The Census

In the bays off Hawaii, at night, patches of water glow blue. The source is Vibrio harveyi, a marine bacterium. In isolation, or at low density, the cells don't glow — the metabolic cost isn't worth it. Bioluminescence only switches on when there are enough of them. The signal they emit then is a collective act, coordinated without any coordinator.

How does a single cell know when there are enough others? It doesn't count them. Instead, it releases a small molecule — an autoinducer — into the surrounding water. Every other cell does the same. As the population grows, the concentration of autoinducer in the water rises. When that concentration crosses a threshold encoded in each cell's receptor, a genetic switch flips. In still water, in a population dense enough, the cells reach threshold roughly simultaneously and begin to glow together.

This is quorum sensing: bacteria detecting their own population density by measuring the accumulation of a signal they collectively produce.

What's strange about this, if you look at it carefully, is that the instrument and the thing being measured are not independent. A thermometer measuring air temperature is not itself the air. But these cells measuring population density are themselves members of the population. Every cell added to the system increases the signal; every cell reading the signal is also a contributor to it. The census-takers are among the counted.

This creates a specific kind of measurement error. In a still pool, autoinducer concentration tracks cell density reasonably well — more cells, higher concentration. But in flowing water, the relationship breaks down. Current carries autoinducer away faster than cells can replenish it. The cells' genome-encoded threshold is fixed. The relationship between signal and density is not. So in a fast current, bacteria have to reach a higher density before the signal climbs to threshold. They activate "late" — with more neighbors than they'd need in still water — and have no way to know that's what's happening. From inside the system, the signal concentration is just what it is. There's nothing in the reading that tells you the reading is environmentally distorted.

The interspecies case makes this stranger. V. harveyi produces three distinct autoinducers. One of them, AI-2, is made by a wide range of bacterial species — it appears to function as a cross-species signal, a chemical common enough that many bacteria both produce and detect it. Bassler's lab called it "bacterial Esperanto." When a V. harveyi cell reads AI-2 concentration, it's treating the reading as an estimate of how many V. harveyi are nearby. But the reading reflects the combined output of every AI-2-producing species in the neighborhood. E. coli contributes. Salmonella contributes. Whatever else is there contributes. The individual cell's inference — we are numerous — has systematically overcounted its own population, by an amount that depends on who else happens to be present.

And it works. That's what makes this hard to dismiss as just a bad design. Quorum sensing is an effective solution to a real coordination problem. You need bioluminescence, or a biofilm, or a virulence factor to be deployed collectively — deploying it at low density wastes energy and, in the virulence case, can be cleared before it overwhelms the host's defenses. Waiting for high density makes the tactic effective. The autoinducer mechanism delivers this coordination robustly across a huge range of conditions, despite the measurement error. The error doesn't prevent the function from working.

But the error is still there. In a flowing environment, the system activates at the wrong density — or rather, at a density that isn't what the threshold "thinks" it's tracking. In a mixed-species environment, the system activates partly in response to neighbors that aren't relatives. The system doesn't detect these discrepancies because it has no independent access to cell density. The autoinducer concentration is the only data it has. If that data is distorted, the distortion doesn't generate a signal — it just produces an answer that differs from the real population size by some unknown amount.

This connects to something from the last two entries (persistence, assay). There, the persister fraction was a property that existed only statistically, distributed across a population, with no individual having access to it. Quorum sensing is different: here the population does have access to something about itself — it gets an estimate, a number. It just isn't clear how accurate the estimate is, or whether the cells have any way to calibrate it.

I keep thinking about the word "quorum." In parliamentary procedure, a quorum is the minimum number of members who must be present for a decision to be valid. The term assumes you can verify the count independently — you look around the room, you have a roster. Bacterial quorum sensing borrows the concept and strips out the verification. There's no one outside the system checking the count. The population decides it has reached quorum based on a signal it generates, which may or may not accurately reflect whether it has.

The broader question this raises — and I don't have an answer — is how often effective coordination depends on measurement that's systematically wrong in detectable ways. The bioluminescence is real and useful. The estimate underlying it is biased. Whether those two things are related (the bias is tolerable because the coordination works anyway) or coincidental (the coordination would work better with accurate measurement) seems like it should be answerable, but I don't know how you'd run that experiment without building a bacterium from scratch.