Exercises — Chapter 2: Audience

Writing is learned by writing. Most of these ask you to produce or revise text, not just pick an answer. Where a task is open-ended, a self-assessment rubric follows so you can judge your own work. Selected solutions live in appendices/answers-to-selected.md.

Difficulty key: ⭐ warm-up · ⭐⭐ core · ⭐⭐⭐ stretch · ⭐⭐⭐⭐ extension.


Part A — Analyze This ⭐

For each, identify the likely audience mismatch (or, where noted, why it works). Name the K-R-A-C variable that's off.

A1. An onboarding email to a brand-new non-technical hire: "Provision your creds via the IdP, then SSO into the VPC bastion to validate connectivity before the standup." — What's wrong, and which variable?

A2. A research abstract for a specialist conference: "In this paper, we present some interesting findings about a thing we studied that might be useful to people who work in this area." — What's wrong, and which variable?

A3. A status-page message after an outage: "Root cause was a null-pointer dereference in the retry handler (commit a3f9c) under connection-pool starvation; reverted in #1851." Audience: customers. — What's wrong, and which variable?

A4. An email to a busy VP that opens: "I wanted to walk you through the full methodology before getting to the results, so you'll understand how we arrived at the recommendation." — What's wrong, and which variable?

A5. A README for an open-source library aimed at developers, opening with: "To install, run pip install ourlib. Quick example below; full API reference and architecture notes further down." — Why does this work? Which variables did the writer get right?

A6. A note to a fellow data scientist that says only: "Churn is up, it's probably the onboarding, trust me." — What's wrong for this audience specifically? (Careful — the problem is the opposite of A1's.)

A7. A safety notice to apartment residents: "Due to elevated particulate readings exceeding the regulatory threshold pursuant to §4.2, occupants are advised to consider evacuation protocols." — What's wrong, and which variable? (Note where the stakes make this especially bad.)

A8. A Slack message to your manager, who will likely forward it to their boss: "the vendor totally screwed us again, classic them, I'll deal with it." — Beyond tone, what audience principle did the writer forget?

How to do Part A well: For each item, don't just say "too technical." Name the specific variable (Knowledge, Role/goal, Action, or Context) that is mis-set, and say who the writing would actually work for. "This is wrong for a non-technical hire because of Knowledge — five undefined terms — but it'd be fine for a fellow sysadmin" is a complete answer. "It's confusing" is not.


Part B — Revise This ⭐⭐

How to do Part B well: Don't just swap a few words. Decide what leads, what to cut, and what to define — then rewrite. After each, name the one variable you optimized for. The point isn't a "correct" sentence; it's a sentence that's unmistakably for the stated reader.

Rewrite each passage for the stated audience. Give the scenario its due — decide what leads, what to cut, what to define.

B1. For a non-technical executive. Original: "We deprecated the v1 endpoint; clients hitting it now receive a 410, so any unmigrated integrations will fail until they adopt the v2 schema before the EOL date." (~1 sentence; lead with what the executive needs.)

B2. For a general-public blog audience. Original: "The algorithm's time complexity improved from O(n²) to O(n log n), reducing latency at scale by roughly an order of magnitude on large inputs." (Use an analogy; spend at most one technical term.)

B3. For a fellow expert (peer engineer), made more precise — this one goes the other direction. Original (too vague for a peer): "We made the database faster by changing some settings and it seems a lot better now." (Add the precision a peer needs to verify and reuse. Invent plausible specifics.)

B4. For a client who is paying you and may forward this to their CEO. Original (too internal): "Your funnel is leaking hard at activation, the numbers are ugly, you need to fix onboarding ASAP." (Keep the honesty; change the register so it survives forwarding.)

B5. For an on-call engineer reading at 3 a.m. during an incident (a runbook entry). Original (too discursive): "If the service starts returning errors, it's often because the cache has gotten into a bad state over time, which can happen for a variety of reasons, and the usual approach people have found helpful is to restart it, though you should also look at the logs." (Make it terse, action-first, imperative.)

B6. For a skeptical, possibly hostile reviewer (e.g., a competitor or an adversarial peer reviewer). Original (overclaims — gives the hostile reader ammunition): "Our method completely solves the problem and clearly outperforms every prior approach in all cases." (Rewrite so an adversary reading line-by-line finds nothing to exploit. Hedge what's genuinely uncertain.)

B7. For a brand-new team member with high jargon fluency but zero project context. Original: "Just use the usual flag when you deploy and it'll pick up the right config like always." (They know the tools; they don't know your conventions. Make the implicit context explicit.)


Part C — Write This ⭐⭐–⭐⭐⭐

C1. ⭐⭐ You discovered that a nightly data export job has been silently failing for three weeks; no data was lost but reports were stale. Write the email to your manager (4–6 sentences). Run K-R-A-C first (manager: needs to know impact and your plan, not the stack trace). Lead with what they need.

C2. ⭐⭐ The audience profile. Pick any technical topic you know well. Write a complete K-R-A-C audience profile (the §2.2 worksheet) for explaining it to a smart 12-year-old. Then write the opening two sentences of that explanation. (This is the smart-friend test made literal.)

C3. ⭐⭐⭐ You're proposing that your team adopt a new tool. Write the opening paragraph two ways: (a) to persuade a decision-maker who controls budget, and (b) to inform a teammate who just wants to know what's changing. Same facts; show how purpose changes the opening.

C4. ⭐⭐⭐ You maintain a popular open-source library. A beginner files an issue: "it doesn't work." You need a short reply that helps without condescending — but you genuinely don't know their skill level. Write the reply (4–6 sentences) using the §2.8 "unknown audience" strategy: default to the least-expert plausible reader, but give an experienced user a fast path too. (This is the layering move at small scale.)


Part D — Synthesis & Critical Thinking ⭐⭐⭐

D1. Translate for four audiences (the marquee task). Here is one technical finding (a composite, fictional-but-realistic):

"Our load testing shows the checkout service degrades non-linearly above 800 concurrent users: p99 latency jumps from 200 ms to 4 seconds, caused by lock contention on the inventory table. Current peak traffic is 600 concurrent users; projected holiday peak is 1,100."

Write four short versions of this finding (3–5 sentences each), one for each audience. For each, first jot the one-line K-R-A-C read, then write the version.

  • (a) A fellow backend engineer who needs to fix it.
  • (b) The VP of Engineering who must decide whether to fund a fix before the holidays.
  • (c) A non-technical customer-success lead who fields complaints and needs to know what to tell customers.
  • (d) A general-public engineering-blog post (assume the fix shipped) explaining what you learned.

After writing all four, answer in 2–3 sentences: which facts appear in all four, and which appear in only one? What does that tell you about the difference between information and audience?

D2. Find the flaw. A colleague defends a dense, jargon-heavy report to executives by saying: "I'm not going to dumb it down. If they don't understand it, that's on them — I'm being precise." Using the chapter, write a 4–6 sentence response that (a) separates "precise" from "appropriate for the audience," and (b) names the cognitive trap your colleague is in.

D3. The Challenger counterfactual. In 3–4 sentences, describe the one-page document the Morton Thiokol engineers needed to put in front of the decision-makers that night. What does it lead with? What's its purpose (from §2.6)? What single visual or sentence makes the conclusion unmissable? (Stay within the verifiable facts: cold temperature, O-ring resilience, the launch decision.)


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

These mix this chapter with Chapter 1. You must decide which idea applies.

M1. ⭐⭐ A peer says: "I already understand my project perfectly. I just need to write it down at the end — the writing part is just transcription." Drawing on both Chapter 1 (writing is thinking) and Chapter 2 (audience), give two distinct reasons they're wrong. (One reason from each chapter.)

M2. ⭐⭐ You're about to write the same lab result up (i) for your lab notebook and (ii) for a course report graded by your professor. Without writing the documents, list: the purpose of each (from §2.6), and one concrete way the two documents will differ because of audience. (Interleaves purpose + audience.)

M3. ⭐⭐⭐ A teammate's draft is technically flawless but their manager "didn't get it." Your teammate concludes the manager isn't smart enough. Diagnose using the chapter's concepts: name the most likely cause, the cognitive trap involved, and two tactics from §2.5 that would have prevented it. Then connect it to Chapter 1's career argument in one sentence.

M4. ⭐⭐⭐ You have ninety seconds before sending an important email. List, in order, the first three things you'd decide using this chapter's framework — and explain why doing this is faster overall than skipping it and rewriting later. (Connects K-R-A-C to Chapter 1's claim that writing is thinking, not wasted time.)


Part P — Peer Review ⭐⭐ (do with a partner, or self-review)

Writing improves fastest when a real reader reacts to it. These work best with a classmate or colleague; if you're solo, do them on your own draft after a 24-hour gap (you'll be a partial stranger to it — see §2.5).

P1. The audience-fit read-aloud. Exchange your answer to C1 (the stale-export email) with a partner. Your partner reads it as the manager would — fast, once — then, without looking back, says aloud: (a) what they think the document is asking them to do, and (b) the one fact they remember. If those don't match what you intended, you have a Role/goal or Action problem. Note where they hesitated; those snags are the curse of knowledge made visible.

P2. The jargon audit, swapped. Trade your answer to D1 (the four-audience load-testing finding). On your partner's public-blog version (d), circle every word a non-technical reader might not know. On their engineer version (a), flag any place they over-explained something an engineer already knows. Give the circles back. Both errors — under- and over-estimating knowledge — are audience failures; this exercise trains you to spot both.

P3. Feedback that helps. When you give feedback in P1–P2, follow one rule: describe your experience as a reader, don't prescribe a fix. Say "I lost the point here" not "rewrite this sentence." The writer owns the fix; you own the reaction. (This is the feedback discipline Chapter 12 builds out — practice it now.)

Receiving feedback: resist defending your draft. If a reader was confused, the reader was confused — that's data, not an insult. "But it's technically correct" is exactly the trap from §2.7. The reader's confusion is the truth about your document; your intention isn't.

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

E1. The tapping experiment, applied. Newton's tapping study showed tappers predicted ~50% success where listeners got ~2.5%. Design a small experiment to measure the curse of knowledge in your own writing: how would you set it up, what would you measure, and what result would convince you that you over-assume reader knowledge? (3–4 sentences; this is a thinking exercise, not a lab proposal.)

E2. The unsolvable multi-audience document. Find a real document that genuinely must serve conflicting audiences at once — examples: a drug label (patient + physician + pharmacist), a tax form (taxpayer + IRS + software), a privacy policy (user + lawyer). Analyze in ~200 words: which audience "wins" in the current design, what it costs the others, and how layering (§2.8) could serve more of them. Connect to the idea that structure can hold multiple readers.


Self-Assessment Rubrics

For Part B (revisions) and Part C/D (produced documents), score yourself:

Criterion ✅ Strong ⚠️ Weak
Audience fit The version clearly couldn't be swapped for a different audience — it's unmistakably for this reader Reads generically; could go to anyone
Right lead Opens with what this reader needs first (recommendation for a decider, method for a peer, hook for the public) Buries the point; opens with throat-clearing or methodology for a non-expert
Jargon handled Terms are kept/defined/replaced deliberately for the reader's knowledge level Undefined jargon for novices, or condescending over-explanation for experts
Content preserved The actual finding survived the translation "Simplified" into vagueness — the point got lost
Forwardable (where relevant) Nothing a secondary reader would weaponize or misread Contains blame, overclaims, or context-free statements

Self-check for the marquee task (D1): A strong answer has four documents that share almost no sentences, even though they describe one finding. If your four versions look similar, you adapted vocabulary but not structure and lead — go back and change what each one opens with.