Guides
Explanatory essay guide for science and technology topics
A working guide to writing science and technology explainers that give readers a usable mental model, honest numbers, clear context, and stated limits.
What to take away
- Write the explanation target as one sentencewhat a curious nonspecialist should be able to explain afterward.
- Lead with the model readers need, then define terms as they do work.
- Give every number a unit, a baseline, a time frame, and a source.
- Name the type of uncertainty and say what it changes about the claim.
- Ask an expert for factual errors and a target reader to explain the model back.
A good explainer hands the reader a model they can use. They finish able to say what the thing is, how its parts relate, what evidence supports the account, where it applies, and what is still unknown.
Simplicity is not the removal of hard facts. It is the removal of avoidable work.
Set the explanation target
Write one sentence completing this frame:
Audience Need vs Useful Depth
Audience need
- Understand a headline
- Definition, scale, evidence, uncertainty
- Compare products
- Mechanism, conditions, performance, tradeoffs
- Assess a policy
- Boundaries, distributions, alternatives, evidence
- Learn a technical process
- Parts, sequence, inputs, outputs, failures
Useful depth
- Understand a headline
- Compare products
- Assess a policy
- Learn a technical process
After reading, a curious nonspecialist should be able to explain...
"Quantum computing" is a subject. "Why a quantum computer does not test every answer at once" is an explanation target. The narrower version names the misconception and the model needed to correct it.
Define the audience by prior knowledge and by the decision they face, not by age or a label like "general public."
| Audience need | Useful depth |
|---|---|
| Understand a headline | Definition, scale, central evidence, uncertainty |
| Compare products | Mechanism, conditions, performance measure, tradeoffs |
| Assess a policy | System boundaries, distributions, alternatives, evidence quality |
| Learn a technical process | Parts, sequence, inputs, outputs, failure states |
Establish the main message early
Lead with the answer or model readers need, then define terms and build support. Do not open with a century of history unless the history explains the current mechanism.
Opening Sequence for Explainers
- What it is
- Why the reader is asking now
- The main explanatory model
- The first limit or condition
The National Academies' discussion of communicating science effectively notes that findings may be provisional, limited to particular contexts or populations, or unsettled where audiences want a firm answer. It recommends presenting numbers, reducing unnecessary inference, explaining unfamiliar quantities, and drawing attention to what matters most.
Use a short opening sequence:
- What it is.
- Why the reader is asking now.
- The main explanatory model.
- The first limit or condition.
Choose the right explanation type
Definition
Draws boundaries around a term and separates it from nearby concepts.
Process
Shows how something changes through stages.
Mechanism
Explains how organized parts interact to produce an outcome.
Causal
Explains why an outcome occurred and weighs rival causes.
System
Maps flows, feedback, constraints, and interactions across a boundary.
Many essays need more than one type. Name the primary job so side paths do not displace it.
Build from known to new
Introduce one new relationship at a time. Define a technical term immediately before it does work. Repeat the exact term when a synonym would imply a different concept.
Mechanism Explanation Order
- Input or trigger
- Relevant parts
- Interaction sequence
- Output
- Feedback or control
- Common failure
- Conditions that change the model
For a mechanism, use this order:
- input or trigger;
- relevant parts;
- interaction sequence;
- output;
- feedback or control;
- common failure;
- conditions under which the model changes.
Sequence is not causation. A stage list says what follows what; a mechanism explains why the transition occurs.
Treat numbers as claims
Every number should answer:
Number Interrogation Checklist
- What was measured or estimated?
- In which unit?
- For which population or boundary?
- Over what period?
- Compared with what?
- With which uncertainty or range?
- Typical, maximum, median, modeled, or observed?
- What was measured or estimated?
- In which unit?
- For which population, system, or boundary?
- Over what period?
- Compared with what?
- With which uncertainty or range?
- Is it typical, maximum, median, modeled, or observed?
Translate scale only after preserving the original quantity. An analogy such as "enough electricity for 1,000 homes" depends on a stated household-use assumption and period.
Explain uncertainty by source
Different uncertainty calls for different language:
- measurement uncertainty;
- sampling variation;
- model assumptions;
- unknown future conditions;
- disagreement among studies;
- incomplete knowledge of a mechanism;
- natural variability.
Say what the uncertainty changes. Does it affect the direction of the result, its size, its application to one group, or only the final decimal place?
Use analogies with an exit
State the correspondence and the break.
A cell membrane can be compared to a selective boundary because both regulate passage. Unlike a staffed checkpoint, its selectivity arises from molecular properties and gradients, not decisions.
Drop an analogy when explaining its exceptions takes longer than explaining the mechanism directly.
Pair words and visuals
A diagram should expose relation, sequence, scale, or comparison. Label boundaries, direction, units, and omitted parts. The caption should say what the visual shows and what it does not.
A decorative scientific image is not evidence. A microscope photograph without scale, method, stain, and sample identity cannot support a quantitative claim.
Run expert and reader checks
Ask a subject expert to mark factual errors, missing conditions, and misleading simplifications. Ask a target reader to explain the model back without prompts. These checks answer different questions.
The federal plain language guides treat testing as a stage of its own, beside writing and design, on the principle set by the Plain Writing Act of 2010 that content for the public is written for its specific audience. Testing answers the one question rereading cannot: whether this audience understood it.
Revise when either person supplies a plausible but wrong model. Clarity includes preventing predictable misunderstanding.
Common questions
How much jargon is acceptable?
Use the terms readers need to think accurately. Define each one at first use, and cut status language that does no work.
Should uncertainty appear near the end?
Place each limit beside the claim it affects. Summarize the main uncertainty near the conclusion only if the piece needs it.
Are analogies necessary?
No. A labeled sequence, a worked example, a diagram, or a direct definition may explain more cleanly.
Can an explainer make an argument?
Yes, but keep the explanatory model separate from any policy or value judgment built on it.
In this guide
- How to explain a technical mechanism, with a bicycle derailleur as the modelChoose a definition, process, causal or systems explanation by the reader's question, then explain a rear derailleur with a boundary, labelled verbs and a failure test.
