To print: the essay guide on two pages · the one-page checklist · the parts, in order

The task

You choose one of six titles that the IB publishes for your exam session, about six months before the deadline. You write an essay of at most 1600 words that explores it. The examiner asks one question of your essay: Does the student provide a clear, coherent and critical exploration of the essay title?¹

• Copy the title exactly, as the title of your essay. A reworded title is marked against the original, and the difference costs relevance.
• Discuss the areas of knowledge the title names. If it says “the arts and the human sciences,” those two; if it says “two areas of knowledge,” you choose two. Not three.
• The 1600 words include quotations. They exclude footnotes and references, but footnotes may not carry argument. The examiner stops reading at word 1601.
• You meet your teacher three times, recorded on the TK/PPF form: to discuss the titles, to discuss your plan, and to get comments on one full draft. The form does not affect your mark, but it is submitted with your essay.

The rubric’s top band reads: the discussion “has a sustained focus on the title and is linked effectively to areas of knowledge. Arguments are clear, coherent and effectively supported by specific examples. The implications of arguments are considered. There is clear awareness and evaluation of different points of view.” Everything below is about how to meet those five sentences.

1. Read the title before you answer it

The key-term swap is the most common reason a competent, well-organised essay ends up at 4 or 5. The title’s key term is quietly replaced by something easier, and the essay answers the easier question. It takes several forms:

• An easier neighbour. Reliability becomes usefulness; power becomes accessibility. But a tool can be useful and unreliable, so the essay now answers a different question.
• A literal misreading. A title asking whether “all things are numbers” is answered with evidence that things can be labelled with numbers (colour codes, letter positions). Everything is true, and none of it answers the claim that number is what things are.
• Several meanings at once. Flawed observation covers imprecise instruments, deliberate fraud and propaganda in one essay. With three meanings, no argument can build up.
• Stretched until nothing counts against it. Doubt is widened until every counterexample fits inside it, so the thesis says nothing.
• A bar set impossibly high. Justification is defined as “beyond all doubt,” so the answer “no” follows from the definition, not from the argument.

The test. Write down your working definition of the title’s key term. Then name one example that would come out the other way under that definition. If you cannot, your definition has already decided the answer, and no evidence, structure or polish will rescue the essay.

Do this before you write anything else, and bring the result to your second meeting with your teacher. A swap made in the first paragraph usually survives into the final version unless someone catches it there.

Some other questions to ask the title:

• Does it ask whether or how much? “To what extent” asks for a degree and the conditions that change it, not “yes” or “no.”
• Does it ask what is, or what should be? A title about whether something is justified is not answered by showing that people in fact believe it.
• What does the quoted claim assume? A title built on a quotation asks you to test the claim, not to illustrate it.
• What does the command word ask? “Discuss” does not mean “describe”: something is at stake, and you have to find the tension. “To what extent” and “Do you agree” ask for a judgement.
• Which words are doing the work? In In the production of knowledge, does it matter that observation is an essential but flawed tool?, three words carry the title: matter, essential and flawed. An essay that only shows observation is flawed has answered a third of it.

2. Plan by positions, not by examples

Most middle-band essays have the same structure: tell an example for a paragraph, add a one-sentence moral (“This shows that…”), move to the next example. Call it the case-file structure. Its variants include an essay with eight examples and no argument between them; an essay that marches through the headings of the knowledge framework (“The scope of history involves…”); and an essay that starts with no introduction and states its thesis only in the last sentence.

Top-band essays are organised around positions: competing readings of the title, competing criteria, or a conditional thesis tested in several places. The examples are recruited to serve the positions.

The outline test. Look at your plan.

• If it reads example 1, example 2, example 3, it predicts a 4 or 5.
• If it reads reading A, reading B, what decides between them, it can reach 8.
• If it reads list all pro, list all contra, I have more pro so I argue pro, it predicts a descriptive essay.

A strong thesis is often conditional: it says when or to what extent the answer is yes, and why. For example: “interpretation produces reliable knowledge to the extent that a discipline’s methods let other people check it.” That one sentence gives a mechanism, applies to both areas of knowledge, and can be tested against every example.

3. Pair your cases and find what separates them

This is the most reliable single move from 5 to 8. Inside one area of knowledge, put a case where the method worked next to a case where it failed, and name the variable that separates them. Then look for the same variable in your second area of knowledge.

The course has given you pairs like these already. Millikan kept only the oil-drop measurements that fitted his expectation and was right; Ehrenhaft kept every measurement and was wrong (Natural Sciences, Lessons 2 and 4). Wegener’s continental drift was correct and rationally rejected for forty years; the Higgs boson was predicted, searched for and found (Natural Sciences, Lesson 5). Kahneman and Tversky’s findings on judgement under uncertainty have largely survived the replication crisis; the social-priming research built on top of them has not (Human Sciences, Lesson 2).

The detail in a paired case is not decoration. In the Wegener case, the missing mechanism is why the rejection was rational, and the arrival of seafloor spreading is why it ended; that detail is the variable at work. If you can delete a detail and the argument still stands, the detail was narration.

4. Decide: balance is not an answer

The commonest structural trap: one example for and one against in each area of knowledge, and the symmetry offered as the conclusion. The essay never asks which side of its own structure wins, or under what conditions. Some related habits:

• Counterclaims dismissed by decree. A “however” paragraph that ends “the thesis is confirmed.”
• Ping-pong between imagined people. “A scientist would say… an artist would argue…” and no verdict.
• Agreeing with everyone. Concluding that both views should be used, without deciding the case where they conflict.
• Leaving instead of resolving. “This opens up a wider discussion,” and the discussion never comes.
• Asking instead of answering. In the Oxford companion’s words, “a raft of rhetorical questions is not a substitute for analysis.”²
• Choosing areas of knowledge because they disagree. If you pick one area that says yes and one that says no, the essay writes itself and decides nothing. Similarities between areas of knowledge are as interesting as differences, and you should explain where a difference comes from.

The decision sentence. Somewhere in your essay there must be a sentence that asks which of these is right, and why? and answers it with something other than “both,” “balance” or “it depends.” You have written this sentence in every lesson of this course: commit before we compare. If you cannot find it in your draft, the essay is not finished.

Deciding does not mean being one-sided. A conditional answer (“yes, as long as…”, “more in history than in the arts, because…”) is a decision. Henly and Sprague even note that a clear essay that weighs both sides can outscore “a one-sided, convoluted, unclear or poorly supported” one.³ What loses marks is weighing without ever stating the result.

In drafts, circle every verdict that has no measure attached (“doesn’t matter,” “matters to a great extent,” “the ideal is balance”) and write by what measure? beside it.

5. Give the other side its best case

The rubric asks for “awareness and evaluation of different points of view.” Awareness is easy: mention an objection. Evaluation means you state the objection in its strongest form, ideally in the words of someone who actually holds it, and then show why your position survives, or how it has to change.

Name real thinkers (Popper, Kuhn, Oreskes, a named historian) instead of inventing opponents. And do not escape a strong rival by a definition: if your thesis says doubt is not central to science, you cannot answer Popper by defining central as “present at the beginning.” A rival dismissed by a definition has not been answered.

6. Say what follows

The rubric names implications in its top band, and they are the part of the essay most often left out. This is the cheapest band-level gain in the essay.

Write one paragraph that begins: If this is right, then… What follows for the knower, for the discipline, or for public disputes about knowledge? If your thesis is that historical interpretation is reliable to the extent that sources can be checked against each other, it follows that whole periods with one surviving kind of source (one empire’s records, one side’s letters) are known less reliably than our textbooks suggest. That is an implication. It also checks your essay: you cannot write this paragraph until you have actually reached a conclusion.

7. Examples: what the example decides

“Be more specific” is the most common piece of essay feedback and one of the least useful. An essay can have specific, well-chosen, properly sourced examples and still sit at 5. Two things matter more than specificity.

How much you analyse compared with how much you narrate. In middle-band essays the example takes five to ten sentences and the point drawn from it takes one (“This shows… / This demonstrates… / This illustrates…”). That is a caption, not analysis. Count your sentences on each side. Sue Bastian gives a simple rule: analysis “should be at least half of the text in the paragraph.”⁴ Sprague says the same from the other side: “Examples are opportunities to illustrate a point, but they are not, by themselves, explanations of a point.”

Whether the example could be swapped. Replace it with another case of the same kind. If the paragraph still works unchanged, the example illustrates; it does not argue. The better question to ask of each example: what does this example decide that another example would not?

Three more rules:

• No invented examples. “A historian could investigate…” or “if someone measured…” is a placeholder, not evidence. A real example has a name, a date and a source.
• Avoid the worn-out examples. Examiners read thousands of essays. “People used to believe the Earth was flat” (educated people have not believed this for over two thousand years), “Galileo against the Church” told as good against evil, and “the Nazis were bad” all signal that you have not looked for your own case.
• Use your own experience where it helps. Your own perspective as a knower counts: something you saw in your own lab work, your own translation, your own sport. One personal example, analysed, is worth more than a famous one retold.

8. Stay inside the area of knowledge

Essays often drift out of the area of knowledge the title names: a mathematics section that is really about technology, a natural-sciences section that is really about engineering, three areas of knowledge where the title asks for two. The relabelling test: could you change the section heading from “mathematics” to “technology,” or from “the natural sciences” to “applied ethics,” without changing a word? Then you are not discussing the area of knowledge.

9. Sources set a floor, not a ceiling

Poor sources hold an essay down; good sources do not lift it. A bibliography made only of study-guide websites and encyclopedia summaries signals a weak essay. But a bibliography of journal articles cannot raise an essay above 5 if sections 1 to 8 are not addressed. Treat sources as a hygiene check, and spend your time on the argument.

• Check every fact in the example that carries your argument. A misstated law, a garbled formula or a discredited dataset costs most when it sits inside the case that is meant to prove the point.
• Write suggests where you mean suggests. “Proves” is almost never true in TOK, and examiners notice.
• Quote only what the source says. Paraphrase inside quotation marks is a misquotation.
• Clean up. Remove tracking tags (utm_source=…), stray links and draft notes. See “Using AI tools” in The Two Assessments.

10. Introduction and conclusion

Write the introduction last, when you know what the essay argues. It does three things: says how you read the title (including your working definition), states your answer, and shows the route. Avoid dictionary definitions: they suggest that a dictionary settles a question that the essay exists to explore.

Do not use the first paragraph to warm up. Two tests help. Your thesis should read naturally straight after the title. And if you write the main claim of each paragraph on a card and read the cards in order, they should answer the title on their own.⁵ Signposts must say how one idea relates to the next (because, but only if, this fails when). “Moreover” and “Another point is” only announce a list.

The conclusion answers the title in the title’s own words, and then gives your implications if they are not already in the body. It introduces no new examples, and it does not apologise (“this essay could have used more examples”): that tells the examiner the essay is unfinished.

A worked example: from caption to argument

Title: In the production of knowledge, does it matter that observation is an essential but flawed tool? Discuss with reference to the natural sciences and one other area of knowledge.

The caption version (the kind of paragraph that sits in the 4–5 band):

Robert Millikan measured the charge of the electron between 1909 and 1913 by watching charged oil drops in an electric field. Later, the historian Gerald Holton found that Millikan’s notebooks recorded 175 measurements, but he only published 58. He left out the ones that did not fit what he expected. Millikan won the Nobel Prize in 1923, and his value was close to the modern one. This shows that observation is flawed but can still lead to knowledge.

Five sentences tell the story; one sentence, the last, draws a point, and the point would fit any example of a scientist who was right.

The argued version:

The Millikan case shows that a flawed observation does not matter in the way the title suggests, because what decides the outcome is not the observation but the judgement about which observations to trust. Millikan published 58 of his 175 measurements and discarded the rest as faulty runs; his rival Felix Ehrenhaft published everything and announced “sub-electrons” that do not exist. If keeping every observation were the safeguard, Ehrenhaft would have been right. He was not. What separates the two men is not how carefully they looked but the theory each brought to his measurements. This makes the flaw matter more, not less: Millikan’s judgement was vindicated only later, by independent measurement, and at the time nothing distinguished his trust in his expectation from Ehrenhaft’s trust in his data. In the natural sciences, then, flawed observation matters to the extent that no independent check is yet available — which is exactly the condition under which the community has to decide.

The second paragraph uses the same facts. It pairs two cases, extracts a variable, tests a tempting view (keep every observation) and rejects it, and ends with a conditional claim that the next section can test in another area of knowledge.⁶

The three meetings

Bring something written to each one.

1. Choosing the title. Bring all six titles, each with two or three lines: your first reading of the key term, and one possible answer. Choose the title where your answer is most interesting, not the one where you know most facts.
2. The plan. Bring your working definition, the example that would come out the other way, your thesis, and an outline in positions. This is the meeting that matters most: almost every fault in this guide is visible in a plan.
3. The draft. You get comments on one full draft. Then make the changes. Comments on a draft are only useful if they reach the final version. Before you upload, go through them one by one and point to the sentence where you dealt with each.

From 8 to 10

If your draft is already strong, these four things usually separate a good essay from an excellent one.

1. Give the strongest real opponent its best advocate, then evaluate it. Do not escape it by a definition.
2. Test your own criterion with one hard case, where your yardstick gives the wrong answer or where the method you trust still failed (the replication crisis is a good place to look), and then resolve it.
3. Argue your implications. A full paragraph, not a closing sentence.
4. Be precise everywhere. No garbled sentences, no “proves” for “suggests,” no careless bibliography. At this level, carelessness is the difference between 8 and 9.