Not all information changes equally as it moves through an institution. Ordinary compression removes detail because decision-makers cannot absorb everything. But some information faces an additional force: its content creates social, professional or organisational discomfort. Bad news can become systematically softer as it travels towards positions of greater authority.

This is not simply a communication problem, and it does not require anyone to lie. A weak signal may become “something to monitor”. A serious delay may become “implementation pressure”. A failing programme may become “challenging but recoverable”. At each stage the description remains defensible, yet the cumulative movement can reduce the severity that reaches the final decision-maker.

This is directional distortion. The distinction from generic information loss is essential. GENERIC COMPRESSION ≠ INCENTIVE-SHAPED INFORMATION DISTORTION. Ordinary compression can affect good news, bad news and neutral detail alike. Directional distortion is asymmetrical: some information is more likely to be moderated because transmitting it creates friction.

That friction can come from many sources. Managers may fear appearing unable to control their area. Staff may worry about damaging relationships with colleagues. Teams may become attached to programmes they designed. Contractors may hesitate to challenge clients. Officials may avoid escalating uncertainty until evidence becomes stronger because false alarms carry reputational costs. No single actor needs to intend deception for the institution to become systematically over-optimistic.

Hierarchy can amplify the effect because each transmission point creates an opportunity for moderation. A frontline employee describes a serious concern to a manager. The manager reports a significant issue to a director. The director describes a manageable risk to senior leadership. Each transformation may seem modest, but the difference between the original observation and the final representation can be substantial. The result is a peculiar institutional asymmetry: the people with the greatest formal authority may receive the most processed version of reality.

This helps explain why leaders can be genuinely surprised by failures that many people lower in the organisation partly anticipated. The information existed, but the severity did not survive transmission.

This phenomenon is different from bad news failing to receive attention. An institution may pay close attention to a report once it arrives. The problem here is that the content arriving is no longer cognitively equivalent to the content that began the journey.

Nor is the problem solved simply by instructing employees to “speak truth to power”. Individual courage matters, but a system that repeatedly makes accurate escalation personally costly should not depend on exceptional courage as its principal information architecture. Institutions need to examine the incentives surrounding transmission.

What happens to the messenger when a problem is escalated? Are early warnings valued even when they prove unnecessary? Does reporting uncertainty make a manager appear competent or unprepared? Are teams rewarded primarily for meeting targets, creating pressure to reinterpret inconvenient evidence? Can people bypass hierarchical layers when information is unusually consequential?

These questions reveal why reporting architecture and organisational culture affect institutional cognition. Digital systems can help by creating direct channels for information, but they do not automatically remove distortion. A dashboard may reduce opportunities for manual moderation, yet its categories can still sanitise reality. Automated status reports may classify complex situations as green, amber or red, forcing ambiguity into simple states.

Artificial intelligence can introduce further transformation. AI-generated summaries may inadvertently soften or intensify language depending on prompts, training patterns or optimisation for concise professional communication. A system asked to produce an “executive summary” may remove precisely the contextual detail through which urgency was visible.

The relevant governance question is therefore not only whether AI summaries are accurate but whether directional severity is preserved. Institutions can build countermeasures. Independent reporting channels can reduce local reputational pressure. Senior leaders can ask explicitly for disconfirming evidence and unresolved uncertainty. Review processes can compare source reports with later summaries to detect systematic attenuation. Organisations can reward early escalation rather than only accurate prediction.

But countermeasures must avoid creating a different distortion in which every problem is exaggerated to secure attention. The objective is fidelity, not alarmism. This is why institutional incentives matter. If people learn that only dramatic warnings receive action, they may amplify. If they learn that uncomfortable warnings produce punishment, they may soften. The reporting environment teaches actors how information must be presented to survive.

Over time, these adaptations can become cultural. People no longer experience themselves as distorting information; they simply learn what counts as an appropriately “balanced”, “constructive” or “senior-ready” message. The vocabulary of professionalism can then become a mechanism of cognitive moderation.

A reflexive institution should therefore occasionally trace important information backwards. What did the senior decision-maker hear? What did the director hear? What did the manager hear? What did the frontline observer originally see? Where did severity, uncertainty or disagreement change? The purpose is not to identify a culprit. It is to observe the transmission architecture.

Formal authority inevitably depends on information produced elsewhere. That creates a structural vulnerability: the people empowered to act may know reality mainly through representations shaped by actors who experience different incentives. An institution cannot assume that bad news reaching the top is the same bad news that entered the hierarchy. Sometimes the most important cognitive question is not whether the warning travelled upward, but what happened to the warning while it travelled.