communicoupling · concept 25 of 26
In the natural sciences the thermometer does not change the temperature. In social life the forecast changes the outcome: the theory of the market becomes a force in the market, the metric becomes a target, the model of the crowd is used by the crowd. When the observers are inside what they observe, describing a thing can move it — Soros called this reflexivity.
Predict a bank will fail and you may cause the run. Publish a model that says a stock is underpriced and the buying corrects the price. Rank the universities and they reshape themselves to the ranking. Knowledge of a social fact is itself a social act with consequences for the fact. There is no detached observation deck.
The natural scientist's dream — describe without disturbing — is structurally impossible where the described can read the description and act on it. Below is an honest engine: a true fundamental value, a crowd that trades on its model of the trend, and one dial — reflexivity — for how hard belief writes back into reality. At zero it is a thermometer. Crank it and watch the same small shock become a bubble.
Live engine · a fundamental, a crowd trading on its model, and their reflexive coupling
the observer inside
A thermometer reads a temperature it cannot change; the mercury and the air are decoupled, and that decoupling is what lets the reading be trusted. The dream of a detached science rests on it — an observer on a deck outside the system, recording without touching. But move to social life and the deck collapses. The people you are describing can read your description, and once they can read it they can act on it, and once they act the fact you described is no longer the fact. A run on a bank is a forecast that manufactures the insolvency it names. The map is bound into the territory by a wire the natural sciences do not have: the described can talk back.
The engine above draws that wire explicitly. There is a genuine fundamental — a real underlying value, drifting on its own noise — and a crowd that never sees it directly, only its own model: the recent trend, extrapolated. The crowd trades on the model, and here is the crux the reflexivity dial controls — their trading feeds back into the fundamental itself. At R = 0 the two are decoupled, thermometer-style: price is tugged home to a fundamental that does not care what anyone believes. Turn R up and belief begins to write reality — a rising price cheapens capital, flatters the story, draws the buyer whose buying confirms the model — until, for a while, the theory of the market really is a force in the market.
What to try
Leave reflexivity at 0 and press shock the price: the spike decays straight back — self-correcting, "scientific." Now drag reflexivity toward 1 and shock again. The same disturbance no longer decays; it feeds on itself, and a bubble climbs out of a nudge.
Press publish a bullish forecast. The crowd adopts the number and trades toward it. With high reflexivity the buying is validated and the forecast fulfils itself; with low reflexivity arbitrage drags price home and the same forecast defeats itself. Same call — opposite fate, decided by the feedback sign.
Toggle metric → target. Now effort flows to the number instead of the thing: the metric (price) is talked and gamed upward while the true quality it was meant to track quietly rots beneath it. The gap becomes a lie you can watch widen — Goodhart's law, live.
Ease reflexivity up slowly with the noise running. There is a threshold where wobble stops decaying and starts compounding: below it, a placid instrument; above it, booms and crashes from nothing but noise. The regime badge flips from anchored to boom to crash.
boom, bust, and the snap
Watch a reflexive boom in slow motion and you can see the loop turning. A price rise generates momentum; the crowd's model reads the momentum as signal and buys; the buying lifts the price; and because reflexivity is on, the rising price bends the fundamental part-way up to meet it — cheap financing, confident spending, the story growing true enough to keep the buyers buying. Belief writes reality, and reality then confirms the belief. For a stretch this is not a mistake anyone can arbitrage away, because the thing the sceptics would short keeps, maddeningly, coming true. This is why reflexive systems have fat tails and regime changes that the tidy "efficient" picture keeps missing: nothing in a self-correcting model of prices predicts a decade-long mania.
But the anchor never fully lets go. Reality is bent, not abolished — the bedrock fundamental keeps pulling, and the further belief has run, the harder that unbent remainder pulls back. Momentum is finite; the marginal believer runs out. When the rise merely stalls, the model that read "up" now reads "down," and the whole loop runs in reverse, now amplifying the fall: selling begets lower prices begets a decaying fundamental begets more selling. That reversal is the snap — the moment the gap between belief and fundamental, stretched past what the story can hold, tears. In the engine you can force it (shock a running bubble) or just wait: every reflexive boom above carries, in the size of its own gap, the crash that ends it.
living with reflexivity
Once you have felt the wire, you cannot unfeel it, and it changes what it means to describe a social world. A forecast, a metric, a rating, a model — each is an intervention: an input the system can read and respond to, a nudge dressed as an observation. The credit rating that downgrades a country raises its borrowing cost and helps bring on the default it predicted. The league table meant to measure schools becomes the thing schools optimise, until it measures optimisation. To publish a model of a social system is to hand that system a new thing to do. Robert Merton named the benign case the self-fulfilling prophecy in 1948; Soros generalised it into a theory of markets that are never in equilibrium because the participants' thinking is part of what they are pricing.
This lays a responsibility on anyone who describes a social world from inside it — which is everyone. If your description is also a lever, then choosing to publish, how to frame, which metric to elevate, is an ethical act and not merely an epistemic one. And yet reflexivity is not omnipotence, which is the honest limit the engine keeps insisting on: the bedrock fundamental is still in there, pulling, and it bites in the end. Belief can stretch reality for a long time; it cannot suspend it forever. A model of a social system is a force within it, but the world it describes is not made only of models — and the ones who forget that are the ones the snap is coming for.
The mapping
| In the model | In social life |
|---|---|
| the fundamental | The underlying reality — a firm's real prospects, a school's real quality, a country's real solvency. |
| the model / belief | The theory acted on — the extrapolated trend, the forecast, the rating everyone trades toward. |
| reflexive feedback (R) | Action changing the reality it describes — buying that cheapens capital, panic that drains a bank. |
| low reflexivity | A natural system, thermometer-like: the observer does not disturb; shocks decay back to the fact. |
| high reflexivity | Self-reinforcing booms and busts: belief drives the fact far from bedrock, then far the other way. |
| the snap | The gap between belief and fundamental stretched past holding, and breaking — the crash, the run. |
Where it tears
The wire exists; the deck really has collapsed. But the seductive over-reading — that since belief moves reality, reality is only belief, and fundamentals are a fairy tale — is exactly the thought that gets you long at the top. In the engine the anchor never disconnects, and it wins in the end. Reflexivity loosens the coupling between belief and fact; it does not sever it. Treating a stretched gap as the new permanent normal is not sophistication. It is the mistake the snap is built to punish.
The model draws a crisp bedrock value that belief deviates from — but in much of social life there is no such number. What is the "true" price of a reputation, the fundamental value of a currency, the real quality of a novel? Often the value is constituted by the very beliefs and conventions the model treats as mere noise around a fact. Where there is no bedrock, "belief versus fundamental" becomes "belief versus other belief," and the tidy gap the readout reports is drawing a line the world may not actually have.
Reflexivity can be weaponised. If describing a bank will start the run, someone can describe it on purpose — talk a book, manufacture a panic, pump a coin, seed the rumour that makes itself true. The same diagram that lets you see the self-fulfilling prophecy tells a bad actor how to build one. So the ethics are not decoration: an honest account of reflexivity has to admit that it hands power to whoever grasps it first, and that the responsibility of the observer-inside is not only to describe well but to reckon with who will use the description, and how.