The objection that should win
Take the strongest form of the case against this whole argument. An incident commander running a wildfire is not a termite. She has doctrine, a chain of command, twenty years of fireline judgement and a radio net that lets her tell a crew exactly where to go. Stigmergy was coined to explain how creatures with no plan and no memory build a cathedral out of mud pellets. Importing that model into fire command mistakes a coping mechanism for insects into a design principle for professionals who can simply be told what to do.
Put more sharply: the reason termites coordinate through the mound rather than through messages is that a termite cannot hold a plan in its head. An incident commander can. She has maps, a common operating picture, a briefing at 0600 and a chain of command that resolves disputes in seconds. If coordination failures happen on a fireline — and they do, fatally — the fix is better communication discipline, not a theory borrowed from soil insects. Calling her command structure "stigmergic" flatters an insect metaphor at the expense of an actual profession.
This objection deserves to be taken seriously before it is answered, because it is largely correct about what an incident commander does all day. She commands. Nobody is claiming otherwise.
What Grassé actually found, and what he didn't
Pierre-Paul Grassé watched Bellicositermes workers raise arches that met in mid-air with no worker able to perceive the far pillar. His explanation, published in 1959, was that the structure already built stimulates the next act of building: a soil pellet carrying pheromone is itself the instruction, read by whichever worker passes next. He called it stigmergie — mark, work. The term travelled in the 1980s and 1990s into swarm robotics and ant colony optimisation, where Marco Dorigo turned pheromone evaporation into a dial you could tune for search problems.
What Grassé's finding actually licenses is narrower than "insects are simple, so simple coordination is insect-like." It licenses a claim about where state lives: in the trace, not in any single agent's head. That claim is orthogonal to how clever the agents are. A single sophisticated commander reading a continuously updated fuel-moisture grid is running a fully stigmergic process, because the grid — not her memory, not a radio call — is what carries the state forward from one hour to the next. The mistake, on both sides of this debate, is to treat stigmergy as a thesis about intelligence rather than a thesis about control. It says nothing about whether the reader is a termite or a fire chief. It says something about whether the thing being read is still being written.
What the incident commander is actually reading
An incident command post on an active fire is a stigmergic environment whether anyone names it that. Fuel-moisture sensors report the dryness of duff and canopy at intervals of hours. Satellite thermal passes — polar-orbiting, often twice-daily rather than continuous — mark hotspots as pixels of temperature anomaly. Wind models forecast gust direction and speed for the next six to twelve hours, revised as new observations arrive. Crew positions, tracked by GPS unit or radio check-in, mark where the line actually is rather than where the plan said it would be. None of these four streams is a message to the commander. Each is a deposit on a shared surface: sensor readings on a moisture map, thermal anomalies on a satellite overlay, isobars on a wind chart, dots on a tracking board. The commander's job is to read that surface and act, and the next set of deposits will already reflect what she and her crews did an hour ago.
The characteristic failure of this system is well known on any fireline: an ignition is detected after the wind has already shifted, not before. A thermal pass flags a hotspot; by the time the pass is processed and plotted, the wind model that explained its likely spread has been superseded by an observation the model hadn't ingested yet. The crew that gets sent to hold a flank is holding against yesterday's wind. This is not a communication failure in Grassé's sense — nobody garbled a message. It is a decay failure: the trace being read had already gone stale by the time the read happened, and nothing on the map distinguished the fresh deposit from the evaporated one.
Answering the objection: command doesn't remove the trace layer
Here the strongest objection partly survives and partly collapses. It survives in this form: an incident commander is not doing what a single foraging ant does, and nothing here should be read as reducing incident command to pheromone-following. Command exists precisely because someone has to resolve which trace to trust when two streams disagree — the sensor grid says humidity has risen, the crew on the ground reports the brush is crackling dry. That resolution is judgement, not deposit-reading, and doctrine, training and hierarchy earn their keep doing it.
It collapses on the scaling argument. Direct, planned coordination costs links: every crew, every division, every air asset that needs to know what every other one is doing multiplies communication overhead roughly with the square of the number of participants. A large fire can run twenty or more crews, several aircraft, and mutual-aid engines that were not on the roster at the start of the operational period. Nobody messages all of them individually about every wind update. What actually happens is that the shared operational picture — the map, the moisture overlay, the tracking board — gets updated, and each unit reads the version relevant to it. That is the trace layer doing the coordination work that direct messaging cannot afford at that scale, and it is present whether or not anyone calls it stigmergic. Sophistication in the commander does not remove the need for a shared, current surface; it raises the cost of that surface being wrong.
Answering the second objection: decay is not visible from inside a snapshot
A related objection concedes the mechanism but shrinks its importance: most of what a fire behaviour analyst relies on is stable. Slope, aspect, historical fuel type, the layout of ridgelines — these do not change during an incident, and a fixed terrain database serves them perfectly well. Continuous intake, on this view, only buys freshness on the thin, volatile margin — moisture, wind, spot ignitions — that a minority of the analysis actually depends on.
The terrain doesn't move. Half of what we're modelling is geometry, not weather. You don't need live intake to know where the ridge is.
That is true, and it should be conceded without qualification. Slope and fuel-type maps are legitimately closer to a frozen corpus than to a live stream, and nobody should be running continuous satellite reprocessing to find out where a ridgeline sits. But the objection proves less than it appears to, because a snapshot has no way of marking which of its contents are ridgelines and which are wind. Both look like settled fact inside a static database. The only way to learn that fuel moisture has a half-life of hours, that a spot-fire report has a half-life of minutes, and that a slope map has a half-life of decades, is to watch all three change, or fail to, over time. A system built on continuous intake with provenance attached can price each belief by its own decay rate — treat the terrain layer as durable, the moisture layer as needing a fresh pull before every operational period, the wind model as needing revision inside the burn period itself. A frozen intake has to apply one freshness assumption to everything it holds, because it has no mechanism for discovering that assumption is wrong for any particular layer.
The claim that survives
What holds, after both objections have taken their real bite, is narrower than "stigmergy explains incident command" and narrower still than "insects and fire chiefs coordinate the same way." What holds is this: wherever coordination depends on reading a shared surface rather than exchanging direct messages — and a wildfire, above a certain size, has no alternative — the freshness of that surface is not a nice-to-have. It is the condition under which the read means anything at all. An incident commander with excellent judgement, reading a moisture grid twelve hours stale and a thermal pass that missed the last wind shift, will make an excellent decision against a fire that no longer exists in that shape. The failure is not one of command. It is one of intake, and no amount of doctrine repairs a trace that stopped updating before the decision was made on it.