How to Organize Research Papers So You Actually Come Back to Them

You remember the sentence. Not the paper, just the sentence.

Something about the gain disappearing once the context gets too long. You read it in the spring. It changed how you thought about your design.

You open your Downloads folder. 2403.11234v2.pdf. 2403.11234v2 (1).pdf. paper_final.pdf. Forty more like them.

You check your bookmarks. A folder called Read later, last opened in March. You search your notes app. You wrote "good paper, check section 3", and nothing else.

Forty minutes later you find it, by reading three papers again from the top.

Saving a paper is easy. Coming back to it is the hard part.

Most advice on how to organize research papers is about filing: pick folders, name your files, choose a system.

But nobody forgets where they saved a paper. They forget what was in it, and why it mattered.

So organizing research papers was never really about storage. It's about finding the one sentence again, with what you thought about it still attached.

The usual ways to organize research papers, and where they break

Most people end up with a mix of these. Each one works on the day you save. Each one breaks on the day you come back.

The Downloads folder. Fast to save, impossible to search. arXiv file names are numbers, and the number tells you nothing six months later.
Browser tabs and bookmarks. A bookmark keeps the link, not the reading. And the tab you meant to read stays open until the day the browser crashes.
A tidy folder tree. One paper about retrieval and cost has to live in one folder. You'll look for it in the other one.
A notes doc next to the PDFs. Your thoughts in one place, the words they were about in another. "See section 3" points at nothing once you've forgotten which paper.
A reference manager. Excellent at what it's built for: citing papers in what you write. More on that below.

The pattern is the same every time. The paper is kept. Your thinking about it isn't.

What a paper library you'll come back to needs

Strip it back and a research paper library only has to do four things:

Save in one click, so saving never waits for later
Keep your notes on the exact words they're about, not in a separate file
Let you tag as you go, instead of designing a folder tree first
Search papers and notes together, so you don't need to remember which paper it was

And because the moment you remember a paper is rarely the moment you're at your desk, it should come with you.

That's what we built the iReading Room Library to do. Here's how each part works.

Save arXiv papers in one click with the Chrome extension

Install the iReading Room extension for Chrome from the Chrome Web Store. Then open any arXiv paper: the abstract page, the HTML version or the PDF.

A small box appears in the corner: Add this paper to your iReading Room library?

Click Add to library. iReading Room opens in a new tab and adds the paper's official PDF to your Library, with its real title, authors and date. Ignore the box and it disappears after a few seconds. That's it.

The extension only does what you would do by hand. It makes no network requests of its own and runs only on arXiv paper pages and your own Library.

Not on arXiv? In your Library, press Add article, paste the link and press Add to Library:

an arXiv link becomes arXiv's official PDF
a link to a PDF is stored as it is
any other web page is printed to a PDF

Everything becomes a PDF on purpose. A link can change or disappear. A PDF on your shelf reads the same in a year.

And if you follow arXiv with Floodgate, the papers you keep from your daily list land on the same shelf with one click. We wrote about that routine in How to Keep Up With arXiv.

How to annotate an arXiv PDF: select the sentence, write the thought

Open a paper from your Library and it opens in the reader. The text is real text, not a picture of it, so you can select it the way you'd select any sentence.

Drag across the passage that matters. A ＋ Note button appears beside it. Click it and write.

The note keeps the quote and the exact spot on the page. It shows as a highlight every time you open the paper, and opening the note takes you straight back to those words, already highlighted.

Not every thought is about one sentence, so a note can be about one of three things:

A passage: pinned to the words you selected
This article: your verdict on the whole paper
General: a thought about your own work that this paper started

One habit makes all the difference: write what you thought, not what the paper says. The paper already says it.

"Beyond roughly 64k tokens the gain is noise." Weak note.

"This is the number our retrieval budget should be sized against. Check it against our long-context eval before the rewrite." Strong note.

Six months from now, the second one is the reason you'll open this paper again. If you want a quick way to decide what's worth a note at all, start with the five lines in How to Read a Research Paper as a Software Developer.

Tag as you go, not up front

Folders make you decide where a paper belongs before you've read it. Tags let you decide after, and let it belong in two places at once.

While you write a note, type a tag. If it doesn't exist yet, press Create and it does. The same tags work on whole papers and on single notes, so one tag can bring back both.

A few kinds of tags that hold up well:

the project, like search-rewrite
the method, like retrieval or long-context
a status you'll act on, like to-try or cited-in-design-doc

Pick tags for the question you'll ask later, not for the shelf you'd put it on. "What did we decide to try?" is a better tag than "NLP".

You can also link a note to any other note you've written, in this paper or in one you read months ago. Two labs agreeing. A result that contradicts the one you kept. The argument stops living in your head and starts living in your library. More on that on the note-taking page.

Find it again: search your papers and your notes together

This is the moment all the rest is for.

On the Search page, one box searches your papers and your notes at the same time. You don't have to remember which paper the sentence was in. You only have to remember a word you wrote, or a word you highlighted.

Or pick a tag, or several, and choose whether results must match any of them or all of them. Papers and notes come back side by side.

The Library itself helps with the paper you were reading last week:

Search by title or author across everything you've kept
Recently updated puts the paper you were arguing with on Friday back on top on Monday; Recently added shows what's new
Shelves that count themselves: all articles, web articles, and papers you kept from feeds
On a wide screen, Jump back in keeps the three papers you had open last at the top

Each paper's card shows how many notes it carries, so you can see what you've already worked through before you open it.

Remember the sentence about the context length? One search for "context" finds the note you wrote, and the note opens the paper on the highlighted line. Forty minutes becomes ten seconds.

Take your paper library to your phone

You remember papers in odd places: on the train, in a meeting, five minutes before a call.

iReading Room installs on your phone straight from the browser, with no app store. On an iPhone, open the app in Safari, tap Share, then Add to Home Screen. On Android, open it in Chrome and tap Install. It opens full screen from your home screen, on your Library.

It's the same library and the same reader, with every highlight you made at your desk. The mobile app page walks through the install in under ten seconds.

Is iReading Room a Zotero alternative?

For some people, yes. For others, it's something to use alongside it. Here's an honest way to tell.

Zotero is excellent if you write papers. It's free and open source. It collects papers from your browser, has its own PDF reader with highlights, and, above all, puts citations and a bibliography into Word, Google Docs or LaTeX. If your output is a paper with a reference list, keep Zotero. iReading Room doesn't format citations.

iReading Room is built for people who read research to build something: a product, a model, a system, a team decision. For them, the output isn't a bibliography. It's the note that says what to try next.

What changes when that's your job:

Papers come to you. Floodgate reads each day's new arXiv papers against a plain description of your project and puts the few that touch it on top. Your library fills with papers you'd otherwise have missed. Here's how that works.
Your notes are the point, anchored to the words, tagged and searchable with the papers.
Your AI agent can use it. If you work with Claude Code, Codex or Cursor, it can fetch your recommended papers through our MCP server.

Your notes stay yours. From Settings, you can export your whole library, every paper and every note on it, as one file.

A ten-minute habit that keeps a library alive

A library stays useful only if saving into it costs nothing and coming back to it pays. Here's the whole habit:

Save the paper the moment it looks useful: one click on arXiv
Write one note in your own words on the sentence that made you save it
Add one or two tags, for the question you'll ask later
Once a week, sort by Recently updated and close the loop: try it, cite it, or let it go

That's how you organize research papers without ever organizing them. No folder tree, no filing session on Sunday, no forty-minute search.

A library is what you can find again

Everyone keeps more papers than they read. That's fine. The papers were never the problem.

The problem is the sentence you can't find, and the thought you had about it that's gone.

Keep the two together, and your library starts working like a memory instead of a pile.

Save your next paper in one click. Start your library.