Signal

In every room, there are some people whose three sentences contain the entire problem and others whose twenty minutes only partially overlap with it. Two engineers describe the same bug; one's description fits in a paragraph, the other's expands across a meeting. Two analysts read the same report; one summarizes it cleanly, the other can only restate it. Two friends recount the same evening; one's account is a text message, the other's is a phone call. The underlying material is the same. What differs is the ratio of information to words.

The first explanation is that some people think more carefully than others. This is true at the extremes and false across the bulk of the population. Most of the difference between dense and diluted speech is not in the speaker's thinking. It is in what the speaker is using the speech channel for.

In 1996, the British anthropologist Robin Dunbar published Grooming, Gossip, and the Evolution of Language, in which he argued that human language did not evolve primarily as a system for transmitting information. It evolved as a substitute for grooming in primate groups that had grown too large to maintain through physical contact. The original function of language was bonding, the use of the voice stream to maintain a sense of relationship across many simultaneous pairs. Information transfer in the modern sense, the encoding of facts and arguments into compact propositions, was a later, secondary repurposing of a channel built for something else.

The repurposing is incomplete. In any sample of contemporary human speech, a meeting, a phone call, a dinner conversation, a public talk, only a fraction of the bandwidth is doing information work. The rest is doing the older work: maintaining relationship, signaling status, reaffirming identity, regulating emotional state, accommodating multiple audiences. These are not noise to the system. They are the system. But from the listener's perspective, in the moments where information is what they came for, they are noise relative to the information component.

This reframes the question. The person who is "pure signal" is not someone whose mind is denser than the rest. They are someone who has moved the non-information functions out of the speech channel and onto other substrates, freeing the channel to carry information almost exclusively. The pure-signal speaker is doing one thing in the channel where the noisy speaker is doing six. The difference between them is not cognitive, but allocative.

The Six Concurrent Functions

Any utterance does some combination of the following, drawing simultaneously on the same finite bandwidth.

Information. The actual content the speaker intends to deliver. The data, claim, argument, instruction. This is what survives if all other functions are stripped out.

Status. The placement of the speaker in the hierarchy. Vocabulary calibrated to indicate education, references calibrated to indicate exposure, sentence structures whose complexity is the point and not the medium of the point, hedges of the form "as I was saying to so-and-so." None of this carries information about the topic. All of it consumes channel bandwidth.

Identity. The reaffirmation of who the speaker is, their political alignment, professional affiliation, taste, posture, allegiance. Most utterances contain at least one small marker that says "I am the kind of person who would say this." The marker does not advance the topic. It maintains the speaker's self-model, in their own eyes and the listener's.

Anxiety regulation. The hedges, qualifiers, and pre-emptive concessions that exist to prevent the speaker from being attacked for what they have said. "It's just my opinion." "I might be wrong, but." "Take this with a grain of salt." Each is insurance, paid in bandwidth, against the possibility that the substantive claim is challenged. The substantive claim is the information; the insurance is the noise around it.

Audience management. The simultaneous accommodation of multiple listeners with divergent priors. Speech directed at one specific listener can encode for that listener's model. Speech directed at several at once, or broadcast to an unknown audience, hedges against each potential audience's potential objection. The hedges multiply with the audiences. Public-facing speech is therefore noisier by construction than intimate speech, regardless of the speaker's clarity.

Bonding. The phatic component, the small talk, the warmth signals, the rhythms that reaffirm "we are in good standing with each other." This is the original Dunbar function, the grooming substrate. It carries no information about anything except the relationship itself.

Most utterances do at least three of these concurrently. The information component is one thread among several. From the listener's perspective, the other threads are noise relative to the information, even though, from the speaker's perspective, they may be the more important content.

Externalization

What pure-signal speakers have done, often without ever describing it this way, is to move the other five functions out of the speech channel and onto substrates that do the same work without consuming speech bandwidth.

Status moves to track record. A trader whose P&L is public does not need to perform competence in conversation. A surgeon with a verifiable record does not need to assert authority. A founder whose company exists does not need to drop names. The status work has been done in another medium, and the speech channel is no longer required to do it. People without external records of competence have to do the status work in the channel, and the work shows.

Identity moves to stable selfhood. People whose sense of who they are is settled do not need to insert markers of allegiance into every sentence. People whose identity is contested, in transition, or under threat use speech as the substrate for keeping it intact, and the maintenance shows. The settling happens in places that have nothing to do with speech, in resolved conflicts, in long-running practices, in lives that have answered enough of their own questions to stop interrogating themselves in public.

Anxiety regulation moves to exposure. The hedge is what someone says when they have not done the work that would let them speak without hedging. The work is usually one of two things: actually knowing the subject deeply enough to be unembarrassed by being wrong about parts of it, or having been wrong publicly enough times that another wrongness no longer threatens the self. Both are accomplished outside the speech channel, in the time spent learning or in the experience of survived public failure. People who have done either kind of work hedge less. People who have done neither use the channel to insure against the possibility of needing the work done, and the insurance is the noise.

Audience management moves to clear targeting. Pure-signal speakers usually speak as if to one specific listener, even when many overhear. The encoding is for that listener's model. Listeners outside the model receive less, but the addressed listener receives more. The narrowness of address is what makes the density possible. Noisy speakers are usually trying to reach everyone within earshot, which forces the message to be encoded for the largest common subset of priors, which means encoding for almost no one's actual model, which is why the message becomes simultaneously less specific and longer.

Bonding moves to other channels. Time spent together, attention given without demand for reciprocation, action taken on the other person's behalf, physical co-presence. These do the relational work that phatic speech otherwise does, and they free the speech channel to carry information instead of warmth. People whose relationships are maintained by these other channels can use speech for information without the relationships eroding. People whose relationships are maintained primarily through speech cannot strip the bonding component without seeming to disinvest from the relationship itself.

When all five non-information functions have been moved out of the channel, what remains is signal. The speaker has not become smarter. They have stopped using one channel for six things.

The Price

This is not a free move, and the people who fail to make it are not always wrong to fail.

The pure-signal speaker is often perceived as cold, distant, blunt, intimidating, or socially unsafe. They have suppressed the warmth-signaling function of speech, and listeners who depend on phatic cues to feel connected feel disconnected. The suppression has not made the speaker hostile. It has redirected their bonding work to other channels, channels the listener may not see. From the listener's side, what registers is the absence of warmth in the speech, not its presence in the action or the time. Pure-signal speakers therefore have, on average, narrower social networks and a higher rate of being misread as rude.

Their audiences are smaller. Decompression-required speech narrows the set of receivers who can decode it in real time. The receiver must hold the speaker's model accurately enough to fill in what the speech leaves unsaid. Most receivers cannot. Pure-signal speakers are rarely popular speakers, regardless of the quality of what they say. The medium they have chosen excludes most listeners by construction.

Their public speech is fragile. Reduced hedging exposes them to attacks that hedged speech would have insured against. Specific claims fail publicly when ambiguous claims would have escaped resolution. They pay the cost of being wrong in public. Most people will not pay this cost, and they should not be expected to in environments where being publicly wrong is consequential.

There is also a deeper cost, harder to see. The neural substrate that processes language is interleaved with the social brain. The functions that share the speech channel do so partly because they share the underlying machinery. Suppressing the social functions of speech can attenuate the social capacities themselves. The person who has trained themselves into pure information transfer has often, in the same training, weakened the relational reflexes that ordinary speech exercises in passing. Pure signal is a redirection of biological resources, not a free skill.

Counterfeit Signal

The pattern above generates two failure modes that show up often enough in everyday life, especially in professional environments, to deserve naming.

False conviction. Most professional environments reward confidence and punish hesitation. Saying "I don't know" carries a cost. Saying something firm and definite, even when one barely understands the topic, often does not. A specific habit develops in response: people learn to speak with authority about things they have not actually thought through, because the alternative, hedging or admitting ignorance, makes them invisible in the room.

This is the executive who has strong, decisive opinions on a technical question they could not begin to evaluate themselves. The consultant who diagnoses an entire business after three days of slides. The colleague who never qualifies a claim, even claims invented on the spot. The pundit who has equally confident takes on every topic in rotation. The participant in a meeting who says something, anything, with conviction, because saying nothing means being passed over and saying something hesitant means being dismissed.

What is happening is that confidence has come uncoupled from the work that originally justified it. In rooms where the listener cannot easily verify the claim, and where sounding sure reads as knowing, the move that wins is to drop the hedges regardless of whether the underlying knowledge is there. The surface form of pure signal becomes a posture available to anyone willing to adopt it. The point gets across, the speaker is treated as authoritative, and the actual information content of what was said is close to zero.

This proliferates specifically in environments without fast resolution. Trading floors do not tolerate it for long because positions resolve into profit and loss. Operating rooms do not tolerate it because the patient either improves or does not. Boardrooms, panels, conferences, strategy meetings, and most office conversation tolerate it indefinitely, because the claims made there are rarely tested in a window short enough that the speaker can be tracked back to them. By the time the decision turns out to have been wrong, the meeting is six months in the past, the people who confidently advocated for it have moved on to advocating for something else, and the link between the original conviction and the eventual outcome has been quietly lost.

Performative behavior. The second failure is doing things specifically to be seen doing them. The activity exists, often visibly so, but its purpose is the visibility, not the result the activity is nominally for.

This is the colleague who works obviously late hours, more focused on being seen working than on what is produced. The executive whose calendar is filled with meetings that exist mainly to be attended. The volunteer whose involvement intensifies when there is a camera and recedes when there is not. The "ally" whose support consists of public statements rather than private follow-through. The expert whose primary output is the maintenance of their personal brand rather than the work the brand is supposed to be a record of. The product launch designed for the press release. The CC line added so the boss will see the email. The conference appearance whose function is the photograph at the conference.

In each case, an activity that could in principle have produced something real has been hollowed into a display of itself. The form is intact: the late hours, the meetings, the launches, the appearances. The substance has been replaced by its own image. From outside it looks like work; on closer inspection, it is mostly the appearance of work. The performer is producing a strong signal in the social channel — "I am working hard," "I am committed," "I am the kind of person who does this" — and producing very little in the substantive channel that the social signal is supposed to point at.

The harder version is when real work is done, but its shape is bent toward being seen. The presentation gets more polish than the analysis it summarizes. The project is selected because it will read well in a quarterly review, not because it would actually accomplish what the review claims. The hire is promoted for being legible upward rather than for being effective in role. The substrate is real, but the substrate has been deformed by visibility, and the work that ends up getting done is whichever portion of it produces a recognizable signal, not the portion that would have produced the underlying value the signal pretends to represent.

In both failure modes, the test that separates real from counterfeit is resolution. The confident claim either survives follow-up, or it does not. The visible work either produces a real outcome, or it does not. The track record of the false-conviction speaker, when actually checked, is full of confident assertions that did not pan out. The output of the performative worker, when measured against what was supposedly being done, is a fraction of what the visibility implied. Both can sustain themselves for years inside environments where no one is keeping the score, and both collapse quickly inside environments where someone is.

The practical move, for the listener, is to weight resolution more heavily than tone. Confident speech and visible activity are cheap to produce; what is expensive is the underlying claim or output that the confidence and visibility were supposed to indicate. The right question is not whether the speaker sounds certain, or whether the worker looks busy. It is what becomes of their claims and their work when reality is given a chance to settle the account.

Defensive Noise

If the price is real, why is the question worth asking?

Because most of the noise in most people's speech is not bonding noise. It is defensive noise. The status assertions, identity markers, anxiety hedges, and audience accommodations are not building relationships. They are managing threats the speaker has not addressed elsewhere. None of this work is purchased deliberately. It happens because the underlying conditions that would make it unnecessary have not been established.

The hedger has not made peace with being wrong. The status signaler has not built a record that would let them stop signaling. The identity maintainer has not stabilized the self. The audience manager has not narrowed the address. Each is the consequence of work that has not been done, not a feature of the person.

When the work is done, the corresponding noise drops out of the speech automatically. The speaker has not learned to communicate better. They have removed the threats that were forcing the channel to do defensive work. The channel becomes quieter on the dimensions where defense is no longer needed, without any conscious change in how the speaker is choosing to speak.

This is why pure-signal speech is overrepresented in people who have been through specific kinds of pressure. Traders sizing positions on actual money. Surgeons making decisions on living tissue. Soldiers under fire. Founders running out of runway. Pilots flying single-engine in weather. The pressure does not directly improve their communication. It removes the option of using the speech channel for defensive work that would not survive contact with the situation. The pilot cannot hedge against radio static. The surgeon cannot perform identity in the middle of a bleed. The trader cannot status-signal a position that is moving against them. The non-information functions are stripped out by the environment, and the channel is left carrying what it can still afford to carry.

The same pattern explains why pure-signal speech often increases with age in people who have lived consequential lives, and does not increase with age in people who have lived insulated ones. Time alone does not strip the defensive work. Survived consequence does.

Implications

The most useful inversion is the one applied to oneself.

A person whose speech is dense with information has not necessarily thought more carefully than a person whose speech is dense with hedges. They have differently allocated the same finite bandwidth. To produce more signal, the project is not to think harder before speaking. It is to identify which of the non-information functions one's speech is doing, and to externalize them: build the record that makes status assertion unnecessary, settle the identity that does not need continuous reaffirmation, accept the wrongness that obviates the hedge, narrow the address to one specific listener at a time, do the bonding work in non-verbal channels.

The reverse is also true. Speech that is full of warmth and social padding is not failing at communication. It is succeeding at the original Dunbar function. The information component is a small thread inside a fabric whose main purpose is to maintain relationship. That fabric does work that pure signal cannot do, and a society of pure-signal speakers would lose more than it would gain. Pure signal is a specialization, not an upgrade.

The contemporary twist is that one of the historical functions of speech, pure information transfer, is now also produced by non-human systems. Large language models output information without status, identity, anxiety, audience management, or bonding, because the model has none of these to do. This is competitive on the information dimension precisely because human production of information is encumbered with the other five functions and the model's is not. Humans who want to compete on information density face the same trade-off pure-signal speakers have always faced, externalize the social functions or accept lower information density. Humans who want to retain the social functions will find that information transfer is increasingly handled elsewhere, and that human speech becomes more clearly what it has always primarily been: a relational channel that incidentally carries some information.

The choice between signal and noise is therefore not a choice about clarity of thought. It is a choice about what one's speech is for. The pure-signal speaker has decided that their speech is for information and has paid in social capital for that decision. The noisy speaker has decided, almost always without ever consciously deciding, that their speech is for the other things. Both are using the same finite channel. Each is paying for what the other is not.

What looks from the outside like the difference between a sharp mind and a soft one is most often, on closer inspection, the difference between a channel doing one job and a channel doing six. The mind underneath may be the same.