I did some reading and annotation and learned something deplorable about Perusall. What grade does your AI give me?


I’m excited to join Dan Allosso‘s book club on How to Take Smart Notes as a means of turning my active reading, annotating, and note taking into papers, articles and books using Obsidian.md and Hypothes.is


cc: Ian O’Byrne, Remi Kalir

Another really clever reimagining of the @Hypothes_is UI from @peterhagen_. Lots of color! A graph of annotations over time! And you can annotate on the page.

Try it at https://annotations.lindylearn.io/page/?url=https://www.theatlantic.com/magazine/archive/1945/07/as-we-may-think/303881/

Alternate Hypothes.is UI featuring the Atlantic article As We May Think with rainbow colored highlights on one side and annotations to the right. At the top is a graph of the article's popularity showing the number of annotations over time.

Different types of notes and use cases

In taking notes and making annotations recently, I’ve started a list of some of the broad semantic types I’ve come across. 


New ideas spurred by reading, potentially for future expansion and refinement.


Questions relating to the the text. What’s missing? What should have been asked or addressed? What biases exist that should be addressed?


Paraphrases and [[progressive summaries]] of articles or portions of articles. Restatements of ideas which may be reused in other contexts.


Basic, usually new, facts highlighted for future use and/or [[spaced repetition]]


Old school sententiae, aphorisms, and quotes for use in the future


Direct communication with others

Phatic notes and Reactions

Reactions, exhortations (Ha!, funny, ROFL, LOL, etc.), reacji, !, ?, ⭐, basic signs of life while reading


Are there any big holes I’m missing based on your experiences?

Annotation Colors Using Hypothes.is: Problems and Solutions

I frequently see the feature request from Hypothes.is users to have the tool add the ability to choose different colors for highlights.

The first occurrence of the issue is probably documented here: https://github.com/hypothesis/vision/issues/123, by the head of the company.

I suspect it may take a while before such a color feature might be built in, if ever. (Here I’ll note that I don’t work for or speak for the company or any of the other open source developers on the project, but I am one what one would consider a “heavy user”.) If they have the time (I know they’re very busy), perhaps they may chime in with a potential roadmap or other ideas.

Color highlights are a difficult user interface problem

While I’m thinking about it, in an academic context for students, colors may be slightly better indicators of different users’ annotations of a particular text as a means of differentiating one annotator from another more subtly, particularly on texts that are extensively marked up.

Just this difference points out what a mixed bag of functionality colored highlights brings from a usability, design, and user interface perspective. While colored highlights is a seemingly “simple” sounding feature in the analog world where a single document is only annotated by one user, mapping it into a digital shared context is a difficult engineering problem to navigate and solve for. What if your color meanings aren’t the same as those of another?, for example.

While colors can be useful for individuals, do they have the same place in a social annotation product?

I already find it difficult to annotate heavily annotated pages that all use the same color, much less a rainbow of others’ colors. (If this is also you, I’ll note that there’s a handy “eye” icon in the annotation drawer that will allow you to turn them on/off.)

Potential Color Highlight Hacks

While the value of colors may be useful in some contexts, you could potentially use a few other features, functionality, and methods to creatively achieve a similar feature in Hypothes.is for yourself today. Below are a few potential creative “hacks” that some might try.

Use the tagging functionality

You could use the tagging system to create specific tags to stand in for your desired colors: As an example, in some systems I might use the following color designators:

  • Yellow—general highlights and highlights which don’t fit under another category below
  • Orange—Vocabulary word; interesting and/or rare word
  • Green—Reference to read
  • Blue—Interesting Quote
  • Gray—Typography Problem
  • Red—Example to work through

Instead of colors in Hypothesis, for example, one could use the tags “words” or “vocabulary”, “reference” or “citation“, “typo“, “quotes“, or “examples” to stand in for these particular “colors” respectively. I sometimes practice some of these which you can find by clicking on the links, though you may note that in practice I use other tags for them.

In some sense, this is what the software would be doing, particularly with regard to search for these after-the-fact. If you wanted a list of all your “citations” for example, you’d have to search for the color for that and be able to find them all, presuming this search functionality existed with such a color feature. This isn’t really much different than simply tagging all those particular highlights with words like “citation” or “reference” in the first place.

Use the Group Functionality

You could created different “groups” (private or public) to stand in for the colors you wish you had, thus a “yellow group” could be used for one “color” of highlight and a “green group” for another. ( See Annotating with Groups for more details.)

Switching between groups for annotating isn’t going to be drastically different than a user interface for switching colors of highlighter. The one drawback (or perhaps it’s a feature?) here is that you will only be able to see one “color” at a time.

Roll your own solution with open source

As ever, with some work, you could self-host the open source software and modify your copy to add this functionality in for yourself.

Some clever hacks in your browser with CSS might also give you your preferred output. I know some users have done custom work to the Hypothes.is UI in the past: eg. https://tomcritchlow.com/2019/02/12/annotations/, see his gist at the bottom of the post.

Another custom solution which may give you ideas can also be found at https://web.hypothes.is/blog/do-it-yourself-anchoring-and-the-evolution-of-the-hypothesis-toolkit/.

Perhaps adding custom classes on the tags or usernames might allow people the ability to target highlights on a page so that one could define custom CSS rules for each highlight using either usernames of tags as well? Of course, just like the “eye icon” described above, I’m sure there are times that people will appreciate the ability to turn these colors on and off. I personally don’t want the clean interface dressed up in Josephs Amazing Technicolor Annotation Dreamcoat.

Other solutions and problems?

Are there ideas or potential solutions for color highlights I’ve missed? How about design problems that might be encountered in implementing color-coded highlights in the older single document/single user model being transferred to a multi-user space with infinite scale? Is color the best and most accessible solution? Are there better things that could be done with color in the product?

Feel free to comment below with your ideas or links to examples.

Interested in other Hypothes.is hacks, tips, and ideas? Try browsing my Hypothes.is archive.

Annotated a tweet by @AmberRegis (Twitter)
I’d probably have done it digitally with @Hypothes_is to share with others, but kudos to those who can still fathom the analog.

When will we have real print-on-demand?

I’ve been thinking it for a while, but have needed to write it down for ages—particularly from my experiences with older manuscripts.

In an age of print-on-demand and reflowing text, why in goodness’ name don’t we have the ability to print almost anything we buy and are going to read in any font size and format we like?

Why couldn’t I have a presentation copy sized version of The Paris Review?

Why shouldn’t I be able to have everything printed on bible-thin pages of paper for savings in thickness?

Why couldn’t my textbooks be printed with massively large margins for writing notes into more easily? Why not interleaved with blank pages? Particularly near the homework problem sections?

Why couldn’t I buy my own hardcover, custom edition of Annotation with massive five inch margins to really make having a handwritten easier? (C’mon MIT Press, I know it’s part of a pre-existing series, but editorial considerations should have necessitated leaving at least an inch!)

Why can’t I have more choice in a range of fonts, book sizes, margin sizes, and covers?

When are publishing platforms going to give us this?!? 

Annotated About by Mandy BrownMandy Brown (A Working Library)
books are a means of listening to the thoughts of others so that you can hear your own thoughts more clearly. 
to which I might add:

And annotation helps you save those thoughts, share them with others, and further refine them.

Replied to a thread by Annie Murphy Paul and Howard Rheingold (Twitter)
Anecdotally, I’ve seen this sort of behavior happening in the margins of ethical educational tools like @Hypothes_is. Some of them are in smaller groups and private groups, which stay hidden. I try to do much of my thinking & commenting there in public: https://hypothes.is/users/chrisaldrich
Holiday gift idea for your favorite annotation fan/nerd: Used Books: Marking Readers in Renaissance England by William H. Sherman (University of Pennsylvania Press, 2007) 

(There should be a discount code KISLAKCTR21-FM to get 40% off & free shipping as part of the Schoenberg Symposium for the next few days Nov. 17-19, 2021)

Annotated Curating a Public Conversation about Annotation (commonplace.knowledgefutures.org)
For academics, annotation is also essential to scholarly communication and knowledge production. With Annotation, we eagerly accepted a social and scholarly responsibility to spark, curate, and facilitate discussion about annotation. 
The tools for thought crowd should all be reading Kalir and Garcia’s book Annotation.
Read Ebooks Are an Abomination by Ian BogostIan Bogost (The Atlantic)
If you hate them, it’s not your fault.
Ian Bogost has a nice look at the UI affordances and areas for growth in the e-reading space.

A🧵 of annotations

What any individual infers about their hopes and dreams for an e-reader derives from their understanding of reading in the first place. You can’t have books without bookiness. Bookiness. That’s the word Glenn Fleishman, a technology writer and longtime bookmaker, uses to describe the situation. “It’s the essence that makes someone feel like they’re using a book,” he told me. Like pornography or sandwiches, you know bookiness when you see it. Or feel it? Either way, most people can’t identify what it is in the abstract.

definition: bookiness

Does this only come out because there’s something that’s book-tangential or similar and it needs to exist to describe the idea of not-book, book-adjacent, or book-like on some sort of spectrum of bookishness.
Annotated on September 18, 2021 at 12:28PM

The ancient Romans sometimes connected wax tablets with leather or cords, suggesting a prototype of binding. Replacing the wax with leaves allowed many pages to be stacked atop one another, then sewn or otherwise bound together. 

early book prototypes
Annotated on September 18, 2021 at 12:30PM

In other words, as far as technologies go, the book endures for very good reason. Books work. 

Aside from reading words to put ideas into my brain, one of the reasons I like to read digital words is that the bigger value proposition for me is an easier method to add annotations to what I’m reading and then to be able to manipulate those notes after-the-fact. I’ve transcended books and the manual methods of note taking. Until I come up with a better word for it, digital commonplacing seems to be a useful shorthand for this new pattern of reading.
Annotated on September 18, 2021 at 12:33PM

If you have a high-quality hardbound book nearby, pick it up and look at the top and bottom edges of the binding, near the spine, with the book closed. The little stripey tubes you see are called head and tail bands (one at the top, one at the bottom). They were originally invented to reinforce stitched binding, to prevent the cover from coming apart from the leaves. Today’s mass-produced hardcover books are glued rather than sewn, which makes head and tail bands purely ornamental. And yet for those who might notice, a book feels naked without such details. 

It is an odd circumstance that tail bands are still used on modern books that don’t need them. From a manufacturing standpoint, the decrease in cost would dictate they disappear, however they must add some level of bookiness that they’re worth that cost.
Annotated on September 18, 2021 at 12:37PM

One site of that erosion, which may help explain ebook reticence, can be found in self-published books. For people predisposed to sneer at the practice, a lack of editing or the absence of publisher endorsement and review might justify self-published works’ second-class status. That matter is debatable. More clear is the consequence of disintermediation: Nobody takes a self-published manuscript and lays it out for printing in a manner that conforms with received standards. And so you often end up with a perfect-bound Word doc instead of a book. That odd feeling of impropriety isn’t necessarily a statement about the trustworthiness of the writer or their ideas, but a sense of dissonance at the book as an object. It’s an eerie gestalt, a foreboding feeling of unbookiness. 

Having helped others to self-publish in the past, I definitely do spend a bit of time putting the small sort of bookiness flourishes into their texts.
Annotated on September 18, 2021 at 12:41PM

The weird way you tap or push a whole image of a page to the side—it’s the uncanny valley of page turning, not a simulation or replacement of it. 

This may be the first time I’ve seen uncanny valley applied to a topic other than recognizing people versus robots or related simulacra.
Annotated on September 18, 2021 at 12:44PM

The iPad’s larger screen also scales down PDF pages to fit, making the results smaller than they would be in print. It also displays simulated print margins inside the bezel margin of the device itself, a kind of mise en abyme that still can’t actually be used for the things margins are used for, such as notes or dog-ears. 

It would be quite nice if a digital reader would allow actual writing in the margins, or even overlaying the text itself and then allowing the looking at the two separately.

I do quite like the infinite annotation space that Hypothes.is gives me on a laptop. I wish there were UI for it on a Kindle in a more usable and forgiving way. The digital keyboard on Kindle Paperwhite is miserable. I’ve noticed that I generally prefer reading and annotating on desktop in a browser now for general ease-of-use.

Also, I don’t see enough use of mise en abyme. This is a good one.

In Western art history, mise en abyme (French pronunciation: ​[miz ɑ̃n‿abim]; also mise en abîme) is a formal technique of placing a copy of an image within itself, often in a way that suggests an infinitely recurring sequence. In film theory and literary theory, it refers to the technique of inserting a story within a story. The term is derived from heraldry and literally means “placed into abyss”. It was first appropriated for modern criticism by the French author André Gide.

Annotated on September 18, 2021 at 12:49PM

Ebook devices are extremely compatible with an idea of bookiness that values holding and carrying a potentially large number of books at once; that prefers direct flow from start to finish over random access; that reads for the meaning and force of the words as text first, if not primarily; and that isn’t concerned with the use of books as stores of reader-added information or as memory palaces. 

Intriguing reference of a book as a memory palace here.

The verso/recto and top/middle/bottom is a piece of digital books that I do miss from the physical versions as it serves as a mnemonic journey for me to be able to remember what was where.

I wonder if Ian Bogost uses the method of loci?
Annotated on September 18, 2021 at 12:53PM

So do all manner of other peculiarities of form, including notations of editions on the verso (the flip side) of the full title page and the running headers all throughout that rename the book you are already reading. 

I do dislike the running headers of digital copies of books as most annotation tools want to capture those headers in the annotation. It would be nice if they were marked up in an Aria-like method so that annotation software would semantically know to ignore them.
Annotated on September 18, 2021 at 12:56PM

Skimming through pages, the foremost feature of the codex, remains impossible in digital books. 

This is related to an idea that Tom Critchlow was trying to get at a bit the other day. It would definitely be interesting in this sort of setting.

Annotated on September 18, 2021 at 01:03PM

“We’ve been thoughtful,” Amazon continued, “about adding only features and experiences that preserve and enhance the reading experience.” The question of whose experience doesn’t seem to come up. 

They’re definitely not catering to my reading, annotating, and writing experience.
Annotated on September 18, 2021 at 01:04PM