From Distorted
The Sound of Authority
What happens between the first report and the statement the public hears?
A team tries a new way to answer service requests. An analyst, whose job is to study the results, writes a report. Some teams did better. Others did worse. A manager asks for a clearer summary. Another reviewer removes people's private details. Then a team prepares the public statement.
Each step has a reasonable purpose. We still need to check the result.
An organization has to turn many reports into something people can use. Does the final message keep the key facts? Does it explain what remains unknown, where people disagree, and what the results can tell us?
The organization, figures, and documents in this example are fictional. A pilot is a small trial used to test an approach before deciding whether to use it more widely.
Follow the memo
The analyst's draft
“During an eight-week pilot, average response time fell from ten working days to eight among participating teams. Two of the six teams reported longer waits. The comparison does not account for differences in case complexity. A longer evaluation is needed before attributing the change to the pilot.”
The average wait fell, but two teams had longer waits. Some cases may also have been harder to handle than others. That is what “case complexity” means. The report did not account for those differences.
These details belong together. The average does not describe every team. It also cannot tell us whether the trial caused the change. We would need more evidence to answer that question.
The department's revision
“The pilot's participating teams recorded a 20 percent reduction in average response time over eight weeks. Results varied by team, and the contribution of the pilot remains uncertain because case complexity was not controlled.”
The wording is shorter. The math is the same: ten days fell to eight, a drop of two days. Two divided by ten is 20 percent. The edit still says that teams had different results and that the cause remains unclear. “Was not controlled” means the comparison did not account for how hard the cases were.
A privacy edit
Suppose extra pages also name service users and describe their cases. An editor can remove those names to protect their privacy. The report can still keep the combined results, explain how they were measured, and state what remains unknown. To judge an edit, we need to know what was removed and why.
The announcement
“Our new approach has improved service performance by 20 percent.”
Now the meaning has changed. “Service performance” could include accuracy or customer satisfaction. The report only measured response time. The announcement no longer says which teams took part. It also claims the new approach caused the improvement, which the comparison did not show.
The sentence is easy to read. It asks us to accept more than the evidence shows.
Which edit most changed your understanding? Was it the shorter wording, the removed names, or the broader claim? Look at what each edit kept or lost.
See what each version keeps
Use this small reading guide alongside the fictional memo. Choose a version and check four details in its wording.
Analyst’s draft
- Response time is the measure
- Included
- Participating teams are named as the group
- Included
- Different team results are acknowledged
- Included
- The cause is described as uncertain
- Included
The draft keeps the measured result and its limits together.
Department’s revision
- Response time is the measure
- Included
- Participating teams are named as the group
- Included
- Different team results are acknowledged
- Included
- The cause is described as uncertain
- Included
The shorter department version keeps these limits visible.
Announcement
- Response time is the measure
- Not stated
- Participating teams are named as the group
- Not stated
- Different team results are acknowledged
- Not stated
- The cause is described as uncertain
- Not stated
“Service performance” is broader than response time. The announcement also says the new approach caused the improvement, which the comparison did not show.
“Not stated” describes this version of the wording. It does not tell us why a detail was left out. The privacy edit is discussed separately above.
What a process can protect
Working together can improve a statement. An editor may catch a math error. A person who knows the subject may spot a missing limit. A reviewer may catch a claim about everyone based on a small trial. A clear summary may help someone see a problem without reading the full report.
Some historical records show how a review can keep different views visible.
On November 6, 1974, L. William Seidman sent President Ford a memo about the Economic Policy Board. He proposed that the board research different choices, avoid repeating work, organize the work that followed decisions, and respond to changing conditions. His plan aimed to improve the information used to make decisions. Read the original memorandum.
In an August 29, 1975 memo, Roger B. Porter recorded different Treasury and Federal Reserve views on rules for handling gold. He named who supported each view and who did not take a side. A phrase like “the advisers recommended” could have hidden those differences. The archive's note also records that Ford approved the Treasury position. Read Porter's memorandum.
These memos record advice given to a president. They do not show how the final public message was worded. They show that a final decision can follow a record that keeps different views visible. We need to read that record before judging how the decision was made.
In Distorted, “The Sound of Authority” asks how an organization's rules and steps shape what we hear. A review may keep a disagreement visible, settle it, hide it, or leave it open. The name of the process will not tell us which happened.
Approved for what
“The statement was reviewed” tells us little about what was checked. We need evidence that the review took place and an explanation of what it covered.
A reviewer may check spelling, math, private information, earlier statements, or supporting evidence. These are different jobs. Passing one check does not mean passing them all.
Ask the same questions about an audit, which is a formal check of records, or an investigation. What was it meant to find out? Which records could the reviewers see? What could they not establish?
Imagine that a reviewer checked the fictional announcement's math. The 20 percent figure is correct. That check still would not show that the trial caused the improvement. Correct math and proof of a cause are different things.
An impressive signature cannot fill that gap.
Read the distance between evidence and wording
Return to the announcement. Ask clear questions that someone can answer:
- What improved: average response time, the experience of every user, or something else?
- Which teams and cases were included?
- What evidence shows that the trial caused the change?
- Where can we read how the result was measured, how teams differed, and what remains unknown?
- Who will update the statement if new evidence changes the answer?
An organization can answer these questions without sharing every private discussion. It needs to show what supports its public claim and what that support leaves open.
The answers may change our judgment. A linked report may contain the missing detail. A later comparison may supply stronger evidence about the cause. But if the organization will not say what it measured, readers cannot check what the announcement means.
Keep your conclusion within the evidence. A statement may claim too much. That does not prove that everyone who worked on it meant to mislead. Polished writing can explain strong evidence. It can also hide a weak claim.
A statement readers can use
The fictional organization could say:
“Average response time fell from ten to eight working days across participating teams during our eight-week pilot. Results varied, and differences in case complexity limit what we can conclude about the pilot's effect. We are extending the evaluation before deciding whether to expand it.”
In plain terms: the trial teams answered faster on average. Some cases may have been harder than others, so the cause is unclear. The organization will study the trial longer before deciding whether to use it more widely. Readers can follow each part of the claim and check later updates against it.
An organization's claim is easier to judge when you can trace it back to evidence. Check the statements you welcome as carefully as the ones you distrust.
Put it into practice: Use Keep a Claim Record to examine one announcement. Save the exact wording. State what its evidence shows. Write one question that would help you judge it.
This guide draws on “The Sound of Authority” and Chapters 11 and 12 of Distorted. The Clarity Framework offers related steps: Investigate Claims, Analyze Narratives, and Engage Disagreement. Use them to check claims, examine the stories built around them, and hear different views.
For more about how this work is done, explore THNK Research.
Save this page
You can copy this link to your notes or bookmarks.