Who it is for

Turn a reading list into notes you keep

Course material disappears when the term ends: links rot, the module page is archived, the seminar reading moves. Markdown files on your own disk do not, and they open in any editor a decade from now with the source URL still inside them.

What happens after the term ends

Halfway through revision you follow your own bookmark and get a login wall, a redirect to this year’s module, or a plain 404. The reading you did in October is gone and what you have instead is a title and a broken link. A bookmark is a pointer, and a pointer is only as durable as the thing it points at – which for university pages is about one academic year.

The fallback most people reach for is worse – screenshots in a phone gallery and a document of pasted text with the site’s menu wrapped around every quote. None of it is searchable, none of it says where it came from, and none of it can be cited in an essay without going back to find the source again, assuming the source is still there.

And the reading is never in one place to begin with. Some of it is on the learning platform, some behind the library proxy, some on a publisher’s site, some in a shared drive a seminar leader set up in second year. Each has its own login, and every one of those logins is switched off some months after you graduate. A folder of .md files on your own disk is the only copy that does not have an expiry date attached to your student number.

What you end up holding

• Select a passage and only the selection is clipped – the way to take one definition without the page around it.
• Every note carries the source URL, so citing it later takes no searching.
• A reading view shows the clip as ordinary text, without Markdown symbols, before you save it.
• Works with Obsidian, Logseq, Joplin or a plain folder of .md files – and with no app at all, since they are text.
• Pages behind your university login clip like any other, because the extension reads what your browser already rendered.
• All of it is free: no student plan to prove, no trial, no account.
• Reading lists and deadline tables come across as tables: 12 of 15 kept on the technical corpus, against 7 for each engine compared.
• Nothing is uploaded and no account exists, so there is no sign-in for your university’s IT rules to have an opinion about.

---
title: "Introduction to Algorithms: Lecture 2"
source: "https://ocw.mit.edu/courses/6-006-introduction-to-algorithms/"
extraction: "dom"
---

## Asymptotic notation

1. Big O – an upper bound on growth.
2. Big Omega – a lower bound on growth.
3. Big Theta – both bounds at once.

Sorting a list of *n* items by comparison takes at least *n* log *n* comparisons.

Setting it up for a term

Do this in the first week and the rest of the term costs one keystroke per reading. The part worth getting right on day one is where the files land.

1. Make one folder for your studies, Uni, with a subfolder per module inside it. Doing this before the first clip is what stops the folder turning into two hundred files with no order.
2. Right-click the extension icon, open Options, and point the destination at Uni. The browser asks once whether the extension may write there and remembers the answer.
3. Set the filename template to {date}-{title}. That is the day you clipped, so the folder ends up in the order you actually did the reading – which at revision time is more useful than alphabetical order.
4. Add a per-site rule for your learning platform – its domain goes to Uni/module, and the library or publisher domain goes to Uni/readings. After that, clips sort themselves.
5. Leave the icon click as the preview window. The window is where the switch between “selection” and “whole page” lives, and selection clipping is the habit worth building.
6. To take one definition rather than a chapter, select it on the page first, then click. The window opens showing the selection, and the switch at the top says which one you are looking at.
7. For a reading you would rather have on paper, use the print button in the clip window. It prints the cleaned text and writes no file.

Settings that suit a course

These differ from the defaults in one direction – everything here is about finding the file again in April, when the module page has been archived and you have four essays due.

| Week | Reading                                  | Due        |
| ---- | ---------------------------------------- | ---------- |
| 5    | Kahneman, *Thinking, Fast and Slow*, ch. 11 | 14 Nov  |
| 6    | Tversky & Kahneman (1974), pp. 1124–1131 | 21 Nov     |
| 7    | Seminar paper, 1500 words                | 28 Nov     |

Submit through the module page. Late work is capped at a pass unless an
extension was agreed in writing before the deadline.

Three moments in a term

One definition, not the whole chapter

A seminar reading defines a term in a single paragraph and you want that paragraph, not the nine screens around it. You select it, click the icon, and the window opens with the selection already showing. The switch at the top confirms you are looking at the selection rather than the page.

Saved that way, the note is one paragraph with the source URL above it. Pasting the same paragraph out of the browser would have brought the site’s furniture with it, and pasting it into an essay later would have brought that furniture into the essay.

Revision on a train with the wifi off

Everything for the module is .md text in one folder, so the search box in any editor covers a term’s reading in one query. There is no page to load, no login to renew and no connection needed, because the words are already on the disk in front of you.

This is the part that a bookmark cannot do at all. A bookmark is a pointer, and a pointer without a network is nothing; the same is true of a link to a library record whose proxy session expired an hour ago.

Citations at eleven at night

The essay is written, the references are not. Each clipped file has the source URL and the publication date in the first four lines, so the reference list is assembled from files on your disk rather than from browser history that was cleared in December.

Where a page published no date, the field is empty rather than filled with the day you clipped it. That is deliberate – an invented date in a reference list is the sort of error that is discovered by the person marking it.

Against how it is usually done

Everyone doing a degree has tried at least three of these. The differences show up in April rather than in October.

When a reading will not clip

The reading is a PDF and nothing happens

The extension converts the rendered HTML of a page, and a PDF opened in the browser’s viewer is not that. Where a reading exists in both forms – many journals and most open textbooks – clip the HTML version. For a PDF-only reading, keep the PDF and clip the module page that lists it, so at least the context and the URL are recorded.

It clipped the login page instead of the reading

The tab was showing the login screen, or the platform redirected before the page finished loading. Log in, make sure the reading itself is on screen, then clip. The extension reads what your browser has already rendered for your session, so anything you can actually see clips normally – including everything behind the library proxy.

The module page came out as a list of links

It did not – the extension will have reported “no article”. A module overview is mostly link labels, and when more than about a quarter of the extracted characters sit inside links the page is refused rather than returned as three hundred entries. Open the individual reading and clip that.

Only the first part of the reading came through

Either the page loads more text as you scroll (scroll to the bottom first, then clip) or the reading is displayed inside an embedded viewer, which is a separate document sitting in the page rather than part of it. Clipping the page around an embedded viewer does not reach inside it.

What it will not do for you

It does not download PDFs or slide decks; it converts the HTML page you are looking at. Formulas rendered as images stay image links, so a maths-heavy page loses those unless you keep images on. It does not summarise, translate or make flashcards. The text is converted, not rewritten. And it clips one page at a time – there is no button that swallows a whole reading list.

Questions