In October 2003, the Columbia Accident Investigation Board released its report on the Space Shuttle Columbia disaster, which killed seven astronauts when the shuttle broke apart on reentry on February 1, 2003. The cause of the physical failure was already known: foam insulation had struck the shuttle's wing during launch, damaging the thermal protection system. The investigation board's deeper finding concerned communication: specifically, how the way information was presented to NASA decision-makers systematically obscured the severity of the risk until it was too late to act.

An engineer at Boeing, analyzing footage of the foam strike, had produced a detailed technical presentation showing that the damage posed a potentially catastrophic risk. That presentation was delivered to NASA management in PowerPoint format. The investigation board found that the slide structure had fragmented and obscured the underlying analysis to the point where key warnings -- information that was present in the underlying data -- did not register with the decision-makers who received the presentation.

The board's report, citing Edward Tufte's analysis of the specific slides, concluded: "As information gets passed up an organization hierarchy, from people who do the analysis to mid-level managers to high-level leadership, key explanations and supporting information are filtered out. In this context, it is easy to understand how a senior manager might read the title [of a slide] and not realize that it reported a disaster in the making."

The Columbia disaster is an extreme case. But the underlying dynamic it illustrates -- critical signal overwhelmed and obscured by surrounding noise -- is a routine feature of organizational and professional communication, with consequences that accumulate invisibly across every decision that depends on transmitted information.

"The commonality between science and art is in trying to see profoundly -- to develop strategies of seeing and showing." -- Edward Tufte, The Visual Display of Quantitative Information (1983). The same principle applies to all communication: clarity is a form of visual and cognitive design.

Signal vs. Noise: Key Sources and Remedies

Noise Type What It Looks Like Why It Accumulates Remedy
Throat-clearing "Today I'd like to share some thoughts on..." Feels polite; signals preparation Cut to signal immediately
Context inflation 20 slides of history before reaching the current situation Communicator confuses what they needed to know with what receiver needs Provide context only for what is necessary to act
Redundant restatement "In summary, to recap, the key takeaways are..." Feels thorough; pads content State once, clearly
Defensive hedging "Of course this is just one perspective and there are many ways to look at this..." Protects communicator from accountability Quantify real uncertainty; remove performative hedging
Irrelevant precision Ten decimal places in a strategic market estimate Accuracy signals effort Match precision to the decision's actual requirements
Filler language "At the end of the day," "needless to say" Habit; fills pauses Remove in editing

What Signal and Noise Mean in Communication

The signal/noise distinction originates in electrical engineering and information theory. Claude Shannon's foundational 1948 paper "A Mathematical Theory of Communication" formalized the concept: in any channel through which information is transmitted, signal is the component that carries the intended message, and noise is everything else that the receiver must filter out to recover the message.

In human communication, the concepts translate as follows:

Signal: Information that advances the receiver's understanding of what matters -- the main point, the essential evidence, the critical qualification, the actionable conclusion.

Noise: Everything present in the communication that does not advance that understanding -- irrelevant context, redundant restatement, hedging that does not add genuine qualification, structural complexity that obscures rather than organizes, filler language, unnecessary background.

The critical insight from information theory, which applies directly to human communication, is that noise is not merely inefficient -- it actively degrades reception of signal. A receiver's cognitive processing capacity is finite. Processing noise consumes capacity that could be used to process signal. When noise is sufficiently dense, signal reception fails -- the receiver can no longer reliably extract the intended message from the surrounding irrelevancy.

This is why the Columbia engineers' real-world warnings were missed. The warnings were present. They were stated. They were somewhere in the slides. But they were surrounded by so much noise -- extraneous context, hedged qualifications, fragmented structure -- that they did not register as the load-bearing information they were.

*Example*: In a landmark 2000 study, psychologist David Kirsh asked participants to sort objects into categories after viewing a workspace. Participants who had viewed a workspace with only the relevant objects performed the sorting task significantly faster and more accurately than participants who had viewed a workspace with relevant and irrelevant objects mixed together, even though the relevant objects were identical in both conditions. The irrelevant objects consumed attention that could not then be devoted to the relevant ones. Physical clutter degraded attention to target objects -- the same mechanism by which communication noise degrades reception of signal.

Sources of Noise in Professional Communication

Understanding where noise comes from makes it easier to identify and eliminate.

Throat-clearing: Preliminary content that delays rather than introduces the signal. "I'm going to take a few minutes today to share some thoughts about our Q3 performance and what it might mean for our strategic planning process" is throat-clearing; "Q3 revenue was down 8% and the trend is likely to continue unless we address two specific structural issues" is signal. The former announces that signal is coming; the latter is the signal.

Context inflation: Providing far more background than is necessary for the receiver to understand and act on the main point. The most common form is the historical review that occupies half of a presentation before reaching the current situation -- information that specialists providing regular updates already know, and that newcomers could be directed to reference material for. Context inflation occurs because communicators conflate "what I needed to know to reach this conclusion" with "what you need to know to act on this conclusion."

Redundant restatement: Saying the same thing multiple times in different words without adding new information or qualification. "In summary, to recap what we covered today, the key takeaways are the following, which reiterate the main points..." is pure noise. It consumes processing capacity while adding no new content.

Defensive hedging: Qualification that serves the communicator's risk-aversion rather than the receiver's need for accurate framing of uncertainty. "Of course, this is just one perspective, and there are many other ways to look at this, and reasonable people could disagree, and this analysis has limitations that should be considered..." is hedging that neither quantifies the uncertainty (which would be useful) nor identifies the specific conditions under which the claim breaks down (which would be useful) but merely signals that the communicator does not want to be held responsible for the content they are presenting.

Structural complexity without organizational benefit: Headers, subheadings, bullet points, and nested lists that create the visual appearance of organization without helping the receiver navigate the content. A seven-level bullet hierarchy is more disorienting than helpful prose would be.

Filler language: "As you know," "needless to say," "for all intents and purposes," "at the end of the day" -- phrases that take up space without conveying content.

Irrelevant precision: Detail that is accurate but not useful at the level of decision being made. A financial analysis prepared for a board decision about whether to enter a new market does not benefit from ten decimal places of precision in market size estimates; the decision will be made on orders of magnitude, not basis points.

Why Noise Accumulates

If noise degrades communication quality, why is so much professional communication noisy? Several organizational and psychological dynamics systematically produce noise:

Noise signals effort: In many organizational cultures, longer communication signals more work, more thoroughness, and more commitment. A one-page executive summary is perceived as less serious than a thirty-page report, even if the thirty-page version contains twenty-seven pages of redundancy around a three-page core. Blaise Pascal's 1657 apology -- "I would have written a shorter letter, but I did not have the time" -- correctly identifies that concise communication often requires more work than verbose communication. Organizations that do not reward this work will receive less of it.

Noise provides defensive cover: Communicators who are uncertain about their conclusions have incentive to obscure that uncertainty in volume. A confident claim that turns out to be wrong is more professionally costly than the same claim buried in so many qualifications and caveats that it is impossible to tell what was actually asserted. Vague communication is often not a failure of skill but a successful defense against accountability.

Noise reflects incomplete thinking: Writing and speaking without prior clarity about what the main point is produces communication that searches for its own conclusion rather than presenting one. The draft that has not been edited is typically twice as long as the final version should be, because the drafting process includes the process of figuring out what you actually want to say. Noise accumulates when that process of figuring out is presented to the receiver rather than retained by the writer.

Tools optimize for production, not reception: Email clients, presentation software, and word processors are designed to make it easy to produce content, not to minimize receiver cognitive load. PowerPoint makes it easy to add a slide; it does not remind the presenter that each additional slide divides the receiver's attention across more information. The default in communication tools is more, not less.

The Economics of Attention

Every communication competes for a finite resource: the receiver's attention. Attention is a scarce resource in a communication-saturated environment. Every unit of attention consumed by noise is unavailable for signal.

This creates an asymmetry that is important to understand: from the sender's perspective, noise has low cost (it is easy to produce). From the receiver's perspective, noise has significant cost (it consumes the attention they would otherwise direct at signal). This asymmetry is one reason noisy communication persists despite being counterproductive: the person who bears the cost (the receiver) is not the person making the production decision (the sender).

Effective communicators internalize the receiver's perspective and price attention accordingly. The question is not "what information is potentially relevant?" but "what information does this receiver need to accomplish the decision or understanding that is the purpose of this communication?" Anything that does not serve that specific purpose is noise, regardless of its inherent interest or validity.

*Example*: Amazon's "Working Backwards" product development method includes a practice called the "six-pager": key decisions are written up as six-page narrative memos rather than slideshow presentations. The discipline of the format forces the writer to do two things: (1) think clearly enough about the subject to state reasoning explicitly rather than gestured at through bullet points, and (2) make choices about what to include and exclude rather than including everything that could potentially be relevant. The format's page limit is explicitly a noise-control mechanism. Products that have successfully gone through this process include Amazon Web Services, Amazon Prime, and Amazon Echo -- none of which were conventional ideas that would have survived a slide-based presentation process that rewarded apparent comprehensiveness.

The Five-Question Signal Test

Before finalizing communication, five questions identify noise that should be cut:

1. What does the receiver need to do or understand after receiving this communication? If you cannot answer this clearly, you have not identified your signal. Every element should serve this identified purpose.

2. For each element of the communication: is this necessary for the receiver to accomplish their purpose? If not -- if they could achieve the purpose without it -- the element is noise. It may be interesting noise, accurate noise, or impressive-sounding noise, but it is still noise.

3. Could any section be removed without the receiver noticing anything is missing? If yes, it is noise by definition -- its absence would not be felt.

4. Does the structure help or require attention? Attention that is spent navigating structure is not being spent processing signal. If structure is consuming more than about 15-20% of processing effort, it has become noise.

5. Who bears the cost of this content? If a piece of content requires significant receiver effort to process but provides minimal receiver benefit, the communicator is externalizing costs. That content is noise from the receiver's perspective regardless of how much effort went into producing it.

Signal Without Enough Context

The counterpart failure to excessive noise is insufficient signal -- communication that is so stripped of context that the receiver cannot interpret what remains.

A one-word email containing only "No" is maximally concise and frequently maximally unclear. Signal without adequate framing context is uninterpretable; the receiver must supply the missing context from their own understanding, which may not match the sender's. The result is misinterpretation that the sender attributes to the receiver's failure to understand clear communication and the receiver attributes to the sender's failure to communicate.

The calibration challenge is to distinguish context that enables signal interpretation (necessary) from context that surrounds and expands upon signal (often noise). The former must be present; the latter should be ruthlessly evaluated.

The why miscommunication happens framework is relevant here: most miscommunication is not a failure of signal transmission but a failure of shared interpretive context. Receivers reconstruct meaning from signal using their existing context; if that context differs from the sender's, reconstruction diverges from the intended message. The minimum context necessary for correct reconstruction is signal; everything beyond that is noise.

Organizational Systems for Managing Signal Quality

Individual communicators can apply signal/noise discipline to their own communication. Organizations face the additional challenge of managing signal quality across communication networks with many nodes and multiple layers of transmission.

Several organizational practices improve signal quality systemwide:

Response-required framing: Rather than sending information and leaving interpretation to receivers, effective organizations explicitly state the decision or action that the communication is intended to support. "For your information" communication has no stated signal destination and produces corresponding ambiguity. "For decision: [specific decision]" creates a target against which signal quality can be evaluated.

Summary-first conventions: Organizations that require communication to lead with the conclusion before presenting evidence eliminate much of the receiver cognitive load involved in processing narrative that reveals its point only at the end. The legal brief format -- statement of position, then evidence -- is a template for this approach that has been refined over centuries.

Active questioning norms: Organizations where it is normal and expected for receivers to push back on noise ("What decision does this serve?", "What do you need from me after reading this?") create market pressure that reduces noise over time. Organizations where receivers process whatever they receive without feedback to senders create no incentive for senders to invest in noise reduction.

Communication audits: Systematic review of internal communication against signal/noise criteria can identify patterns of noise accumulation (particular senders, particular formats, particular contexts) that can be addressed systematically.

Signal in High-Stakes Communication

The consequences of noise are proportional to the stakes of the communication. In routine low-stakes communication, noise creates inefficiency -- wasted time, reduced comprehension -- that is mildly costly. In high-stakes communication, noise can be catastrophic, as the Columbia case demonstrates.

In emergency communication, military command, aviation, medical handoffs, and other high-stakes domains, explicit protocols exist precisely because informal communication norms produce too much noise. Surgical checklists, aviation read-back protocols (the receiver reads back what they understood the transmission to mean), and structured medical handoff formats (SBAR: Situation, Background, Assessment, Recommendation) are all noise-reduction systems. They enforce signal clarity through structure because the cost of noise in those contexts is too high to leave to individual communicator discipline.

The underlying principle is applicable beyond those specialized domains: when the stakes are high, impose structure that forces signal clarity. Do not rely on communicators' natural inclination toward conciseness, because that inclination is typically outweighed by noise-generating forces. Design the structure so that noise is architecturally difficult to include.

What Research Shows About Signal and Noise

Claude Shannon and Warren Weaver's 1949 foundational work established the mathematical basis for signal-to-noise ratio, but researchers studying human cognition have since extended these ideas in important ways. George A. Miller's landmark 1956 paper "The Magical Number Seven, Plus or Minus Two" quantified the central bottleneck: human working memory can hold only about seven items at once. Once a receiver's working memory is saturated with noise, signal reception degrades sharply -- not gradually.

Richard Mayer, a cognitive psychologist at the University of California Santa Barbara, spent two decades experimentally demonstrating the "redundancy effect": adding extra information to a message actually reduces learning compared to the leaner version. In studies published in his 2001 book Multimedia Learning, Mayer found that learners who received narration plus on-screen text learned less than learners who received narration alone -- the text was noise that competed with the spoken signal for the same cognitive channel. This principle, called "coherence," is now one of the most replicated findings in educational psychology.

Gregory Bateson, the anthropologist and systems theorist, approached signal and noise from a radically different angle. In his 1972 collection Steps to an Ecology of Mind, Bateson defined information as "a difference that makes a difference." This reformulation is practically important: signal is not everything that is true or accurate, but only what causes a difference in the receiver's state of understanding. Much organizational communication transmits accurate information that makes no difference -- it is truth, but it is noise.

Paul Watzlawick and his colleagues at the Mental Research Institute in Palo Alto documented in their 1967 book Pragmatics of Human Communication a related phenomenon: in interpersonal communication, the relationship level of a message often overwhelms its content level. A technically accurate message delivered with condescension is received primarily as condescension; the information is noise-drowned. This means signal quality in human communication depends not just on content precision but on the relational framing within which content is embedded.

Real-World Case Studies in Signal Failure

The Columbia disaster, described in this article's opening, is the canonical case, but several other incidents illustrate how noise systematically defeats signal in high-stakes contexts.

In 1977, the deadliest accident in commercial aviation history occurred at Tenerife, killing 583 people. A critical factor identified by Dutch linguist Rene Plomp in subsequent analysis was noise at the channel level: the KLM pilot's statement "We are now at takeoff" was ambiguous between "We are positioned at the runway" and "We have begun our takeoff roll." The phrase was technically present in the transmission; its meaning was drowned in ambiguity. Post-accident reforms in aviation communications, particularly the standardized use of specific phraseology and read-backs, are noise-reduction protocols still in use today.

Amazon's six-page narrative memo practice, introduced by Jeff Bezos in the mid-2000s, represents a designed organizational response to noise accumulation. Bezos explicitly banned PowerPoint presentations from senior leadership meetings, replacing them with structured prose memos read silently at the meeting's start. His reasoning, documented in internal communications that became public, was that bullet-point format fragments reasoning in ways that obscure whether there is actually a coherent argument. The format forces signal quality because incoherent reasoning cannot hide behind visual structure. Products that succeeded through this process include Amazon Prime, AWS, and Alexa.

In 2009, the aviation industry implemented Crew Resource Management protocols specifically to address the signal problem identified in dozens of accident investigations. The finding across cases: junior crew members possessed critical safety information (the signal) but could not get it through hierarchical noise to the people who needed to act on it. CRM training restructured communication norms so that any crew member could and should surface safety-relevant information regardless of rank. A five-year study by Helmreich, Merritt, and Wilhelm (1999) found that airlines with stronger CRM implementation showed substantially lower accident rates -- a direct measure of the value of signal quality improvements.

The Science Behind Noise Accumulation

Why does noise accumulate so reliably despite its cost? Research across several disciplines converges on a set of mechanisms.

Psychological distance theory, developed by Yaacov Trope and Nira Liberman at New York University, explains part of the problem. When communicators think abstractly about their audience -- as a distant, generalized group rather than a specific person with specific cognitive constraints -- they produce more abstract, less concrete communication. Abstraction is a primary noise source: abstract language requires receivers to do additional inferential work to extract a concrete meaning, and each inferential step introduces error. Communicators who have concrete, immediate feedback about their audience (a face in front of them showing confusion) produce less abstract, lower-noise communication than those communicating to a distant, abstract audience.

Sian Beilock at the University of Chicago has studied how expertise affects communication. Expert communicators who attempt to explain their domain to novices systematically underestimate the knowledge gap -- a direct consequence of the curse of knowledge. But her research shows the problem is deeper: experts also use more technical vocabulary, more complex sentence structures, and more implicit references to background knowledge precisely because expertise rewires the brain to process these features automatically. Noise that experts do not notice as noise is experienced by novices as dense interference.

Nassim Nicholas Taleb's analysis of fragile information systems in The Black Swan (2007) identified what he called "narrative fallacy" -- the tendency to construct causal stories from data, adding connective noise that feels explanatory but often obscures the underlying signal. In communication terms, explaining why the data shows what it shows often generates more noise than the data itself, because the explanation encodes assumptions, framings, and background theory that may not be shared by the receiver.

References

Tags

communication signal and noise clarity information quality effective communication

Frequently Asked Questions

What is signal vs noise in communication?

Signal is the meaningful information you want to convey, while noise is anything that distracts from or obscures that message.

How do you identify signal in communication?

Signal is information that directly addresses the core message, answers key questions, or drives understanding and action.

What causes noise in communication?

Noise comes from unnecessary details, jargon, poor structure, redundancy, or irrelevant information that dilutes the message.

Why does reducing noise improve communication?

Less noise means the audience can focus on what matters, understand faster, and retain key information more effectively.

How can you improve your signal-to-noise ratio?

Be concise, prioritize key points, remove redundancies, use clear language, and structure information logically.

Share this article