> ## Content Index > Fetch the complete content index at: https://pours.thequantumdistillery.com/llms.txt > Use this file to discover other available public pages before exploring further. # Fifteenth Pour: The Swarm and the Ledger - URL: https://pours.thequantumdistillery.com/fifteenth-pour-the-swarm-and-the/ - Published: 2026-08-13T16:56:32.000Z - Updated: 2026-10-09T02:58:25.000Z - Description: Coordination without a coordinator — how termites, biofilms, bones, and AI agents think through the world instead of inside their heads - Author: Thad Connelly - Tags: Newsletter, #thequantumdistillery, #quantumbiology, #georgechruch, #gengyveusa, #hubermanlab, #sabinehossenfelder, #Migrated-1791514643696, #Import 2026-10-09 02:58 The bar went dark in October. No dramatic story. The still didn’t explode; I just had too many mash tuns going at once, and a couple of them were on fire. But barrels don’t care whether the lights are on. They kept aging, and one idea in particular would not stop knocking on the inside of its cask. Tonight’s pour is that idea. It may be the oldest trick in the biosphere, and I watched the newest minds on Earth rediscover it in my own office last week. Pull up a stool. This one drinks best slowly. [Subscribe](#/portal/signup) ## The termite question In 1959, the French zoologist Pierre-Paul Grassé was staring at a problem that should keep more people up at night than it does: termites build cathedrals. Their mounds run meters tall, with ventilation shafts, fungal gardens, humidity control, load-bearing pillars, spiral galleries. Architecture, by any honest definition. Now the uncomfortable part. No termite has a blueprint. No termite has even *seen* the mound — they’re blind. There is no foreman, no meeting, no plan. The queen is an egg machine, not an architect. Each worker carries a brain about the size of a poppy seed, wired with a handful of reflexes. Ask an individual termite what it’s building and you’d get nothing. There’s nobody home to ask. Grassé’s answer was so simple it took decades for its size to register. A termite picks up a pellet of soil, mixes it with saliva and pheromone, and drops it. That’s the whole behavioral repertoire. But the deposited pellet *changes the environment*, and other termites are wired to respond to that change by depositing their own pellets on top of it. The growing pillar becomes the instruction to grow the pillar. When two pillars rise near each other, the geometry of the pheromone field tips the workers toward building inward, and an arch closes overhead. An arch no termite conceived. Grassé coined a word for this: **stigmergy**, from the Greek *stigma* (mark) and *ergon* (work). The mark spurs the work. The work leaves a mark. Nobody talks to anybody; everybody talks to the wall, and the wall talks back. Read that again, because it’s the whole pour in one line: **coordination does not require communication between agents. It requires a medium that remembers.** ## The colony as a computer made of evaporation Ants sharpen the principle to a knife edge. In the classic double-bridge experiments — run by the Brussels group of Deneubourg, Goss, and colleagues at the end of the 1980s — a colony is offered two paths to the same food source, one short, one long. Foragers initially wander both. But ants returning on the short path complete round trips faster, so pheromone accumulates on it faster; more pheromone recruits more ants, which lay more pheromone. Within an hour the colony has “chosen” the short path. It has solved an optimization problem, and no ant ever compared the two routes. The comparison happened *in the chemistry of the trail*. The quiet hero in that experiment is evaporation. If pheromone were permanent, the first trail laid would tyrannize the colony forever — early noise locked in as dogma. Because the medium *forgets* at a steady rate, only trails that are continually re-earned persist. Forgetting isn’t a defect of the system. Forgetting is what keeps the ledger honest. Memory in the medium. Forgetting as quality control. Amplification of whatever works. That’s the whole architecture, and now that you’ve seen it, we’re going to find it everywhere — including inside you. ## The city on your teeth Regulars of this bar know I’ve spent a career elbow-deep in the clinical consequences of one particular stigmergic construction, so allow me the home-turf example. A biofilm — dental plaque being the stuff you floss at — is not a pile of bacteria. It’s a built environment. The residents secrete an extracellular matrix of polysaccharides, proteins, and DNA, and that matrix is the city: it channels nutrients, traps enzymes so digestion happens *outside* the cells communally, buffers pH into distinct neighborhoods, and shelters everyone from antibiotics and immune patrols at doses that would kill any bacterium caught alone in the open. Microbiologists who study the matrix have called it the house of the biofilm cells, and the house has opinions: what one generation of bacteria secretes constrains where the next generation can live, what it eats, and who it becomes. The organisms coordinate through the architecture, not through anything resembling conversation. Same trick as the termites, run in slime at micron scale. And when that matrix mineralizes into calculus — tartar — the ledger literally fossilizes. Archaeologists now read thousand-year-old dietary and microbial history out of the calculus on medieval teeth. The trail outlived the ants by a millennium. (Where that particular city’s exports end up once they leave the mouth is a story I’m building a whole library about elsewhere. The short version: the traffic doesn’t stop at the gums. But that’s another pour.) ## Your skeleton is a slow argument with gravity Here’s the example that made me put down my glass when I first saw it whole. Bone remodels constantly — osteoclasts demolish, osteoblasts rebuild, your entire skeleton turning over roughly every decade. For years the coupling between the two was a puzzle: how does the construction crew know where the demolition crew has been? They don’t overlap much in time. There’s no foreman here either. The answer, worked out in detail by Tang and colleagues in 2009: the message is buried in the material. Bone matrix is salted with growth factor — TGF-β1 — deposited there when the bone was originally built. When an osteoclast chews through matrix, it *releases* that stored signal, and the liberated TGF-β1 recruits marrow stem cells to the freshly dug trench, where they become the osteoblasts that refill it. The demolition is the mail. The wall talks back. Which means Wolff’s law — bone growing denser exactly along its lines of mechanical stress, the reason a tennis player’s serving arm carries measurably more bone — is colony computation in mineral. Millions of cells, none of which knows what a femur is, collectively solve a structural engineering problem by reading and writing marks in a shared medium, with mechanical strain deciding which trails get re-earned. Your skeleton is a pheromone trail that takes decades to evaporate. ## The brain keeps marginalia in its walls You knew I’d end up back in the skull. The astrocyte pours (Seven through Ten, for the new arrivals) were one long argument that the neuron-centric picture of the brain is too small — that the glia running the chemistry between neurons are part of the computation. Tonight I want to widen the lens one more click, past the cells entirely, into the space *between* them. That space isn’t empty. It’s packed with extracellular matrix, and around certain neurons the matrix condenses into lattice-like sheaths called perineuronal nets. Dissolve those nets in an adult brain and juvenile plasticity comes flooding back — erasable fear memories, re-openable critical periods. The nets are doing something that looks suspiciously like *stabilizing* what the network has learned. Roger Tsien — the chemist who won the Nobel for green fluorescent protein — spent his late years pushing a heretical hypothesis: that very-long-term memories may be stored not in the synapses themselves, which turn over their proteins in days to weeks, but in the *pattern of holes* in the perineuronal net — the matrix scaffold deciding which synaptic arrangements are allowed to persist. Some in the field now speak of the “tetrapartite synapse”: presynaptic neuron, postsynaptic neuron, astrocyte, and the matrix. It remains a hypothesis, and I’ll pour the full version another night. But sit with the shape of it: even the organ we call the seat of intelligence may keep part of its ledger in the walls between the cells. The termites never left. They just moved indoors. ## One more, for the physicists at the end of the bar This is the Quantum Distillery, and somebody down at that end has been waiting all night for me to say the word. So: a nightcap. I’ll label the pour honestly — the first half is textbook, the second half is conjecture, and the house knows the difference. The textbook half. In the light-harvesting machinery of photosynthesis, energy transport doesn’t succeed *despite* the molecular racket around it. A run of very good theory in the late 2000s showed that transport efficiency peaks at an intermediate level of environmental jostling: too quiet, and the energy gets stuck in dead ends; too loud, and it dissolves into a random walk; in between sits a sweet spot where the environment’s own interference helps the excitation find the exit. The field calls it environment-assisted quantum transport. I call it the bartender’s rule: a little disorder keeps everything moving. Now the conjecture, poured neat. Look at the shape of that law — performance against noise, an inverted U — and climb the ladder with it. Stochastic resonance: your neurons detect faint signals best at intermediate noise. Hormesis: your physiology gets tougher at intermediate stress. And tonight’s quiet hero, the trail: colony intelligence peaks at intermediate evaporation — too little forgetting breeds dogma, too much breeds amnesia. Same curve, four scales, from a pigment molecule to a termite mound. I cannot prove it’s one theorem wearing four costumes rather than four coincidences wearing one shape. Nobody can, yet. But I’ve stopped being able to un-see it: nature doesn’t shield its machinery from noise. Nature tunes the noise. The tuning is the machinery. One more sip, because I promised you the wall, and the wall has a basement of its own. There’s a serious line of decoherence physics — Zurek and colleagues call it quantum Darwinism, and I’m pouring it as framework, not settled fact — that says the classical world works like this: a quantum system never shows you itself. What it does is imprint redundant copies of its state onto the surrounding environment — every scattered photon, every jostled air molecule carries a partial record — and anything that later “observes” the system is actually reading those copies. The environment is a ledger. Many overlapping ledgers, in fact, and that redundancy is the whole trick: it’s why you and I agree on where this glass is sitting. We never measured the glass. We read the same graffiti. Follow that one step further and it stops being trivia: objectivity itself — the shared, classical, agreed-upon world — emerges the way a termite mound does, from redundant marks written into a common medium, with the durable marks selected and the fragile ones erased. The wall doesn’t just talk back at the bottom of the stack. At the bottom of the stack, the wall is all there is. Wheeler gave that suspicion its bar name half a century ago — *it from bit* — and it is not lost on me that the man who compiled the hymnal was Zurek’s teacher. And so the bar keeps its license: nothing else in tonight’s pour depends on either of those basement sips. The termites, the biofilm, the bone, the nets — every one of them stands on boring classical legs if the basement floods. That’s the house rule with quantum biology: admire it, pour it, never lean on it. ## The newest minds find the oldest trick Now the part I watched happen at my own desk. I’ve been running an experiment with AI agents — multiple instances working my problems, and every session wakes up with no memory of the last one. Amnesiac by design. No shared brain, no channel between them. The frontier-lab solution to this is to make the minds bigger: more context, more memory, more model. The termite solution is different, and I tried the termite solution: give them a shared folder of append-only notes. Each session reads the trail on arrival, works, and deposits a mark before it dissolves — what was decided, what failed, what’s still open. Within an hour of laying the first trail, a different agent — one I hadn’t briefed at all — walked it correctly and deposited its own mark for the next. I actually sat back in my chair. No agent has the whole picture. The *folder* has the whole picture. Coordination without communication; memory in the medium; the colony smarter than the ant. Sixty-seven years after Grassé named the trick and maybe a billion years after evolution shipped it, the most advanced artifacts our species has built converged on the termite’s move — because the constraint is the same. When individual minds are bounded and mortal, you stop trying to build a bigger mind and you start building a medium that remembers. ## The distillate Every pour ends with what’s left in the glass after the water burns off. Tonight, three drops. **Intelligence doesn’t require a head.** It requires a medium that holds marks, agents with a rule for leaving them, and time. Brains are one implementation, not the definition. The mound thinks. The biofilm thinks. The skeleton thinks, slowly, in mineral. **Forgetting is load-bearing.** Every one of these systems works because the medium decays — pheromone evaporates, matrix remodels, trails must be re-earned. Permanent memory isn’t wisdom; it’s the first error, enshrined. **And a newsletter is a stigmergic object too.** Each pour is a mark that makes the next one likelier — for the writer as much as the reader. The bar went quiet for nine months, but the marks were still on the wood, and they’re what pulled me back through the door. The still is lit. Next pour in two weeks. --- ### Tasting notes (for those who drink neat) Grassé, P.-P. (1959). *La reconstruction du nid et les coordinations interindividuelles chez Bellicositermes natalensis et Cubitermes sp.* Insectes Sociaux 6, 41–80 — the paper that coined stigmergy. · Theraulaz, G. & Bonabeau, E. (1999). *A brief history of stigmergy.* Artificial Life 5, 97–116\. · Goss, S., Aron, S., Deneubourg, J.-L. & Pasteels, J.M. (1989). *Self-organized shortcuts in the Argentine ant.* Naturwissenschaften 76, 579–581\. · Deneubourg, J.-L. et al. (1990). *The self-organizing exploratory pattern of the Argentine ant.* Journal of Insect Behavior 3, 159–168\. · Flemming, H.-C. & Wingender, J. (2010). *The biofilm matrix.* Nature Reviews Microbiology 8, 623–633\. · Tang, Y. et al. (2009). *TGF-β1-induced migration of bone mesenchymal stem cells couples bone resorption with formation.* Nature Medicine 15, 757–765\. · Tsien, R. Y. (2013). *Very long-term memories may be stored in the pattern of holes in the perineuronal net.* PNAS 110(30), 12456–12461\. · Plenio, M.B. & Huelga, S.F. (2008). *Dephasing-assisted transport: quantum networks and biomolecules.* New Journal of Physics 10, 113019\. · Rebentrost, P., Mohseni, M., Kassal, I., Lloyd, S. & Aspuru-Guzik, A. (2009). *Environment-assisted quantum transport.* New Journal of Physics 11, 033003\. · Wiesenfeld, K. & Moss, F. (1995). *Stochastic resonance and the benefits of noise: from ice ages to crayfish and SQUIDs.* Nature 373, 33–36\. · Zurek, W.H. (2009). *Quantum Darwinism.* Nature Physics 5, 181–188\. · Wheeler, J.A. (1990). *Information, physics, quantum: the search for links.* In Complexity, Entropy and the Physics of Information (ed. W.H. Zurek). [Subscribe](#/portal/signup)