Regarding Highlights

I didn't want to write about “meta” topics (“writing about writing”), but there's a tool issue, a workflow issue that really bothers me. It has to do with highlights.


Traditionally, when you went to buy a used textbook at the campus bookstore, your nightmare was to find one full of somebody's yellow highlights. It used to be that readers bought pastel-colored “highlighters” to mark up text in a book or printed article. If you weren't very discerning, you could highlight too much.

“Oh, that's interesting.”

“Oh, that's important.”

“Can't forget this key term.”

“Can't forget that key date.”

“This will be on the test, for sure.”

Mortimer Adler in his classic How to Read a Book will tell you to mark up your books, have a conversation with the author, write in the margins, and make use of end pages for summaries and notes. Not only can one highlight with yellow, or some other color, or many colors all at once, one can underline, or bracketing off selected sentences, or put asterisks or other signs and symbols in the margin. Highlights combine with commentary, combine with primitive forms of “tagging” (asterisks, check-marks, cryptic abbreviations, and other custom mark-up).

All this reading activity combined to create today's standard for working not only on paper copy books and articles, but on PDFs and eBooks.

Enter the PDF reader app, which encouraged highlighting and commenting. Kindle and eBook devices and apps do the same thing for electronic books. Web highlighters work on published articles online.


With the proliferation of apps that highlight, and with knowledge workers starting to use a wide range of note-taking apps for “personal knowledge management” (PKM), there was a clear opportunity to import highlights from one direction (many sources) and export them to another (ideally one). Enter Readwise, a highlight aggregation service. It imports from all over, exports to your PKM of choice, and for good measure includes a spaced repetition feature so you can review your highlights over time.

Readwise recently released their (amazing) Reader that combines the highlight aggregation of the original Readwise with a cutting edge read-it-later application (innovating on old standbys like Pocket or Instapaper) and a basic RSS feed reader (think Feedly or Inoreader). Readwise Reader has helpful tools for organizing and managing your reading pile, filtering and saving custom views, and — of course — highlighting, commenting, and tagging as you read.

All this is wonderful.

For those of us who still love our hard copy books, we do still need a way to scan, OCR, and save highlights (with source and page number) in order to use the data within a now fully-integrated workflow. Readwise has a function in its mobile app to do take highlights from paper books, but I like the Highlighted app better. (I would love for it to become an import source for Readwise, which would complete the workflow.)

So far, so good.

Working with Highlights

So what can you do with highlights once you have them captured from a source text, imported into Readwise or some other service, and exported from there to your notes database or PKM system? If you use one of the cutting edge PKM systems like Roam, Obsidian, Logseq, or about a dozen others that have come on the market in the last couple years, the answer is: lots!

Each highlight can become its own “atomic note,” which you can then bidirectionally link with other notes. You can filter on tags. You can create interlinked visual graphs or whiteboards. You can work at a block-level or note-level. You can wander through your collection at will to explore connections.

New tools like Napkin, aided by AI-powered auto-tagging, can help make serendipitous finds by resurfacing notes together in surprising ways. And one can continue to make use of spaced repetition, which is built into many notes systems, for recalling forgotten notes and for review.

When you’ve completed your research and are ready to enter the writing phase, you can plan out a draft much like old-time students and scholars used to do by using boxes of index cards — only now it’s all electronic. You could make use of citation managers (e.g. Zotero or many others) for tracking references and auto-creating citations, footnotes, endnotes, and bibliographies.

In theory, this is how it all could work.

In practice, I find things to be much messier.

Even if I could achieve it, the ideal version of such a system is missing some essential things.

Tomorrow I'll propose what I see as the missing elements, ask if there are any other tools that might help, and suggest an alternative to the almighty highlight as the preferred unit of analysis.