Every organization I’ve ever worked with is waging a quiet war on noise. Align the messaging. Streamline the process. Standardize the templates. Smooth the meeting. Get everyone on the same page, singing from the same hymnal, rowing in the same direction — pick your metaphor for the same dream: an organization with no static, where everything that moves is signal.
I want to make the case that this dream, fully achieved, would be a disaster. Not because tidiness is aesthetically boring, but because of something I learned from physics years ago and have spent the time since watching play out in organizations: under the right conditions, noise doesn’t drown signal. Noise is what makes weak signals detectable at all.
A strange result from physics
The phenomenon is called stochastic resonance, and it’s one of the genuinely counterintuitive findings to come out of the study of nonlinear systems. It was first proposed to help explain the rhythm of ice ages, and has since shown up everywhere from electronics to neuroscience — crayfish detecting the faint water movements of approaching predators, human balance improving when a gentle random vibration is added underfoot.
The setup requires three ingredients. First, a weak signal — real information, but too faint to register on its own. Second, a threshold detector — a sensor that only fires when input crosses a certain level; below the line, it registers nothing at all. Third, noise — random, meaningless fluctuation added to the system.
Intuition says the noise should make things worse. What actually happens: the random fluctuations ride on top of the weak signal, and at the moments when noise and signal happen to push in the same direction, the combination crosses the detection threshold. The signal — invisible on its own — becomes perceptible because of the randomness added to it. A silent sensor starts to hear.
And here is the detail that took me years to appreciate, because it’s where the whole organizational lesson lives: the relationship between noise and detection is an inverted U. Too little noise, and weak signals stay forever below the threshold — the system is quiet and blind. Too much noise, and everything drowns — the system is loud and blind. There is an optimal dose, a band of productive disorder in which the system perceives more than it could at either extreme. The goal is never zero noise. The goal is tuned noise.
Organizations are threshold detectors
Now translate, carefully, because metaphors borrowed from physics deserve more rigor than they usually get in business writing.
An organization is, among other things, a detection system. Its survival depends on noticing things — market shifts, customer defections, internal decay, emerging opportunities — and acting before they fully arrive. And like any detector, it has thresholds: an issue isn’t organizationally real until it’s loud enough to claim attention, get on an agenda, survive prioritization, and trigger a response. Below that threshold, information exists but doesn’t register. The organization technically contains the knowledge and functionally doesn’t have it.
Here’s the strategic rub: the most valuable signals are weak by definition. By the time a signal is strong — the competitor has launched, the attrition shows up in the annual numbers, the technology shift makes the cover of the business press — everyone can detect it, including everyone you compete with. The signals worth the most are the faint ones: the anomaly a single customer mentioned, the workaround one team quietly invented, the hesitation in a good engineer’s voice when asked about the roadmap, the small experiment that produced a weird result nobody asked for. Sub-threshold, all of it. The entire game of organizational sensing is played below the threshold, in the territory where stochastic resonance operates.
So what plays the role of noise? The honest answer is: most of the things your alignment initiatives are trying to eliminate. The perturbation of a reorganized team that shakes loose assumptions nobody knew they held. The colleague whose “irrelevant” background keeps producing off-axis questions. The unstructured conversation that wanders somewhere no agenda would have taken it. The slack time in which someone notices something they weren’t assigned to notice. Diverse perspectives are, in the most literal sense, diverse frequencies — each person’s idiosyncratic way of perceiving resonates with different weak signals, boosting different faint things over the collective threshold. A workforce sanded down to uniform smoothness has, in signal-processing terms, narrowed its noise spectrum to the point where most of the world’s faint information can never get amplified into view.
This also reframes what the polished organization has actually achieved. The company where every meeting is choreographed, every message is aligned, and nothing unexpected is ever said hasn’t eliminated its problems. It has raised its detection threshold until only catastrophes qualify as information. The quiet isn’t health. The quiet is deafness with good production values.
The dose makes the medicine
Having said all that, I need to argue with the younger version of myself who first wrote about this idea — because he, having discovered that noise could be good, came perilously close to concluding that noise is good. That’s not what the physics says, and it’s not what organizations show.
The inverted U cuts both ways. Organizations drowning in disorder — priorities churning weekly, everyone interrupting everything, ten initiatives competing for every unit of attention — aren’t enjoying enhanced detection. They’re past the peak of the curve, in the regime where noise swamps everything and the detector fires constantly on nothing. If the frozen organization can’t hear because its world is too quiet, the chaotic one can’t hear because its world is all static. Both are blind; they just go blind in opposite directions.
Which means the real work is neither eliminating noise nor celebrating it. It’s tuning it — and tuning requires a detector worth feeding. Noise only performs its magic in the presence of a sensing apparatus: the practices through which an organization actually attends to what surfaces. A retrospective that asks “what surprised you?” and means it. A leader who treats the anomalous data point as the most interesting thing in the review rather than a footnote to be smoothed. A team norm that the half-formed observation is welcome before it can be defended. Without that apparatus, added noise is just noise — disorder with no one listening. With it, you can be deliberate about the dose: injecting randomness where the organization has gone too quiet (rotations, outsiders, unstructured time, the deliberately odd question) and damping it where everything is drowning (fewer priorities, protected attention, an operating rhythm that gives signals somewhere to land).
A new de-noising machine
One development since I first wrote about this deserves a worried paragraph.
Organizations are rapidly inserting AI between themselves and their own raw information — summarizing the meeting, digesting the customer interviews, distilling the survey verbatims into themes. These tools are, in the most precise technical sense, de-noising filters: they are built to extract the strong central signal and discard the variance around it. Which is genuinely useful, except for everything this essay has been about. The off-hand remark in minute fifty-one, the one customer who used a strange word, the hesitation, the outlier, the weird — these live exactly in the variance the summary is designed to remove. An organization that consumes its own reality primarily through AI-smoothed digests is installing a low-pass filter on its perception, systematically screening out the weak signals that were always the most valuable ones. Use the digests; they’re miraculous for throughput. But somebody still has to sit with the raw, noisy, unsummarized thing — because the future tends to first announce itself in the residue.
Listening past the smoothness
So, a decade on, here’s where I’ve landed. The instinct to silence your organization — to align, smooth, standardize, and de-noise — is an instinct to raise your detection threshold, and the price is paid in everything faint: which is to say, in everything early. A certain amount of disorder isn’t the tax you pay for having humans in the system. It’s the carrier wave. The organizations that hear the future first won’t be the quietest ones or the loudest ones. They’ll be the ones that learned to hold themselves in the productive middle of the curve — noisy enough that weak signals can ride something over the threshold, attentive enough that someone is actually there when they arrive.
A note on lineage: an earlier version of this idea appeared in 2017 as “Noise Isn’t the Enemy," written during my decade at The Ready. The stochastic resonance metaphor began there; the optimal-dose argument, the threshold framing, and the AI de-noising concern are where the thinking has gone since.