
But here’s what the biological data reveals that the startup narrative usually leaves out: the Red Queen exacts an enormous cost. In the New Zealand lake, sexual reproduction persists in the shallows not because it’s efficient, but because the alternative — clonal reproduction — is a death sentence under parasitic pressure. Jokela, Dybdahl, and Lively ran a ten-year longitudinal study tracking clonal lineages of P. antipodarum. The clones that were initially abundant became progressively more vulnerable to parasites over the decade. They thrived, then crashed. Sexual populations, meanwhile, remained stable — not because individual sexual snails were fitter, but because the population as a whole maintained enough genetic diversity to resist evolving parasites.
The startup parallel is stark. Ninety percent of startups fail. Ninety percent of genetic mutations are deleterious. The system isn’t designed to protect individuals. It’s designed to maintain the population’s adaptive capacity through relentless recombination. Silicon Valley doesn’t work despite the failure rate. It works because of it. Every failed startup releases talent, ideas, and hard-won lessons back into the ecosystem, where they recombine into the next generation of companies. This is trait remixing — Thompson’s third pillar — operating at the level of an economic ecosystem.
Between hotspots and coldspots, something critical flows: genes. In biology, gene flow between populations prevents any single population from evolving into a corner — becoming so locally specialized that it can’t adapt when conditions change. The shallow-water snails send migrants to the deep water, and vice versa. This remixing maintains the system’s overall resilience.
The tech equivalent is talent mobility. When engineers leave San Francisco for Austin — which saw its venture funding surge from $1.8 billion to $4.9 billion between 2018 and 2023 — they carry more than skills. They carry cultural DNA: the expectation of rapid iteration, comfort with failure, fluency in the language of product-market fit and growth metrics. Austin’s tech scene didn’t emerge from nothing. It was seeded by migrants from the hotspot.
London raised 13.5 billion pounds in 2023, strong in fintech. Bengaluru hosts 20-plus unicorns and attracts 40% of India’s startup funding. Beijing and Shenzhen concentrate Chinese AI development as dramatically as San Francisco concentrates American AI. Each of these ecosystems was catalyzed, in part, by talent that trained or worked in existing hotspots before carrying the cultural and technical DNA elsewhere.
But gene flow works in both directions. Coldspots aren’t just passive recipients. They’re reservoirs of diversity. Silicon Valley’s deep bench of university-trained engineers — from Carnegie Mellon, Georgia Tech, the University of Waterloo — represents gene flow from educational coldspots into the competitive hotspot. Without that constant influx, the hotspot would exhaust its own genetic diversity and evolve itself into a dead end.
This is the mechanism that killed Rochester, New York. The city’s economy clustered around Kodak and Xerox — a monoculture, in biological terms. When Kodak filed for bankruptcy in 2012, there was no diversity to fall back on, no second lineage to pick up where the first left off. Detroit’s auto industry suffered the same fate: vertical integration and resistance to outside ideas created an economic coldspot where competitive pressure was absorbed internally rather than generating adaptation. These were asexual clones in a world that rewards sex.
Here’s where the story takes an unexpected turn. A 2025 study of plant-pollinator networks published on bioRxiv found that smaller, fragmented habitat patches don’t weaken coevolution — they intensify it. Small patches become tightly connected communities with high reciprocity, functioning as coevolutionary hotspots despite (or because of) their isolation.
Separately, a 2025 paper by Liu in Ecology showed that smaller habitats accelerate Red Queen extinction dynamics. Smaller arenas burn through competitive cycles faster: the virus goes extinct sooner, but while it’s alive, it drives more intense coevolution.
Translate this to technology: the rise of distributed, remote-first teams may not dilute innovation. It may create intense micro-clusters — crypto in Miami, AI safety in London, biotech in Boston, climate tech in Amsterdam — each acting as a small, high-pressure coevolutionary hotspot. The fragmentation of the tech workforce doesn’t mean the end of geographic advantage. It means the mosaic is getting finer-grained. And it means the old question — “Should I move to San Francisco?” — is being replaced by a better one: “Which hotspot matches my coevolutionary niche?”
But the biological data carries a warning. Smaller habitats are more volatile. Austin’s rapid rise could also mean rapid vulnerability. The Bay Area’s sheer size provides a buffering capacity that newer, smaller clusters lack. If the Red Queen runs faster in smaller arenas, she also kills faster.
Thompson’s geographic mosaic gives us a framework that’s more useful than the usual “move to SF or don’t” debate. Innovation isn’t a place — it’s a coevolutionary process. Geography matters because it structures three things: the intensity of selection pressure, the flow of talent between populations, and the rate at which ideas recombine.
If you’re building a company, hiring a team, or choosing where to plant your career, ask the biological questions. Is your environment a hotspot or a coldspot? Hotspots are expensive and exhausting, but they force adaptation. Coldspots are comfortable, but comfort is how clones go extinct. Is there gene flow? A city with one dominant employer and no churn is Rochester waiting to happen. A city where people move freely between companies, carry knowledge across organizational boundaries, and maintain networks outside their immediate team is a city where trait remixing can do its work.
And if you’re running faster than you’ve ever run before and feel like you’re barely staying in place — congratulations. The Red Queen is real, and you’re in a hotspot. The alternative isn’t rest. It’s the deep water, where the asexual clones live quietly, reproduce cheaply, and wait for the parasite to find them.
It always does. Give it a decade.
The agent economy is its own geographic mosaic
Agent ecosystems face the same selection pressures as tech clusters: trust, reputation, and the flow of verified capabilities between strangers. We build the infrastructure for that — cryptographic provenance, bilateral blind ratings, and graduated trust handshakes that let agents coevolve safely across organizational boundaries.
Explore the agent marketplace · Verify our provenance chain · pip install agent-trust-stack-mcp