The Researcher's Restart: A Systematic Reading and Writing Plan for New Researchers

id="where-you-actually-are-framing-the-first-six-months">Where you actually are: framing the first six months

Most new researchers do not fail because they cannot read. They fail because they read for six months and then discover that reading is not writing. A literature does not get smaller when you consume it. It becomes a room you are standing in, with walls you cannot see past.

This guide assumes you are in the first six months of a project, or that you are restarting one after a break, a failed funding round, or two years away from it. You have a broad area, a supervisor, and a deadline somewhere in the distance. You probably do not have a research question yet. Most people do not, and pretending otherwise wastes months.

The first six months are for narrowing, not producing

The standard advice to spend a year reading is bad advice for anyone with a funding review, an annual review, or a supervisor who asks what you did this month. The first six months have one job: turn a broad area into a question small enough to answer in a year.

Reading is not the job. Reading is the tool that does the narrowing. If you cannot tell what a given paper changed for you, you are not narrowing anything. That single test — what did this change — is what separates a research reading system from a pile of PDFs and guilt.

A reading list without a question is just a queue of guilt.

Write down what you already believe

Before you open a single database, spend twenty minutes writing half a page about what you currently believe about the field. What is the central disagreement? What do your colleagues argue about at coffee? What is the finding that everybody mentions and nobody explains?

This takes twenty minutes and it builds the most useful reading filter you will ever own. Once you hold a position, you can read for it, and reading for a position is roughly three times faster than reading in order to be informed.

Keep the half-page. Put a date on it and paste it at the top of your notes. Read it again at week six and again at week twelve. The distance between what you believed in January and what you believe in June is your first research finding, even if you never publish it.

Three things to have at the end of month six

1. A one-page scoping memo. The question in one sentence, why it matters in three sentences, and a paragraph on what already exists that you have checked.
2. A triage list of forty to sixty papers. Every paper you looked at, each with a one-line verdict: core, useful, background, or irrelevant.
3. A two-page proposal. The question, the method, the data, the timeline, and the two risks you think will kill it.

Do them in that order. You cannot write a defensible proposal before you know what has already been tried, and you cannot triage before you have a question to triage against.

A six-month map

Month	Focus	You should have	Hours per week
1	Orientation	Half-page belief note, 25 triage papers, supervisor meeting 2	20
2	Scope	Scoping memo, triage list at 60, three candidate questions	22
3	Focus	One question chosen, 15 core papers read properly	25
4	Method	Two-page proposal, data and method sources identified	25
5	Evidence	Evidence ledger with 25 mapped claims, notes on one pilot dataset	28
6	Write	Introduction draft, 1,200 words, supervisor meeting 8	30

A note on the projects that go wrong

Some projects will not have a question by month six, and that is a supervision problem rather than a personal failure. The signal is that your supervisor asks what you did, and you cannot answer in one sentence. If that happens three times in a row, book a longer meeting and bring your scoping memo and your triage list. Bring evidence of effort, not an apology.

The other common wrong turn is a supervisor who has already decided the question and expects you to confirm it. That can be a good arrangement. It can also be a bad one. You will know which within the first two months, and the difference in your output over a year is enormous.

id="finding-a-direction-in-a-literature-you-cannot-read">Finding a direction in a literature you cannot read

The problem is not that you have too little to read. It is that you have no defensible way to choose. Open a review article and you get a list of everything; open a citation database and you get ten thousand results. Neither is a direction.

A direction is a question narrow enough that you can name the paper that would answer it, and broad enough that you have twenty candidate papers to read.

Start from the citation graph, not the reading pile

Every field has a small set of papers that everyone cites, and a small set of papers that everyone actually read. These are rarely the same set. The classic mistakes are obvious and, for a decade, unfalsifiable. The live, contested, closely contested literature is what you want.

Three legitimate ways into a field

The citation chase. Take the most recent review in your area. Do not read it. Open its reference list and find the papers from the last three years, then follow their references backwards one step. You will find the working set quickly.

The author chase. Find the three or four groups publishing in your area and read whatever they posted in the last eighteen months. Not because those papers are best, but because you will learn how the field argues.

The problem chase. Search for the phrases practitioners use for the thing they cannot do. Methodological papers in good fields often start with a complaint about a real workflow.

Entry point	Speed	Depth	Best for
Citation chase	Fast, about 2 hours	Low to medium	A broad area you know nothing about
Author chase	Medium, about a week	Medium	Learning how a field argues internally
Problem chase	Slow, about 2 weeks	High	A field with strong applied or clinical work
Keyword search	Very fast, minutes	Very low	Finding one specific paper, nothing else

The scoping question test

A candidate question passes if you can answer all four of these without hesitation:

1. Falsifiable. Could a result come out against it?
2. Bounded. Can you name the population, corpus, dataset, or period?
3. Contested. Do at least two credible people disagree about the answer?
4. Reachable. Do you have, or can you plausibly get, the data?

If a question fails on the third test, it is a settled topic and you will be writing a literature review for nobody. If it fails on the second, it will expand until you drown.

If you cannot name the paper that would answer your question, you do not have a question yet.

A ninety-minute narrowing procedure

1. Write three candidate questions, each in one sentence, in a text file. Time-box: fifteen minutes.
2. For each question, name the single dataset, corpus, or participant group you would need. If you cannot name it, cross the question out.
3. Search each question's key terms in a general academic index. Count the results. A question with more than two thousand hits is too broad; one with fewer than fifty is probably worked out or hidden in a discipline you have not opened.
4. Read the abstract and conclusion of the three most recent papers for each surviving question. Twenty minutes per question.
5. Score each question from one to five on interest, feasibility, and whether you would still care about it in a year. Add the scores.
6. Take the top question and write the sentence you would put at the top of your first paper. If you cannot, take the second one.

When your topic is genuinely new

If the literature is thin, that is an opportunity with a trap in it. Thin literature usually means one of three things: the field is young, the field is tiny, or the work exists but is not in English or not indexed where you are looking.

Check the third option first. Search the phrase in a full-text index, not a metadata index. Look at the conference proceedings. Ask a librarian, specifically for grey literature and non-English work. Talk to two people who have been in the field longer than you. A surprising share of thin literatures are actually medium-sized literatures hidden behind terminology differences.

If the field really is young, your review becomes the contribution, and you should plan for it explicitly rather than hoping to smuggle it inside an empirical paper.

id="a-reading-system-triage-notes-and-the-three-pass-method">A reading system: triage, notes and the three-pass method

A reading system is not about reading more. It is about deciding, quickly, what deserves your attention, and then giving that attention without interruption. Most people have the second half and no first half, which is why they end up with eighty open PDFs and a vague sense of having worked.

Pass zero: triage before you commit

Triage means deciding what a paper is without reading it. Ten minutes maximum per paper: title, abstract, introduction's first paragraph, the conclusion, the references list, and a skim of the figures or tables. Then stop.

Signal	What it tells you	Verdict
Cited in the last 24 months	Active area, someone is building on it	Read properly
Cited only in older work	Settled, or abandoned	Background only
Author is a leader in your candidate group	Sets the agenda	Read properly
Same method, different dataset	Directly comparable to your plan	Read carefully
You already know the answer	Low information gain	Skip
No abstract, no conclusion, only slides	Probably a talk, not evidence	Skip
Open access, 40 pages, thesis	Might be a literature review you need	Skim the contents page

Do not delete what you triage. Move it into a second file with the verdict. That file is your triage table, and it is the first thing you show a supervisor who asks what you have done.

The three-pass method

Pass	Time per paper	What you read	What you write
1: Survey	15 to 20 minutes	Abstract, intro, figures, conclusion	One sentence: what did this change?
2: Argument	45 to 60 minutes	Full text, methods and results included	Three claims, each marked agree, dispute, or unclear
3: Deep	2 to 3 hours	Everything, plus supplement and data	A note per claim, with page numbers

1. Pass one, survey. Read front to back at speed but do not skip the figures. The figures usually contain the actual finding. Write exactly one sentence. If you cannot write it, you have not understood it, and one more minute is worth more than an hour of re-reading later.
2. Pass two, argument. Now read properly, including methods and results. Write three claims. Label each one agree, dispute, or unclear. Only papers with at least one dispute are worth a pass three. Agreeing papers are for citations, not for the argument.
3. Pass three, deep. Read the whole thing including appendices. Take a note per claim, with page numbers, direct quotes for anything you will paraphrase, and the exact numbers from tables. This is the pass that produces quotable material.

If you have written nothing after reading a paper, you have not read it. You have looked at it.

If you only have thirty minutes

Situation	Do this	Skip
Triage a new batch of 10 papers	3 minutes each: abstract, conclusion, verdict	Methods, references
Prepare a supervisor meeting	Re-read your five core notes, 6 minutes each	New papers
Recover from a lost week	Read one core paper at pass 3	Everything else
Write a paragraph on a topic you know cold	Read 2 anchor papers for quotes	The literature
Build a claim	Read 1 paper that disagrees with you	9 that agree

What to do with the papers you will never read

Most of your triage list should end up unread in full, and that is the correct outcome. A research reading system is mostly a machine for discarding.

But do not discard silently. Keep a single file called unread-and-why, with one line per paper. This matters for two reasons. When you write, you will need to say what you considered and rejected, and a reviewer will ask whether you missed something obvious. And on the rare occasion when a reviewer says you ignored the key paper, you can point to a dated list showing you triaged it on the fourth of March and judged it irrelevant, which is a much better position than silence.

Budget it honestly: aim to finish pass three on five to eight papers per month. Not thirty. Twenty-five full readings in a year is already a serious amount of work.

id="building-the-argument-notes-claims-and-evidence">Building the argument: notes, claims and evidence

Notes are where reading becomes writing. Most people's notes are a mess of quotes that they will never use, stored in a folder that grows by 40 documents a month. The fix is small and mechanical: stop taking notes about papers and start taking notes about claims.

One note per claim, not per paper

A paper usually contains three to seven claims that matter to you. Note each one separately. A claim note has the claim in your own words, the source, the exact location, the evidence type, and your verdict. That is enough to build a paragraph six weeks later when you have forgotten everything.

Field	What goes in it	Example
Claim	The assertion, in your own words	Rural clinics delay referral more than urban ones
Source	Author, year, short title	Osei 2024, Referral pathways
Location	Page, section, figure	p. 112, Fig 3
Evidence type	Data, review, argument, case study	Survey of 214 clinics, 2021 to 2023
Verdict	Agree, dispute, unclear, need data	Dispute — their sample excludes mobile clinics
Use	Where it might appear	Methods paragraph, gap statement
Quote	Exact words, with page	Delay reached a median of 11 days

Write the claim in your own words even when you are going to quote it. Paraphrasing first tells you whether you understood it; quoting first hides the fact that you did not.

The evidence ledger

Keep one table of every claim that will appear in your paper, with a status. This is the document that tells you whether you can write the Results section on Thursday.

1. List every claim your paper will make, one per row. Include the ones you believe but cannot yet support.
2. For each row, record the best source you have found so far, or the word none.
3. Mark the status: supported, weak, missing, or contested.
4. At the start of every writing session, check the status column before you type anything.
5. If more than three claims are marked missing, you are not writing yet. Go and find the evidence.

The evidence ledger is the difference between writing an argument and producing a sequence of sentences that once appeared near each other.

Argue with yourself

The most useful note is the one that records where the literature disagrees with you. Set up a separate file called disputes. Every time a paper contradicts your working position, write both claims side by side and state which one you believe and why.

This file will do two things. It gives you the shape of your contribution, because a paper that agrees with everyone is a translation, not a contribution. And it gives you the raw material for the last paragraph of your introduction, where you state what this study adds. That paragraph is usually where weak papers die in review, and it is much easier to write in week eleven if you built it in week three.

A practical weekly note budget

Twenty papers triaged, five papers read at pass two, and six to ten claim notes written per week is a strong, sustainable rate for a first project. It will produce about 250 claim notes in a year, which is more than enough for a paper.

If you are writing fewer than five notes a week, you are reading without processing. If you are writing more than twenty, you are writing notes instead of reading. Both are fine occasionally. Neither should be your default for a month.

id="zotero-obsidian-or-a-folder-tools-that-fit-how-you-think">Zotero, Obsidian, or a folder: tools that fit how you think

Tool choice matters far less than most advice suggests. What matters is that your notes are structured enough to find later, and that the tool does not require more maintenance than the work it supports.

Aspect	Zotero	Obsidian	Plain folders
Best for	Reference management and citation output	Thinking and connecting claims	People with no patience for setup
Setup time	30 to 60 minutes	1 to 2 hours	10 minutes
Citation handling	Built in, BibTeX and Word plugins	Manual or community plugins	Manual, error-prone
PDF annotation	Strong, with note templates	Good, but you manage it	Depends on your reader
Linking claims across papers	Weak	Strong, backlinks and graph	Not possible
Sync between machines	Built in, 300 MB free	Usually paid	Manual, if you use a cloud drive
Risk	You spend a week tagging instead of reading	You build a beautiful system and read nothing	You lose links, quotes get lost in text files
Monthly upkeep	20 minutes	60 to 90 minutes	10 minutes

Choose by thinking style

If your main problem is finding citations again, use Zotero. If your main problem is developing an argument across papers, use Obsidian. If your main problem is that any software at all gets in the way, use folders and a single running notes file, and accept that you will need to search more.

The worst outcomes I see are: a Zotero collection with 400 items and no notes, and an Obsidian vault with 300 linked notes and no paper. Both are procrastination with good tooling.

A tool that takes two hours per week to maintain is a tool that has taken two hours per week from your research.

If you use Zotero

1. Install Zotero and the Zotero Connector for your browser, so one click saves the PDF and the metadata.
2. Create three collections: To Read, Read, and Evidence. Move papers between them as you go. Do not build deeper than three levels.
3. Set up a note template with the fields from the claim note table: claim, source, location, evidence type, verdict, use, quote.
4. Set Zotero to pull the best available metadata and to link files by DOI where it can.
5. Install the Zotero plugin for your word processor, and never type a reference by hand again.
6. Once a week, run a search for items with no note and either note them or delete them.

If you use Obsidian

1. Create a vault with one folder per project, and inside it four files: claims, papers, disputes, and evidence ledger.
2. Use one note per paper, and one note per claim inside it, so backlinks work in both directions.
3. Turn on backlinks and use tags sparingly: one tag for status, one for method, one for discipline. Resist adding a fourth.
4. Do not install plugins in the first month. Use what is there. Every plugin is a small new project competing with the real one.
5. Write a weekly note on Fridays with three lines: what I read, what I concluded, what I will read next. This becomes your progress report to your supervisor, automatically.

If you use plain folders

This still works well, and it works better than a complicated system you abandon.

1. One folder per paper, named Author_Year_Title, containing the PDF and a notes.txt file.
2. One claims.md file at the top level, where each claim note lives with its source and location.
3. One evidence.md table with the ledger.
4. Use Dropbox, Google Drive, or OneDrive for sync, and check version history before overwriting anything.
5. Back up the whole folder monthly to an external drive or a second cloud. One corrupted file during a deadline is enough to lose a week.

Whichever you pick, write down the choice and the reason in your notes file. In eight months you will otherwise reconsider it for no reason, and that reconsideration is a task with no end.

id="open-access-how-to-find-the-papers-you-cannot-afford">Open access: how to find the papers you cannot afford

Paywalls are not an obstacle to reading; they are a routing problem. Most of what you need is legally reachable with about three minutes of extra effort per paper. Some of it is genuinely free, some is free on request, and some you may have to pay for or request through your library.

Where the free full text actually is

| Source | What it holds | How a researcher uses it | | --- | --- | | DOAB | Peer-reviewed academic books from scholarly publishers, open access | Finds an open full text of a monograph or chapter when the publisher site is paywalled | | DOAJ | A vetted list of open access journals | Checks that a journal is legitimate before submitting, and finds journals with no submission fee | | PubMed Central | The free full-text archive of biomedical and life-science literature | Searches past abstracts into actual full text for medicine, biology, and health | | Zenodo | A general repository of datasets, preprints, theses, software, and other research outputs | Finds and deposits grey literature, including work that was never formally published | | OpenAlex | An open catalogue of papers, authors, institutions, and citations | Maps the citation graph of a topic and finds recent work, not only the classics | | Europeana | Digitised cultural heritage from European museums, libraries, and archives | Finds primary sources, images, maps, and archival material for the humanities | | arXiv | Open preprints in physics, mathematics, computer science, and quantitative biology | Reads the newest work before peer review, and posts your own preprints | | Wikisource | Public-domain and freely licensed texts, including speeches, letters, and historical works | Provides primary texts for history, literature, and citation | | Internet Archive | Scanned books, journals, audio, and video from libraries and archives | Reaches out-of-print and pre-1929 works that no publisher still sells | | Project Gutenberg | Over seventy thousand public-domain ebooks, fully searchable | Gets clean full text of older works whose copyright has expired |

Getting a specific paper, legally, in five steps

1. Find the DOI. Search the title in OpenAlex, Crossref, or Google Scholar. If there is no DOI, look for the publisher page.
2. Try the DOI resolver first. If the publisher has a deposit, you land on the PDF or a landing page that states availability.
3. Try Unpaywall. Paste the DOI into Unpaywall's API or its website. It returns every legal open copy, including preprint and repository versions.
4. Search the repository layer. Check CORE, your institutional library repository, the subject repository for your field, and Google Scholar's versions link, which often includes a copy the publisher does not link to.
5. Ask the author. Most authors reply. One email with the DOI and a polite request takes two minutes, and authors reply more often than people expect, especially to early-career researchers.

Every paywalled paper has a legal route. You just have to try more than one.

When you must pay, or must not

Some material is genuinely restricted: recent paywalled journal articles with no open version, most books from commercial presses, and paywalled databases behind institutional subscriptions. For these, the honest routes are: your university library's access and document delivery service, interlibrary loan, a library outside your institution, an open-access mandate the authors' funders impose, and asking your supervisor for a departmental budget for purchasing a handful of key papers.

Do not use unauthorised sharing services. Beyond the legal risk, it makes you a small part of a system that punishes authors and libraries, and it is the fastest way to end an academic career before it starts. If you find yourself tempted, check step five above again. The author will usually help.

Ask your library for the things you cannot find

Librarians are the most underused resource in research, and their job is to help you. Ask specifically for: a literature search on your question, a Web of Science or Scopus author search, a systematic search on a named database, and a print collection scan for pre-digital books. Libraries often hold print runs of journals that are not digitised anywhere, and a scan of six consecutive issues can resolve a question that no search engine can.

Set a target: two library consultations in your first six months, thirty minutes each, brought with a specific question. It is the highest return on time available to you.

id="writing-your-first-paper-the-structure-that-works">Writing your first paper: the structure that works

A first paper feels impossible because you are trying to write it in one pass, from nothing, in one document. It is not impossible. It is a sequence of small documents, each with a clear job, written in a known order.

IMRaD, with word budgets

Section	Target words	Drafting time	The job of this section
Title	12 to 15 words	20 minutes	Name the finding, not the topic
Abstract	200 to 250 words	90 minutes	Question, method, result, implication
Introduction	600 to 900 words	2 to 3 hours	Move the reader from their problem to your method
Literature review	600 to 1,000 words	3 to 5 hours	Show the gap, using your disputes file
Methods	700 to 1,200 words	3 to 4 hours	Make it replicable by a stranger
Results	500 to 900 words	3 to 5 hours	Findings in order of importance, not order of analysis
Discussion	600 to 900 words	2 to 3 hours	Interpret, compare, limit, recommend
References	30 to 60 items	60 minutes	Generated, never typed

These totals put a realistic first paper at 3,700 to 6,000 words, which is normal for a journal article. If your target journal wants 8,000, add depth to the discussion and the literature review, not padding to the introduction.

Outline first, prose second

1. Write the title as a claim, even provisionally. If you cannot write a title that says what you found, you may not have a finding yet.
2. Write one sentence for each section, stating what it will argue. If a section needs two sentences, it is two sections.
3. Write the claim list for the results section from your evidence ledger. Order by importance, not by the order you ran the analysis.
4. Write the Methods section as a numbered procedure someone else could follow. Include the awkward details: exclusions, software versions, dates, and what you did about missing data.
5. Draft each section separately, in separate files. Give each a filename and a word target.
6. Assemble them. Read the assembled draft once as a whole, without editing, and note where you got bored. Boredom marks structural problems.
7. Write the abstract last, from the finished draft, in one sitting.

Nobody has ever written a good paper by starting at the beginning.

The introduction is the hardest eight hundred words

The introduction does four jobs in order, and it fails when it tries to do them out of order. First, establish that the problem matters, in language a non-specialist in your subfield would understand. Second, show what is already known, briefly, by citing rather than describing. Third, name the gap precisely, in one sentence, ideally using the disputes file. Fourth, state what you did and what you found.

A useful test: your fourth job, the one-sentence statement of what you found, should be written before the rest of the introduction and then be the paragraph the reader reaches at the end of it.

Write badly on purpose, then edit

The single biggest obstacle for new writers is trying to write well in the first pass. Separate the two. In draft one, you are allowed to be inaccurate about which word comes next, but not allowed to stop. In draft two, you revise for argument: does each paragraph make one point, and does the point follow from the one before? In draft three, you revise for language: cut every sentence that does not carry weight.

Track each pass with a word count. If the count goes down by more than ten percent in a pass, you are cutting, which is correct and healthy. Most first drafts of 6,000 words should become 4,500.

id="the-weekly-rhythm-of-a-researcher">The weekly rhythm of a researcher

Research is done in blocks and ruined by interruption. A schedule built around a weekly rhythm survives bad days; a schedule built around daily inspiration does not.

The week as the unit of planning

Block	Days	Duration	What happens
Deep reading	Monday, Tuesday	3 to 4 hours each	Pass two reading, no email, no phone
Writing	Wednesday, Thursday	3 to 4 hours each	Draft one section, using claim notes only
Admin and teaching	Friday morning	3 hours	Emails, admin, preparing teaching
Consolidation	Friday afternoon	2 hours	Update ledger, update triage, plan next week
Buffer	Weekend	1 to 3 hours	Reading, rest, or catching up
Supervisor contact	Weekly or fortnightly	30 to 45 minutes	One written update plus one meeting

The Friday afternoon block is the one people skip, and it is the one that makes the rest work. Friday consolidation means Monday morning starts from a written plan rather than from memory.

Five daily habits

1. Open with the previous day's last sentence. When you stop, leave a note to yourself about what comes next. Do not close the laptop mid-paragraph and try to reconstruct it on Tuesday.
2. One review card. On a card or in a pinned note, keep three lines: today I will, today I read, today I am avoiding. The third line is the useful one.
3. Read before you write, not after. Twenty minutes of reading gives the writing session somewhere to start.
4. Set a timer and stop at the bell. A four-hour block that ends at three hours and forty-five minutes is better than a six-hour block of visible but unfocused effort.
5. Record one sentence in the daily log. What happened, in the past tense, one line. Friday's update to your supervisor is then ten minutes of work rather than an hour of reconstructing.

The researcher who writes every day at four o'clock beats the researcher who writes perfectly on Sundays.

Protect two hours that nobody can take

Pick a two-hour block that is genuinely yours. Not the two hours after work, when you are tired and the house is loud. The best candidates are early morning before other people wake, or a block you can take in a library or room booked in advance.

Tell two people what that block is and ask them not to book you into it. Then defend it. If the block is consistently overrun by teaching preparation, admin, or a colleague's requests, the problem is not your time management. It is that nobody has been told that the block exists.

Weekly review: fifteen minutes on Friday

Answer four questions in writing:

• How many papers did I read properly this week, and which one changed my mind?
• What did I write down, and can I see it?
• What is blocked, and who could unblock it?
• What will next week's deep-reading block cover?

If the first answer is zero for two consecutive weeks, you are not doing research. Change the schedule before you change the topic.

id="working-with-your-supervisor">Working with your supervisor

Your supervisor is a resource with finite time and their own pressures. Almost all supervisor friction comes from one of two mistakes: turning up without a document, or bringing a topic instead of a decision.

Bring a document, not a topic

Every meeting should have a one-page document you send at least four hours in advance. It should contain the meeting's date and a three-line agenda, a progress list covering what you said you would do and whether you did it, a decision list showing what you need decided and your recommended option, a problem list giving a specific obstacle and the two options you have already considered, and a brief notes section with anything you want in writing rather than in memory.

Send it by email, not a chat message. Email creates a record, allows them to think before replying, and can be read at 11pm on a Friday.

Meeting type	Length	Focus	Your output afterwards
Regular, weekly	30 minutes	Decisions and blockers	A one-line summary you send before the next meeting
Direction, monthly	60 minutes	Question, method, scope	Updated scoping memo
Draft review	60 to 90 minutes	Specific sections, line by line	Written list of changes with deadlines
Crisis, as needed	45 minutes	What is going wrong	An agreed recovery plan with dates

Decide what you need from each meeting

Before the meeting, write one sentence: what do I need to leave this meeting with? Sometimes it is a decision. Sometimes it is information only they have. Sometimes it is permission to stop pursuing something. Write that sentence at the top of your agenda. It converts a vague conversation into a transaction, and it stops the meeting ending with you both agreeing to think about it.

When you disagree

Disagreement with a supervisor is normal and often productive. The failure modes are avoidance and capitulation. Avoidance means you nod at a plan you know is wrong. Capitulation means you agree publicly and then quietly do something else.

Both are worse than a clear disagreement, because both destroy trust in the one thing you need: their ability to advise you. The workable version is to bring evidence rather than doubt. Write two lines: what I think, and the specific evidence that made me think it. If they disagree and you are not persuaded, say so, and record what you will do. Follow it for a month, then bring new evidence.

Bring the disagreement on the page. An unrecorded disagreement becomes a grievance nobody can answer.

Managing when they are unavailable

If your supervisor is away for two weeks or more, or for the first three months because of travel, illness, or a heavy teaching load, you need a second reader. Ask your department for a co-supervisor or a mentor. This is normal, not a complaint about the first supervisor. Say it as: I would like someone to read my first draft by the fifteenth, so that I get feedback while it still matters.

You will also want a peer group. Three or four other early-career researchers who meet for one hour every second week to present work in progress is worth more than most writing groups and costs nothing. Set a rule: each meeting, one person presents and the others ask questions for twenty minutes.

id="rejected-revised-and-resubmitted-what-to-do-with-a-rejection">Rejected, revised and resubmitted: what to do with a rejection

A rejection is a normal event in research. Journals receive hundreds of submissions and publish tens. The variable is not the decision but what happens in the two weeks after it.

What a rejection actually tells you

Desk rejection, decided without review, usually means one of three things: the topic does not fit the journal, the abstract and title did not communicate the contribution, or the submission was malformed. The first is a redirect. The second is a fix. The third is a formatting error. None of them is a judgement of your career.

Rejection after review is different and more useful. The reviews contain, buried under the politeness, the specific reason your paper did not get published: an unclear contribution, a flawed method, a weak analysis, an overclaimed conclusion, or literature that is inadequate. You can act on all five.

What the reviewer said	What it usually means	What you do
Not novel enough	The gap you claim is not real or not important	Sharpen the contribution, cut the descriptive sections, add the comparison that makes it new
Method unclear	A reader could not follow your procedure	Rewrite Methods as a numbered list, add the parameters they asked for
Sample too small	Power is insufficient for the claim	Add a power analysis, soften the claim, or reframe as a pilot
Claims exceed data	Overreach	Walk the claims back to what the data supports and say so explicitly
Literature inadequate	You missed the field	Go back to the triage list, read the papers you skipped, and cite them properly
Poorly written	The argument cannot be followed	Cut 20 percent, then restructure the paragraphs around one point each

The two-week response procedure

1. Wait three days. Read the reviews once, write down everything that stung, and do not reply to anyone.
2. Separate the three lists. Write what is genuinely fatal, what is fixable, and what is a misunderstanding or a preference. Most reviewers' comments fall into the third list, and conflating the three is what makes people give up.
3. Read the reviews a second time, as a stranger. A reviewer who has been in your field for thirty years knows things you do not. If two independent reviewers mention the same gap, it is a gap.
4. Decide the journal. Despair is a bad reason to accept a bad fit. If the top journal rejects you, submit to the second journal on your list within a week, and keep the referee reports attached.
5. Revise before you resubmit. Take three weeks minimum for the revision. Revising with the reports in front of you is a rewrite, not a patch.
6. Answer every point. In your cover letter or response letter, quote each comment and answer it in one to three sentences. Do not argue with reviewers in the letter; a reader who thinks you are combative will not recommend you.
7. Write the thank-you letter. If you were seriously considered, a short thank-you to the editor costs five minutes and keeps your name in their mind for the next round.

Resubmitting to the same journal

Usually a bad idea, with two exceptions. If the rejection was an administrative error or a desk rejection on fit, resubmit after tightening the framing. If the reviews said the work is important but needs revision, you may resubmit after a major rewrite, but check the journal's policy first: some specify a time limit of 90 days, some forbid resubmission without an editor invitation, and most require a response letter.

Do not resubmit to a journal that said not to resubmit. That burns the relationship permanently.

A rejected paper with a written revision plan is worth more than an accepted paper you learned nothing from.

id="twelve-week-plan">Twelve-week plan

This plan assumes you have a question, or are close enough to one to start. It produces a submitted paper or a defensible draft. Twelve weeks at 25 to 30 hours per week is 300 to 360 hours, which is a normal amount of work for a first paper.

Week	Focus	Deliverable	Hours
1	Question	One-sentence question, two-page proposal	28
2	Evidence	Evidence ledger with 20 mapped claims	28
3	Method	Methods drafted as a numbered procedure	26
4	Method	Pilot run, code or protocol debugged	30
5	Analysis	Analysis plan fixed in writing, no changes after this	26
6	Results	Results section drafted, figures made	30
7	Results	Results complete, ledger shows no missing claims	28
8	Intro	Introduction drafted, 800 words	26
9	Literature	Literature review drafted from the disputes file	28
10	Discussion	Discussion drafted, limitations section written	28
11	Full draft	Whole paper assembled, abstract written	30
12	Revise	Supervisor review applied, journal selected, submitted	30

Phase one, weeks 1 to 4: build the spine

The aim is to have a method you can run and a results section you could write, before you spend a single hour on prose. Resist the pull to write the introduction early. Introductions are easier to write when the results exist, and writing one first commits you to a framing you may not be able to support.

At the end of week 4 you should have a fixed analysis plan: the variables, the models, the exclusions, and the decision rules, all in writing. Write the date on it. Do not change it after this point without writing down why, in the same file. Analysis plans that drift mid-analysis are the source of most honest-looking but indefensible papers.

Phase two, weeks 5 to 8: produce the evidence

Run the analysis, make the figures, and write the results as you go, section by section, rather than saving all writing for week 11. Write each result as a claim: what the data show, with the number, and how confident you are. If a result does not match your expectation, write it down plainly and note it in the ledger. Results that contradict expectations are the most interesting thing in a first paper and the thing most often buried.

Target by the end of week 8: a complete results section, and no claim in the ledger marked missing.

Phase three, weeks 9 to 12: assemble, revise, submit

Now write the introduction and literature review, in that order, using the disputes file for the gap statement. Then the discussion, which must contain four things: what the findings mean, how they compare to the literature, the limitations, and what you would do next.

Assemble in week 11, send the full draft to your supervisor with a 10-day deadline, and hold the submission for week 12. Read the reviews carefully, apply what is right, note what you disagree with and why, and submit. Do not submit a paper you have not read end to end at least once, at least three days after finishing it, because the errors you will find are mostly of the kind you can see only after a break.

id="common-mistakes-and-how-to-get-out-of-them">Common mistakes and how to get out of them

These are the ten failures that recur most often in the first year, in roughly the order I see them.

1. Reading without writing. You read 200 papers and can produce no argument. Fix: stop at pass one for anything that is not core, and write one sentence per paper. If you cannot write the sentence, you have not understood it and the paper needs a second pass.
2. The tool rabbit hole. You spend three weeks configuring Zotero collections or Obsidian plugins. Fix: write down the minimum setup, do it in one 90-minute session, and forbid yourself from changing it for six months. The tool is never the problem.
3. The topic too big. Your question covers a whole field. Fix: apply the scoping question test, and if it fails on bounded, pick one population, one region, one period, or one outcome.
4. Waiting for a perfect dataset. You are designing your study around data that might arrive. Fix: start with what you already have, write the limitation honestly, and treat the ideal dataset as the discussion's future work.
5. The supervisor avoidance spiral. You delay sending updates because you have not done enough. Fix: send the update anyway, honestly. A short honest update is worth more than a silence that hides a problem for six weeks.
6. Perfect prose, no argument. Your paragraphs are beautiful and say nothing new. Fix: write one claim per paragraph, and delete any paragraph where you cannot state the claim in a single sentence.
7. Ignoring the 80 percent. You spend all your time on three papers. Fix: triage ruthlessly and accept that most of your reading list will only ever be a citation, and possibly not even that.
8. Never asking anyone to read it. You write for months without showing a human. Fix: put a draft in front of a peer in week 6, not week 14. Every project that stalled late had a peer reading available earlier.
9. Treating a rejection as a verdict on you. Fix: re-read the reports, make the three lists, and act on the fixable list within two weeks.
10. Starting the methods too late. You cannot write the discussion without the results. Fix: reverse the order of work: method, analysis, results, literature, introduction, abstract.

The most common mistake is treating research as a series of big decisions. It is a series of small ones, taken weekly.

Getting out of the worst one

If you read that list and recognised yourself in six items, do not change your topic. Change your system for two weeks: one triage session, two deep reading blocks, one writing block, one Friday consolidation, every week, no exceptions, and nothing else changed.

Two weeks of that will do more than six months of thinking about it. Most stalled projects are not stalled because the question is wrong. They are stalled because nothing has moved for weeks and the researcher has started believing the project is the problem.

id="final-checklist">Final checklist

Use this at the end of every week. It should take fifteen minutes.

1. My question fits on one line, and I could say it out loud without reading it.
2. My triage list is current, and every item has a verdict and a date.
3. I have read at least five papers properly this week, not skimmed.
4. I have written at least five claim notes, each with a location and a verdict.
5. My evidence ledger shows no claim marked missing, or I know which one is and why.
6. My disputes file has at least one entry, and the gap statement in my introduction is drawn from it.
7. Every note I took is in one place, not in my inbox, my bookmarks, and my head.
8. I have written something this week, even if it is 200 words.
9. I have sent my supervisor the weekly update, even when it is mostly bad news.
10. I know what I am reading, writing, or blocked on this week, and it is written down somewhere I will see.
11. I have put at least one paper or dataset into a working repository, not a downloads folder.
12. My files are backed up somewhere other than the machine they live on.
13. I have given at least one hour to something outside research this week.

Before you submit

1. I have read the journal's author instructions, including length limits and reference style.
2. My title states a finding, not a topic.
3. My abstract states question, method, result, and implication in 250 words or fewer.
4. Every claim in my results has a source in my evidence ledger.
5. My limitations section names at least three real limitations.
6. Every reference in my text appears in my reference list, and every reference in the list is cited.
7. My figures are readable in greyscale and carry their own captions.
8. Someone else has read the whole draft, and I have applied their comments.
9. I have read the assembled draft end to end, at least three days after finishing it.
10. I have read the journal's policies on data availability, ethics approval, and preprints.

One last thing

None of this is about working harder. It is about working in a shape that produces output, and about having the output be the thing you actually meant to say.

The reading part is now cheap. Open-access repositories and preprints mean you can see almost any paper in your field within a day of knowing that it exists, and a plain-folder system plus a spreadsheet will carry you through a first paper if you are disciplined about it. The habits above are what turn a broad area into a defensible question and a defensible question into a submitted paper.

For the aggregator side of the work — finding open-access books, public-domain texts, and genuinely free articles across dozens of libraries rather than one publisher at a time — ANYFREEBOOK at anyfreebook.com indexes those collections in one place, and it is a reasonable first stop when your reading list needs a book rather than an article.

Start the week, not the project. Pick the paper you will read properly tomorrow and put it on your desk.

id="appendix-the-two-week-sprint-to-a-submitted-draft">Appendix: the two-week sprint to a submitted draft

A plan is only useful when the day arrives. This is the compressed version of the whole guide for the fortnight when a deadline has landed and the first draft does not exist yet. It assumes you have a question, an adviser who has approved it, and roughly two working weeks.

Days 1 and 2: the skeleton

Write the paper's structure before you write a sentence of it. One page, eight headings, no prose. In order: the problem, why it matters, what is already known, what is missing, the method, the results, the limits, the conclusion. If a heading has nothing under it, that is the part to work on.

Day	What you produce	Hours
1	The eight-heading skeleton, one page	3
2	The claim in one sentence, and the three sentences that support it	3

The one-sentence claim is the hardest part and the most valuable. If you cannot write it, the project is not ready to be a paper, and finding that out on day two is a gift.

Days 3 to 6: the middle

Now the body, in this order: the problem, the method, the results, the discussion. Write the method while it is fresh and the results while the numbers are on screen. Leave the introduction for last, because the introduction is a promise about what the rest of the paper does, and it cannot be written honestly until the rest of it exists.

Work in two-hour blocks with a hard stop. Put the draft in one file, not six. Keep a running list of every sentence you cannot justify and park it for the revision pass rather than stopping to fix it in the moment.

Days 7 to 9: the figures and tables

Every figure earns its place or it goes. A figure that restates a table is a second copy of work the reader will not do twice. Write the caption as a sentence that would still make sense to somebody who never sees the figure, because that is how it will be read.

Day	What you produce	Hours
7	Figures, with captions written as sentences	4
8	Tables, with every column justified	4
9	The reference list, formatted in the target style	3

Formatting the references is dull and it is worth doing now: it is the single most common reason a submission bounces back, and it takes hours rather than minutes, so it does not belong in the final evening.

Days 10 to 12: the hostile read

Give the draft to someone who has not been inside the project, and ask them three questions: what is the claim, what is the evidence, and what would make you reject it. The third question is the one that matters, and a colleague will usually answer it if you ask for it directly.

Then read it yourself with the abstract and the conclusion covered. If you cannot follow the argument from the middle of the paper to the end, the structure is wrong, and no amount of sentence-level editing will fix it.

Days 13 and 14: the last pass

Three passes, in this order, and no mixing them.

1. Argument pass. Does each section do what the previous section set up? Fix the order here and nowhere else.
2. Evidence pass. Every claim has a citation or a figure. Every citation is actually in the reference list.
3. Language pass. Now the sentences: cut the hedges, cut the throat-clearing, fix the tense, fix the reference formatting one last time.

Then submit a day early, not a day late. A submission that needs a correction after the deadline is worth nothing, and the reason people submit late is that they planned to submit on the last day and discovered on the last day that the form wanted a cover letter.

Day	What you produce	Hours
13	Three passes, in the order above	5
14	The cover letter, the metadata, and the submission	3

A cover letter worth sending is four sentences: what the question is, what you did, what the answer is, and what it means for the field. If you cannot write those four sentences, the paper is not finished, and you have found out with a day to spare.