How to Write a Literature Note

Braindump

The objective is the contract

A literature note answers one stated question. The question is written before any other prose — it is the contract the rest of the note has to honour. A reader scanning the note must be able to tell, in one line, what the note is going to settle and what it is not. Without that contract, the note becomes a dump: there is no test for whether a given paragraph belongs.

The objective also bounds the work. A note finishes when its question is answered, not when the source has been read in full. Material that is interesting but off-topic earns its own note with its own objective; appending it here would dilute the contract this note is going to be audited against. Hereafter, audit refers to the act of reading the note carefully and checking it against each of the rules that follow.

The reader’s flow is sacred

The reader must understand what we are talking about, the chain of though. Only claiming obscure statement without a clear (but concise) explanation of why it helps reaching the objective is as good as no claim at all.

Neutrality is key

We don’t care about your opinion or what you think you know. The document must contain a neutral point of view.

Neutrality starts upstream of the prose, in the formulation of the objective itself. A question — “what does the spec require for X?” — admits any answer the source supports. A thesis dressed as a question — “why does the spec require X?” — has already chosen the answer, and every step downstream inherits the choice: what is read carefully, what is quoted, how the quote is paraphrased. The independent reader does not catch this, because they can only check the quotes the author selected, and the author selected to confirm. The objective is therefore phrased as an open question, and any framing that presupposes its own answer is rewritten before the source is opened.

The chain runs from objective to answer

Sections follow the path the reader needs to take from the question at the top to the conclusion at the bottom. Each one moves the argument by exactly one step: it introduces the next concept the answer depends on, or rules out a candidate the objective could have been confused with. A section that does neither is noise, regardless of how interesting it is in its own right.

The order of sections is not the order in which the reading of the source happened. The note is rewritten, after the source is digested, so the chain is the shortest path from objective to answer. If a detour was needed during research and is not needed in the final argument, it is dropped — its absence in the chain is the audit. A reader who has to reconstruct the argument from a meandering note is doing the work the note was supposed to do.

The reader’s flow is sacred

The reader should glide through the prose uninterrupted, glancing at a footnote whenever they want to verify a claim against its source. Everything else — detours, meta-commentary, secondary justifications — is noise and must be removed. If something is valuable but breaks the flow — a side comparison, a self-contained derivation whose result the argument will consume — move it to its own note and link to it from the main chain at the moment its result is used.

Brevity serves flow. The longer a sentence, the more its message dilutes — attention is narrow, and a convoluted clause already costs a glance back. Where a bulleted breakdown carries the idea more directly than a paragraph, prefer it.

Flow is more than absence of friction. The reader should feel the writer’s pull toward the answer, eager to find out how the question resolves. Style matters: a note that is technically rigorous but emotionally flat loses readers as surely as one that breaks flow with detours.

Put a lot of links — to other notes, to the canonical source URL of a cited text, to the matching footnote. If the reader wants to follow a concept referred to and described somewhere else, they must be able to simply click and go there.

A new term enters the document where the document has just laid out what motivates it — not earlier. In a literature note this rule is acute, because the source vocabulary is dense and indexical: article numbers, case identifiers, doctrinal labels, regulatory acronyms. Dropping a section title like « 666 plutôt que 653 » on a reader who has not yet been told what each one says, or why one might apply rather than the other, forces them to suspend reading and chase the footnote before they can parse the heading itself. Flow is broken on the section title.

The fix is almost always reordering — introduce the grounding sentence first, then the term that names what was just established. A section title that names a question — « quelle présomption s’applique au grillage ? » — pulls the reader in instead of asking them to recognize an identifier they have not yet met. Once the body has laid the ground, the article number can arrive in a sentence the reader can already parse: « deux présomptions distinctes coexistent — l’une pour les murs (art. 653), l’autre pour les autres clôtures (art. 666) ».

When the structure constrains the sequence, defer the reference itself: name a more general concept the reader already has, and let the specific identifier arrive when its time comes. The point is not to hide identifiers but to ensure each one lands on a reader who can receive it.

This applies everywhere the reader’s eyes go — section titles, body sentences, footnote labels, link anchors. The reading flow doesn’t care whether a name sits in a heading or a paragraph; an ungrounded name breaks it wherever it appears.

More broadly, a literature note is a story, not an inventory. Stakes (the open question), tension building toward a result (the chain of arguments), a payoff at the end (the answer, or a documented unresolved disagreement) — these are what carry rigorous content through. Substance without narrative rarely lands; the writer’s pull is what makes the citations count.

Every claim is sourced or absent

Statements about the world — what a spec says, what a system does, what a method requires — carry a footnote with the quote that supports them. A claim without a footnote is the author’s inference, and is written as such (“it follows that…”, “this implies…”) so the reader can tell distilled judgement from cited fact. Anything that is neither sourced nor explicitly inferred is a guess and does not belong in the note.

This rule is what makes the note durable. A future reader — most often a future-you — who doubts a line resolves the doubt by reading the footnote, not by re-reading the source. A note whose claims are unsourced is a note whose claims have to be re-derived every time they are doubted, which is the same as having no note.

Footnotes carry the verbatim quote, not a paraphrase

The footnote contains the original passage, the original wording, and a link to the source. It does not summarise. A summary in the footnote re-introduces the failure mode the citation was meant to prevent: the reader has no way to tell whether the body’s claim follows from the source or from the author’s compression of the source. Verbatim is what makes the note auditable in isolation, without re-fetching the document.

The body’s prose is allowed — and encouraged — to compress and re-express. The contract is that any compression in the body is backed, immediately, by the un-compressed text in the footnote. When the source is later revised and the wording shifts, the diff between body and footnote becomes visible at the next read.

Off-topic is exclusion, not appendix

Material that does not move the chain from objective to answer is cut, even when it is correct, even when it was hard-won during the reading. A literature note is not a record of what was read; it is the distilled answer to one question. A reader who scanned the objective and is looking for the answer must not have to skip past unrelated paragraphs to find it.

The cut material is not lost: it earns its own note, with its own objective, and is linked from any chain where it is actually load- bearing. This is also how the note set grows without each individual note becoming a swamp. A note that tries to hold every adjacent fact ends up holding none of them well, and the link graph between notes — which is where the value compounds — never forms.

Source quality is part of the claim

A footnote pointing at the primary source — the specification, the standard, the maintainer’s own documentation, the source code of the system under discussion, or a blog post or forum answer from a well-known and respected actor of the domain — supports a claim differently than a footnote pointing at a forum thread or a blog post by an unknown author. Both are citations; only one is evidence. The note prefers the primary source whenever it exists and is reachable, and names it as such: the RFC for a protocol, the official documentation for a product’s guarantees, the published paper for a result.

Secondary sources are admissible only when they interpret a primary source the note cannot host — paywalled, deprecated, untranslated — and they are named as secondary, with the reason the primary is not cited directly. Anonymous forum posts, undated blog entries, screen- shots without provenance, and AI-generated summaries are not citations: they are unsourced claims wearing the costume of one, and they do not survive the audit. A note built on them collapses the moment the sourcing chain is followed.

The channel matters as much as the source. Each intermediary between the note and the source — a search engine, a reformatter, a third-party mirror, a cache — is a place where wording can drift, an outdated version can be served, or the wrong document can be returned. The note prefers the channel with the fewest intermediaries: ideally a fetch addressed by the source’s own canonical identifier, with no search step in between.

A dedicated channel that the source itself exposes — an official API, an authenticated documentation endpoint, a structured archive addressable by a canonical identifier (Legifrance for French legal texts, the official documentation API for a product, the published archive for a specification; for an agent, an MCP tool is the typical instantiation) — is the channel of choice whenever one exists. It returns the text the source itself publishes, addressed by its canonical identifier, and returns the canonical URL alongside it, so the footnote points at exactly the document the quote came from. A verbatim quote produced this way is the strongest form of citation the discipline can produce. Web fetches and search engines are the fallback, used only when no such channel covers the source, and they inherit a higher burden of scepticism: the URL is verified, the version date is read, and the text is compared against any independent reference before it is trusted as the primary it claims to be.

Independent verification before the claim is used

Citation is not verification. A claim can be sourced, the source can be primary, the verbatim quote can sit in the footnote — and still the body’s wording can drift from what the quote actually says. The author who wrote the body and selected the quote is the worst-positioned reader to catch the drift: their attention is bent by the conclusion they were heading towards, and the same bias that picked the quote is the bias that reads it as supporting more than it does. Before the claim is relied on, an independent reader — no context, no stake in the conclusion — is given the quote and the body’s claim side-by-side and asked whether the claim is what the quote supports. The note advances only on that reader’s confirmation.

The check is mandatory for the claims the rest of the chain depends on, not for every line. The failure mode it catches — wishful interpretation, selective quoting, a verbatim slice that contradicts its own surrounding context — is precisely the failure mode a self-review will not catch, because it is the failure mode of self-review. Outsourcing the read to someone with no investment in the conclusion is what makes the citation step actually adversarial, and adversarial is what “verified” has to mean here. For an agent, a neutral subagent is the typical instantiation of that independent reader; for a human writer, a peer or a sceptical colleague.

The same principle extends to the conclusion the chain produces. Before the note closes, the author tries to break the conclusion: what evidence would falsify it, what alternative reading fits the same sources equally well, where the chain could have gone wrong. The instruction to the adversarial reader is not “find flaws in the presentation” but “find reasons the conclusion is wrong.” Failing to contradict the conclusion after a genuine attempt is itself evidence: if the counter-argument cannot be made — opposing sources are weak, alternative readings forced, objections dissolved by the same citations that support the conclusion — the conclusion is stronger for having survived. A conclusion that has not been challenged is not confirmed; one that has been challenged and held is. The note records the attempt: what objection was raised and why it did not hold, so a later reader who finds a counter-argument can see exactly where the adversarial pass stopped.

One pass is rarely enough. Each correction the adversarial reader prompts is itself a new draft, and the new draft inherits none of the assurance the previous one earned: it has not yet been read by an independent reader. The cycle — adversarial read, correction, adversarial read again — runs until a pass returns no objection that holds. Stopping earlier confuses no objection raised yet with no objection survives; only the second is the contract. The note is stable when adversarial pressure stops moving it, not when the author runs out of patience. Iteration is not a sign that the earlier draft was poor — it is the mechanism by which the note earns the confidence the discipline promises.

Doubt is recorded, not hedged

When the sources themselves disagree, the note states the disagreement and cites each side. Hedging in the prose — “some people say”, “it is generally accepted”, “I believe” — is the failure mode this whole discipline exists to prevent: it lets an unsourced claim wear the costume of a sourced one. If a position cannot be cited, it is named as the author’s inference; if it can, the citation replaces the hedge.

A note that ends in unresolved disagreement is still a useful note, provided the disagreement is the answer the objective reached. The reader is left with two cited positions and the reason the question is open, which is the state of the world at the time of writing — far more useful than a confidently single answer that papers over the conflict and has to be unwound later.

Permalink