Exercises — Chapter 38: Ethics and Responsibility in Technical Writing

Writing is learned by writing. These exercises ask you to detect dishonesty in real-looking documents and to produce honest ones. Where a task is open-ended, use the self-assessment rubric provided instead of looking for a single right answer.

A reminder on the cases: keep the Challenger, 737 MAX, and medical examples to the verifiable public record. Do not invent quotations, internal text, dates, or casualty figures in your answers (Chapter 11 citation discipline).

Part A — Analyze This (Spot the Ethical Problem) ⭐

Each passage below contains no outright false statement, yet each is dishonest. For each, name the violation (overstatement, lie of omission, buried conflict, false confidence, exclusionary language, or spin) and point to the exact words that carry it.

A1. "Our new battery chemistry delivers industry-leading performance, dramatically extending device life. Early results are extremely promising and validate years of investment."

A2. "A study confirms our air purifier removes 99.9% of airborne particles." (The study, run by the manufacturer's own lab, tested removal of a single particle size in a sealed 1-cubic-meter chamber over 24 hours.)

A3. "The migration will be straightforward and low-risk. We recommend proceeding immediately to capture the substantial efficiency gains." (The team's own notes record an unproven dependency on a legacy billing integration and an expected 2–4 week slowdown during cutover.)

A4. "When the user makes an error, he should consult the manual. A sanity check by a normal operator usually prevents this."

A5. "Results were statistically significant, demonstrating the treatment's effectiveness." (n = 22; the reported difference was 1.1 points on a 10-point scale; participants were unblinded.)

A6. "We take your privacy seriously and have implemented robust security measures." (Written in a breach-notification email; no specifics about what was accessed or what the reader should do.)

A7. "The system has been thoroughly tested and is ready for production." (Test coverage excludes the payment module and all error-handling paths; this is not stated anywhere.)

A8. "Patients tolerated the medication well, with no serious adverse events in the trial." (Three patients withdrew due to side effects; "no serious adverse events" is technically true under the trial's definition of "serious," but the withdrawals are not mentioned.)

Self-check: A1 overstatement ("industry-leading," "dramatically," "extremely promising," "validate"—all unquantified intensifiers). A2 buried conflict (manufacturer's own lab) + misleading conditions ("99.9%" under one particle size in a sealed chamber, presented as general). A3 spin / hidden risk (the unproven dependency and slowdown are omitted; "straightforward and low-risk" contradicts the team's own notes). A4 exclusionary language ("he," "normal operator," ableist "sanity check"). A5 overstatement ("demonstrating effectiveness" from n=22, tiny effect, unblinded). A6 false reassurance + omission (a breach email must say what happened and what to do; "robust" and "seriously" are content-free). A7 lie of omission ("thoroughly tested" hides the untested payment and error paths). A8 lie of omission via definitional cover ("no serious adverse events" is true under the narrow definition but the three side-effect withdrawals are material and hidden).

Part B — Revise This (Make It Honest) ⭐⭐

Rewrite each passage so the whole document leaves a true impression. Keep the legitimate persuasive intent—you are not gutting the message, you are making it honest. State your assumptions if you need to invent plausible specifics.

B1. Revise A1 (the battery claim) into a calibrated statement that reports an actual magnitude and names what hasn't been tested yet.

B2. Revise A3 (the migration recommendation) using the §38.5 pattern: lead with the recommendation, quantify the benefit honestly, and surface the one genuine risk—without making the recommendation collapse.

B3. Revise this limitations note so a non-specialist can use it (§38.6 "state the limitation and its consequence"):

"Limitations: the sample was a convenience sample; the follow-up period was 30 days; the model was not externally validated."

B4. Revise A6 (the breach email) into an honest disclosure: what is known, what is not yet known, and the specific action the reader should take. Keep it short and calm.

B5. Revise A8 (the trial summary) so the side-effect withdrawals are disclosed without overstating the harm—calibrated in both directions.

B6. This sentence understates a real, well-supported finding into invisibility. Rewrite it to claim exactly what the evidence licenses—no more, no less:

"It may possibly be the case that the intervention could potentially have had some modest effect on the outcome in certain circumstances, though more research is of course always needed." (The trial was large, pre-registered, blinded, and found a clear, replicated 18% reduction.)

Self-check rubric (B1–B6): (1) Does the whole passage now leave a true impression, not just true sentences? (2) Is each claim calibrated to its stated evidence—neither inflated nor hedged into mush? (3) Are decision-relevant limitations surfaced and paired with their consequence? (4) Did you preserve the legitimate message (the recommendation still recommends; the finding still reports a real effect)? If you removed the persuasion entirely, you over-corrected.

Part C — Write This (Produce the Document) ⭐⭐–⭐⭐⭐

C1. Disclose the uncertainty honestly. You are Dana, a data scientist. Your churn model: ~70% accuracy on held-out data; validated only on the last two quarters; noticeably worse for enterprise accounts (your most valuable segment). Write the two-to-three-sentence executive-summary opening that is confident enough to be useful and honest about the uncertainty. Then write the one-line version a less scrupulous analyst would write, and in a sentence, explain what the honest version costs and buys.

C2. The honest recommendation. Write a 150–200-word recommendation memo (BLUF, Chapter 4) proposing your team adopt a new tool. Invent plausible specifics. Requirement: include at least one real cost and one genuine risk, stated prominently, and still make the case. Then mark, in the margin or a footnote, the sentence a spin-writer would have deleted—and why deleting it would have been dishonest.

C3. The safety-critical instruction. Write a short set of numbered steps (4–6) for a task where a mistake is dangerous—invent the scenario (e.g., resetting a medical device, handling a chemical, an emergency shutdown). Apply Chapter 22 and Chapter 36: warnings before the step they govern, unambiguous units, one action per step, no abbreviation that can be misread. Then write one sentence naming the specific ambiguity you most had to design out.

C4. The honest limitations section. Take any analysis you have actually done (a class project, a work report). Write its Limitations section using all four §38.6 techniques: prioritized by decision-relevance, each limitation paired with its consequence, findings separated from uncertainties, language calibrated once. Aim for 120–180 words.

C5. The conflict disclosure. You are writing a product comparison, and your company sells one of the products. Write the two-to-three-sentence disclosure that an honest reader would want—where it should appear, and what it should say—so the comparison can still be useful rather than dismissed.

C6. The calibrated abstract. Here is a research abstract's closing sentence, overstated: "Our method solves the long-standing problem of real-time fraud detection and will transform how the industry operates." The actual contribution: a measurable improvement (you choose a plausible number) on one benchmark, under specific conditions, not yet tested in production. Rewrite the sentence so a peer reviewer (Chapter 14) would find it both honest and worth publishing—calibrated, specific, and still making a real claim to novelty.

C7. The honest changelog / status update. A deployment introduced a bug that lost some users' unsaved work for about 40 minutes before rollback. Write the two-to-four-sentence user-facing notice (Chapter 21, Chapter 34). It must state plainly what happened, who was affected, what you did, and what (if anything) the reader should do—without either minimizing ("a brief hiccup") or catastrophizing. Then write the dishonest version a nervous PR instinct would produce, and label the specific evasions in it.

Part D — Synthesis and Critical Thinking ⭐⭐⭐

D1. The three-reader translation, with an ethical twist. Take this finding: "A new screening test detects the condition in 95% of cases but produces a false positive in 8% of healthy people." Write the one-paragraph summary three ways—for clinicians, for patients, and for a marketing page—and keep all three honest. Then write a short paragraph on which version is hardest to keep honest, and why. (Hint: the marketing page is where the lie of omission is most tempting.)

D2. Find the flaw. Here is a fluent, well-structured, scannable proposal opening. It is exactly the kind of writing this book taught you to produce—and it is spin. Identify every place clarity is being used to defeat the reader's judgment rather than serve it:

"This initiative represents a transformative opportunity to modernize our platform and unlock significant value. The approach is proven, the timeline is aggressive but achievable, and the ROI is compelling. Industry leaders have already moved in this direction. We strongly recommend approval to avoid falling behind."

D3. The Challenger lesson, three ways. In Chapter 4 the Challenger case taught structure; in Chapter 9 it taught data display; in this chapter it teaches ethics. In one paragraph each (and using only the verifiable record), explain what the same case teaches from each of the three angles—and then, in a final paragraph, explain why the three lessons are ultimately one lesson.

D4. Where is your line? Write a one-page reflection (this is for you, not for grading): name three specific things you will not write, delete, or sign, no matter the pressure. For each, write the early, in-writing, constructive sentence you would use to raise the concern before the dramatic version became necessary (§38.8). Be concrete—imagine the actual document and the actual ask.

D5. Accuracy in both directions. The chapter insists accuracy runs two ways: don't overstate, and don't understate a real finding into invisibility. Find (or write) one passage that commits each error on the same underlying result—a strong finding inflated past what the evidence licenses, and the same finding so over-hedged the reader can't tell anything was found. Then write the single calibrated version between them, and explain in one or two sentences why both errors are misrepresentations of the same evidence.

Self-assessment rubric (D1–D4): Strong responses (a) keep every version honest while genuinely adapting to the audience (D1); (b) name specific words that carry the spin, not just "it's vague" (D2); (c) cite only the verifiable record and resist inventing detail (D3); (d) produce concrete, usable language, not abstract principle (D4). A response that confuses "honest" with "unpersuasive" has missed the chapter: the honest versions should be more trustworthy, not weaker.

Part M — Mixed Practice (Interleaved) ⭐⭐–⭐⭐⭐

These problems mix this chapter with earlier ones. Part of the task is choosing which skill the situation calls for.

M1. A README's installation section says "Setup is easy—just run the install script." On a fresh machine the script fails without a dependency that is never mentioned. Is this primarily a tutorial-accuracy problem (Chapter 26), an instructions problem (Chapter 22), or an ethics-of-omission problem (Chapter 38)? Defend your answer, then fix the sentence.

M2. A data memo's chart is accurate but its caption reads "Figure 2. Revenue by quarter." The memo's whole argument depends on the reader noticing a 23% Q4 drop. Apply Chapter 9 (interpretive captions) and Chapter 38 (the obligation to make the decision-relevant fact unmissable). Write the caption.

M3. An executive summary (Chapter 20) leads with a confident recommendation but omits that the supporting pilot failed. Which chapters are in tension here, and which obligation resolves the tension? Rewrite the opening two sentences.

M4. A peer-review comment (Chapter 12 / Chapter 34) says: "The conclusions are overstated given the sample size, and the funding source isn't disclosed." Rewrite the authors' overstated sentence into a calibrated one and draft the one-line conflict disclosure they should add.

M5. A blog post for a general audience (Chapter 28) explains a new study with a vivid analogy. The analogy is clear but subtly overstates what the study found. Diagnose the collision between Chapter 28 (use analogy) and Chapter 38 (don't overstate), and propose a fix that keeps the analogy honest.

M6. A status page during an outage reads: "We are investigating reports of an issue affecting some users." Internally, the team knows the cause and that all users in one region are fully down. Name the obligation being violated and rewrite the message honestly without causing needless panic.

Part E — Extension ⭐⭐⭐⭐ (optional; Deep Dive track)

E1. The "reasonable reader" audit. Take a real public document—a privacy policy, a drug leaflet, a release note, a financial disclosure. Apply the §38.6 "reasonable reader" standard: what would a competent member of the intended audience need in order to judge, and does the document provide it? Write a 300–400-word critique identifying one accuracy issue, one transparency issue, and one accessibility/inclusivity issue—citing only what the document actually says.

E2. The ethics of the persuasion you've already learned. Pick one earlier chapter whose techniques are the most double-edged (candidates: Chapter 4 BLUF, Chapter 9 interpretive captions, Chapter 20 executive summaries, Chapter 30 assertion-evidence slides). Write a short essay: how could a dishonest writer abuse this exact technique, and what is the single check that keeps it honest? This is the chapter's thesis applied to the book's own toolkit.

Selected solutions and rubrics: appendices/answers-to-selected.md. For the open-ended writing tasks (Part C, D, E), use the rubrics above for self-assessment; the test in every case is whether the whole document leaves a true impression a reasonable reader could rely on.