Information Is Authority: Connecting Operational Reality and Strategic Intent.

Leaders often tell people closest to the work to make decisions. Then they leave the information needed for those decisions somewhere else.

A bus that looks available

Consider a constructed example in a large city transit system. A bus stops during the morning rush. A street supervisor sees stranded riders and an empty bus five blocks away. Sending the spare seems like the obvious way to restore service.

The operations screen says the spare is available. Yet a mechanic noted an intermittent door fault late yesterday without changing that flag. Dispatch has also counted on the vehicle for another route, but the commitment sits in a shift handoff note. The nearby driver may have too little duty time to complete the detour. The street supervisor can see the immediate problem clearly and still lack the facts needed to choose responsibly.

Meanwhile, the control center sees network commitments but cannot hear the fault, see the riders waiting near a busy intersection, or know that someone needs assistance transferring. Its short incident code is a compressed account, not the event itself.

These details are illustrative, not a report of a real transit incident or an assertion about how transit agencies normally operate. An actual authority would have its own maintenance-release, labor, dispatch, and service rules. The example exposes a design question that appears across industries: where are the facts, where is the authority, and who can connect them in time?

The gap runs both ways

The person at the point of action often knows what happened locally. Each retelling can lose details, even when everyone is acting in good faith. The person farther away may understand the organization’s commitments and priorities better, yet receive a thinner account of the scene. Proximity is an advantage for local observation, not a guarantee of complete judgment. Central position can offer a broader view, not a guarantee that a dashboard reflects the work on the ground.

Both sides need to understand the limits of their own view. The driver might say, “I saw the doors fail twice, but I cannot tell you whether that spare is cleared or needed elsewhere.” The control center might answer, “I can see the fleet and route commitments, but I cannot see whether the passengers can wait safely where they are.” A question from either direction should close a knowledge gap, not automatically be heard as blame.

The useful exchange separates what was observed, what was inferred, what remains an assumption, and what another person or record has verified. The field supplies conditions and physically possible options. Central operations supplies intent, competing commitments, constraints, and consequences elsewhere. Then the authorized people can decide whether the choice belongs locally, jointly, or centrally.

I learned this operating principle before I heard the term shared consciousness. In a civilian role supporting Department of Defense and Department of State efforts in the same theater of operations, I saw the value of people having a broad, common understanding so they could act within the mission. Later, Stanley McChrystal and his coauthors’ account of shared understanding and empowered execution in Team of Teams gave me a name for that pattern (McChrystal et al., 2015). I am describing an influence on my leadership thinking, not claiming to have participated in JSOC’s internal redesign.

Shared consciousness is not everyone receiving the same update. It is the ability to connect relevant local facts, wider objectives, dependencies, uncertainties, and decision boundaries quickly enough to act. Research on dispersed knowledge and coordination helps explain why neither a distant approval chain nor an isolated local choice can solve every problem (Hayek, 1945; Jensen & Meckling, 1992; Malone & Crowston, 1994).

The record can be accurate and still mislead

I keep a picture of a World War II aircraft marked with bullet holes in my office. It reminds me of Abraham Wald’s work on aircraft vulnerability. The damage visible on returning planes did not show damage to planes that never returned. The familiar picture simplifies the technical analysis, but it prompts a valuable question: what is missing because of the way we collected the data (Wald, 1943/1980)?

In the transit example, studying only completed trips misses riders who never boarded after a cancellation. Counting a parked bus as “available” can be accurate under one definition while hiding a maintenance hold. Measuring punctuality only at terminal points can miss people stranded between them. A short radio report may omit details for different reasons than the aircraft sample; I use Wald as an analogy about incomplete views, not as an identical statistical case.

Leaders should not wait until riders are standing in the rain to discover that dispatch and maintenance mean different things by “available.” They can identify recurring decision triggers in advance, walk the actual work with operators and support teams, test records against practice, identify who owns each critical fact, and decide what evidence would change a choice. This is preparation for judgment, not a demand for perfect information.

A practical alignment test

1. Decision rights: Who may decide, within what boundaries, and when must the decision widen? “Use your judgment” does not define a boundary.
2. Usable information: Can that person see relevant facts, their source and age, competing commitments, and what may be missing?
3. Competence: Has the person learned to interpret those facts, anticipate effects, and know when to bring in dispatch, maintenance, safety, or a leader with a wider view?
4. Accountability and learning: Can the organization review what was known and assumed at the time, distinguish a hidden system failure from an ignored hold, and repair the information path?

This is a practitioner diagnostic, not a validated scale. Information by itself does not confer formal authority or make a person competent. The four elements need to meet at a specific decision.

When the decision belongs at the center

Suppose the field confirms riders are safe and a cleared spare is ready. Central operations sees simultaneous disruption on two other routes, one relief driver, and an emergency service obligation. Even if the street supervisor can see all that information, no single local actor may have the mandate to decide which neighborhoods lose service. Central operations should make the allocation under the authority’s stated priorities after hearing the street account.

More visibility does not mean every choice must be decentralized. A safety hold is a boundary, not a tradeoff that a manager can quietly waive. Restoring one route may worsen service for another group of riders, particularly people with fewer alternatives. Authorized leaders have to set and explain priorities rather than allow an easily measured delay to stand in for the whole objective.

What AI can change

AI implementation often starts with business objectives: better service, lower cost, faster response. Those objectives matter. They do not tell a system which vehicle status lags reality, how a dispatcher uses a verbal handoff, or why an operator calls a mechanic before trusting the screen. Operational concerns and strategic objectives must be investigated together.

If an AI tool reads only the dispatch board, it can recommend the wrong spare faster. If leaders have mapped how the work actually happens, the tool may assemble maintenance release, driver coverage, route commitments, rider needs, and their last verified updates. It should show what is stale, absent, or contradictory and let the supervisor reach the underlying record and a knowledgeable person.

A large language model could make that context easier to ask about in ordinary language or by voice. A supervisor might ask, “What happened the last three times we moved the final spare from this corridor?” and then, “Which of those cases differs from this morning?” The interface can shorten a search and support follow-up questions. It cannot create a missing inspection release or turn an outdated note into a current fact. Fluency is not authority.

That gives leaders more work to do before expecting technology to speed decisions responsibly: determine which observations people can capture without disrupting operations, verify source ownership and update times, preserve the reason behind consequential choices, and establish how field corrections reach the records and the tool. The National Institute of Standards and Technology’s AI Risk Management Framework supports deliberate governance and measurement of such risks; it does not establish that this particular use will improve decisions (NIST, 2023).

A pilot could follow comparable service disruptions from first report to restored service. It would examine decision time, rider impact, safety exceptions, reversals, missed dependencies, escalation quality, and whether people could explain what they knew. A faster answer is not an improvement if rework or harm rises. Weather, staffing, demand, and incident severity complicate comparisons; this article reports no measured outcome.

The Human-in-Complexity connection

None of the underlying communication problems is new. Increasingly connected operations and faster technology make the consequences of a poor information path more immediate. Awareness asks what is actually happening and what the record omits. Judgment weighs the local and systemwide effects. Responsibility defines who may act and who owns the outcome. Learning Forward repairs the information path after an error. Community makes it safe for people to report uncertainty and ask questions without treating clarification as an accusation.

My working proposition is that organizations can make more sound decisions close to the problem when they connect operational reality to strategic intent before action and know when a decision must move to a wider level. Conversational AI may help people reach and question that context faster, provided the information reflects the work as done. Those relationships remain propositions to examine, not findings established by this illustrative case.