How to Highlight Text on Web Pages: Why Most People Do It Wrong
Kazuki Nakayashiki
Jun 25, 2026 7:14 AM
Summary
-
(00:02) Episode premise: defending highlighting on the web – highlighting isn’t inherently useless; it just has to be done correctly to actually help learning.
-
(01:05) What a highlight does: highlights capture exact sentences rather than vague bookmarks, making it clearer what actually mattered for your future self.
-
(01:38) Two architectures explained: (a) highlight on the live page attached to the URL; (b) a separate app scrapes the article and saves a copy on its servers (read-later model).
-
(02:25) Pros/cons of read-later copies: offline access and stable snapshots vs. reading a copy on someone else’s server; highlighted data can be premium in these tools, and a real-world risk is illustrated by Pocket’s shutdown and data loss (03:03–03:21).
-
(03:58) Glasp setup and workflow: quick two-minute install; highlight a sentence and choose a color; notes attach to highlights; a searchable profile library; exports to Readwise, Notion, Obsidian, Roam.
-
(06:33) The study nuance and best practices: the famous Dunlosky 2013 critique hinges on how highlighting is used; when used with selectivity, added notes, color-coding, and weekly review, highlighting earns its keep on the web.
Transcript
Welcome back to Glasp Deep Dive. I'm Hope, here as always with Adam, and today we're doing something slightly risky. We're defending highlighting.
Risky is the right word. Telling people to stop highlighting might be the single most repeated piece of study advice on the internet.
And I've read those articles. Science says highlighting is useless. It's the junk food of studying. So convince me. Why are we spending a whole episode on how to highlight text on webpages?
Because that advice is a misreading. The study everyone cites didn't test highlighting. It tested highlighting badly. The same research literature shows that done right, it genuinely works.
Okay, you'll have to pay that off later in the show. But let's ground people first. When we say online highlighter, we're talking about a browser extension, right?
Right. It adds a marking layer on top of any webpage. Select a sentence, click a color, it's saved. Chrome, Safari, Edge, Brave, Opera, they all support these.
And the pitch over plain bookmarking is that you capture the passage, not the page. A bookmark says, "This article was useful somehow. Good luck remembering why." A highlight says, "This exact sentence. This is the thing."
Which is why people who read a lot on the web end up here. Your future self doesn't want the URL. It wants the three lines that actually mattered.
My bookmarks folder is a graveyard, so that lands. Now, you told me before the show that these tools are not all the same under the hood.
Not even close. There are two completely different architectures, and the choice matters more than any feature list.
Two architectures. Okay. Walk me through the first one.
Architecture one: highlight on the original page. The extension lives in your browser, and your highlights attach to the real URL. Visit the page again, your marks are sitting right there.
So you're always on the live page, ads and all, but the highlights follow you back. And sharing is natural because you're pointing at the actual web, not at a copy.
Exactly. Architecture two flips it. The app scrapes the article, saves a clean copy on its own servers, and you read and highlight that copy. That's the read later family: Instapaper, Matter, Readwise Reader.
And to be fair to that camp, the upside is real. Ads stripped out, proper offline reading, a stable snapshot that survives page changes, though highlighting is often a premium feature over there.
Right. The deeper downside is that you're not on the web anymore. You're in their app reading a frozen copy on someone else's servers.
Which brings us to Pocket. And I wanna tell this one because if anyone thinks a frozen copy on someone else's servers is a theoretical risk, it is not. Pocket was the biggest read later app on the planet, a decade and a half of people's saved reading.
And then Mozilla shut it down. The site went dark in October 2025.
But here's the part that still gets me. It wasn't just the app going away. Saved libraries were deleted from the servers. Years of articles, years of highlights, gone.
If you didn't export before the deadline, gone for good. And honestly, the lesson cuts across both architectures. Your reading library lives on somebody's servers either way.
So the feature that matters most isn't colors or tags. It's the export path.
That's my one rule now. If this company vanished tomorrow, could I walk away with my data in a format another tool can read? If no, keep looking.
Okay. Let's get concrete and actually walk through a setup. And full disclosure before we do, Glasp makes this show, and Glasp is a web highlighter, the original page kind.
So judge our enthusiasm accordingly. The setup really is about two minutes though. Create a free account, install the extension, done. Chrome, Safari, Brave, Edge, Opera are all covered.
Then I just go read like normal. I select a sentence, and what happens?
A small popup appears with four colors. Click one and the text is highlighted, and that sentence also lands in a sidebar collecting everything from the page as you go.
And I can attach a note to any highlight, which, small spoiler, turns out to be where the real learning value lives.
We'll get there. Everything you save also collects on your profile page, which becomes a searchable library of your reading. Search by keyword, browse by article, all in one place.
And the Pocket question, because I'm not letting that go. When I want my stuff out, what are my options?
You can download plain text, markdown, HTML, or CSV, or sync straight to Readwise, Notion, Obsidian, or Roam. That's the first thing I'd verify in any tool now, and it's there.
All right, honest limits time. Nothing is perfect, so tell me what breaks.
Two things worth knowing. If a site heavily redesigns a page, the in-place display can break, so a highlight might not render where it used to be. The text stays saved in your profile regardless, but it's a real trade-off of the original page approach.
Okay, and you said two things. What's the second?
Glasp is social by default. Your highlights are public unless you change that, and you should know it going in. Oh, and PDFs are a different game entirely. Different technology. Today is strictly webpages.
Let's sit on the social part for a second because it sounds like a privacy quirk, but you keep insisting it's actually the main event.
For learning, it kind of is. Open an article and you can see what other readers highlighted on the same page. You get multiple perspectives on one text instead of just your own.
The what did they see that I missed moment, which now that I say it out loud, is basically free retrieval practice. We did a whole episode on this, the active recall one.
Exactly. It plugs straight in. Researchers call the bigger effect distributed cognition, and there's work out of Cornell showing social annotation improves comprehension and retention.
Okay, time to pay off my challenge from the top of the show. The famous study, the one every stop highlighting article cites. Make your case.
2013, a researcher named Dunlosky led a huge review of study techniques and rated highlighting low utility. The headlines wrote themselves, but the fine print matters. But the fine print matters. They rated it as most students actually do it.
Which is what? Dragging the marker across the entire page like I did in college?
Nearly every line, no deeper processing, and we've known that fails for ages. 1974, a study by Fowler and Barker found test performance actually dropped as the amount of highlighting went up.
Mark everything, differentiate nothing. The page is all yellow, so nothing stands out, and you never made a single decision while you were reading.
And the decision is the whole game. A 2015 study showed that choosing what to mark forces you to weigh every sentence as you read, and that evaluation is active processing. It measurably improves recall.
So the decision is the work, and the color is just the receipt that you did it. And there's a wrinkle. A 2024 study found that on paper, comprehension dropped as highlighting increased. But on digital text, the pattern reversed.
Because digital tools give you precise selection, easy search, easy review. The medium finally caught up with the technique.
So highlighting is acquitted, but with conditions. Let's spell out the conditions. Practice number one has to be selectivity.
Read the paragraph first, then go back and mark one sentence, maybe two. Across a whole text, ten to twenty percent, absolute ceiling.
Read first, highlight second. I love that because it kills the paint everything reflex on its own. What's practice two?
Add a note while the thought is hot. A bare highlight can feel meaningless three weeks later. A highlight with a note attached, something like, "This contradicts what I read last month," that stays gold.
Three is color-coding, and I actually run a system like this: yellow for the main idea, green for evidence, blue for questions I want to chase down, pink for connections to other things I've read.
And the system is the point, not the colors. Choosing a category forces a classification step while you read, which is more processing. Plus, your library becomes scannable at a glance.
And four is review because of that famous forgetting curve from Ebbinghaus. Without review, you lose most of what you read within a week. So my move is once a week, open the recent highlights and try to recall the context before I reread anything.
That little struggle before you peek is retrieval practice, the technique that same review rated among the very best. And a weekly recap email can do the remembering for you.
One more habit question. When all of this lands in Notion or Obsidian, how should people organize it?
By topic, not by source. Highlights grouped by URL are a graveyard. The same highlights grouped by theme are a knowledge base you can actually think and write from.
Quick questions before we close, the ones people actually type into search. First, does this work on every website?
Almost every one. Pages that block extensions won't cooperate. Some banking sites, the Chrome Web Store itself, plus the odd strangely rendered app. Regular articles and blog posts, basically all of them work.
Second one, do my highlights actually stick around when I come back to a page next month?
Every revisit, as long as you're logged in with the extension installed. And even if the page changes underneath you, the saved text is in your profile either way.
And the one I hear the most, can other people see my highlights?
In Glasp, yes. By default, that's the point of the social layer. Other tools offer private modes. Pick whichever matches how you want to read. Just pick it knowingly.
So here's my verdict, and I'm comfortable with it now. Highlighting was never the problem. Passive highlighting was the problem.
Mark less, note more, review weekly. Do those three things, and the oldest study technique there is earns its keep on the web.
Want to go deeper? Full article's on glasp.co. Just search how to highlight text on web pages.
See you in the next deep dive.
Share This Audio Transcript & Summary 📚
How to Highlight Text on Web Pages: Why Most People Do It Wrong
00:00
11:15