Section on Genius annotation. The full whitepaper this is from (lots of stuff on social annotation, digital editions, community reading!) is at dr.amandavisconti.com.
The Genius annotated version of Ulysses comes closest to my vision of a public conversation around Ulysses: visitors can highlight words or passages in the text and add their interpretations, questions, and comments1. Genius grew out of the "Rap Genius" site, a project allowing rap enthusiasts to annotate lyrics with interpretations and contextualizations on the same level as other types of poetry. Some of its design metaphors are more understandable knowing they arose from a music tradition rather than a print text annotation tradition.
Readers can up- or down-vote existing annotations as well as "promote" particular annotations onto the profile pages of users who follow their account. Genius grants locked-down, custom "classroom" versions of any text to a teacher and the users the teacher allows; these versions are stripped of all existing annotations so that a reading group can begin with a fresh slate (e.g. my own, recent classroom version of the first episode of Ulysses). Video, audio, and hyperlinks are all allowed as part of an annotation, which means that users have more ways of helping each other understand what Stephen's hat looks like or the sounds you might hear walking along Sandymount Strand. Account verification and annotation cosigning add levels of moderation and approval to user participation. In contrast to the Wikibooks Ulysses, the Genius Ulysses has attracted adequate attention (48 contributors and over 36,000 views) and relies on strong policies and instructions on what makes a good and ethical annotation.
Where does the Genius Ulysses veer from my goals for Infinite Ulysses? Like many public-aimed digital Ulysses projects, the Genius text was unreliable (pages were unclear as to which text of Ulysses they represented)—an unsurprising trend given that until recently, the Project Gutenberg e-text of Ulysses, created by volunteers without a clear methodology or oversight, was the only freely available digital transcription of the novel. The Modernist Versions Project (MVP) recently solved this problem by providing public, free, reusable Modernist texts as digital facsimiles and digital transcriptions checked by scholars. Both the text of the 1922 Shakespeare and Co. edition built on by Infinite Ulysses and Samuel Roth's "pirated" edition of Ulysses (which appeared in the U.S. Two Worlds Monthly magazine in 1926) are available from the MVP, so digital Ulysses projects can now easily start from a reliable text.
My takeaway from the Genius Ulysses for Infinite Ulysses was that features allowing participatory creation of annotations needed to be complemented with personalization features that pull each reader's sound from the noise. The Genius Ulysses deals with this by allowing educators to control their own limited-access versions of a text, and by preventing more than one annotation per piece of highlighted text (readers can reply to that one annotation, but these replies are arranged hierarchically beneath the annotation that got there first). I didn't see the Yellow Block Syndrome on Genius' Ulysses (when most or all of a text is highlighted in yellow for annotation purposes2), but that seemed due to Genius' only allowing one annotation for any piece of text, moderation by teachers on classroom versions of the text, and a limited number of contributors actually adding annotations. For a platform like Infinite Ulysses that invites a conversation around the text, every user needed the option to add their own annotation to any piece of the text, and readers needed to see annotations displayed not in an arbitrary hierarchy but sorted by preferences like topic and author.
Annotation tagging on Genius' Ulysses is absent. Infinite Ulysses' allows the user to hide or display reader comments depending on their topic. Annotated reading is, by definition, being mentally pulled out of the main text repeatedly; since the earliest incarnations of this project (UlyssesUlysses.com), I've tried to design the reading page so that such distractions are limited to desired information: the definitions, interpretations, and comments a given reader is interested, and nothing else. With the Genius interface, you'd need to weigh whether you were willing to click each highlight, with the chance of the annotation not being worth the interruption to you—or of missing annotations that would benefit you, because you were tired of gambling on whether a distraction was worthwhile.
A related issue on Genius was that a given annotation's author is invisible unless you click on the highlight and then click on a "x contributors" link to toggle a view of the associated authors. The display of contributions on a page via bar charts shows relative amounts of activity, although it is not clear whether this activity compared number of annotations, word count among a user's annotations, percent of the page annotated, or something else. While I sympathize with the drive to keep annotation display minimal and thus reduce reader distraction, a platform relying on user-generated content should keep credits to those authors visible to the casual reader without requiring special action.
The Genius Ulysses design was text-agnostic: the same design as any other book, poem, or rap song on the site. Visual design always carries an argument and a bias, and interfaces to texts are interpretation. Seeing that people could happily interact with Ulysses without a visual design critically crafted to that particular text reminded me of my experience with an early prototype, UlyssesUlysses.com. The design on that early site was specific to Ulysses, but was created at an early stage of my web design learning and wasn't particularly visually supportive of the text; nevertheless, readers contacted me having used the prototype and wanting to know when additional chapters would be available. Despite strong feelings about visual design rhetoric, these two examples helped me set a reasonable scope for the Infinite Ulysses project: thinking about what I can reasonably add to the digital Ulysses conversation in the course of a dissertation, rather than trying to build everything I wish were available and design the perfect interface to compliment my interpretation of the text.