--- layout: presentation title: FOOBAR --Week N-- description: Accessibility class: middle, center, inverse --- background-image: url(img/people.png) .left-column50[ # Welcome to the Future of Access Technologies Week N, FOOBAR {{site.classnum}}, {{site.quarter}} ] --- name: normal layout: true class: --- # Important Reminder ## This is an important reminder ## Make sure zoom is running and recording!!! --- [//]: # (Outline Slide) # Learning Goals for today --- # Topics - Video conferencing accessibility - Understanding different access needs of different participants + access conflicts - Identifying built-in accessibility features - Access practices to navigate inaccessibility / role of sociocultural https://dl.acm.org/doi/pdf/10.1145/3479578 - Connections: - Privacy concerns with captioning - Machine learning bias in speech recognition - Collaborative coding / writing - https://dl.acm.org/doi/abs/10.1145/3359293 - https://dl.acm.org/doi/abs/10.1145/3517428.3544812 - Crowdsourced accessibility (not sure if this fits here but mostly thinking of how cscw systems can help provision access as well) - SoundPrint, DeafFriendly, BeMyEyes - Opensource projects: NVDA - Datasets: VizWhiz, ASLCitizen --- # Study of End User Content Creation Although guidelines for tools and for authorship exist, inaccessible content is still created. - How do communities react - How could authoring tools help this? .quote[I wan't able to read your entire post. Maybe you could go back and break up your paragraphs so they are easier to read?] ??? This quote is taken from a study investigating communities that take an active role in fostering a culture of accessibility, using both formal and informal practices of moderating inaccessible content. --- # Case Study 1: Wikipedia Sample size: 179 discussions, 82 contributors Of 82, 10 “moderators” – declared members Qualitative analysis (coding  themes) --- # What Accessibility Process does Wikipedia Have? Formal discussion of policy such as Manual of Style Process of feedback / moderation .quote[[Marking articles as accessible] would be as if people went and marked templates as ‘This is an article with a Neutral POV’ and ‘This is an article that is well referenced’] --- # Key Activity: Moderation When moderating content accessibility, a moderator provides advice, guidance, or critique relating to an author’s content through - Talk pages - Flags --- # Key Activity: Moderation Authority is important for moderators, because sometimes authors did not want to comply with moderator requests. Moderators sometimes idealized the process .quote[It isn’t that hard to make sure a page is accessible - just use conventional formatting and heading standards -- M2] --- # Key Activity: Moderation In reality, moderators and authors spent significant time on tensions between accessibility goals and other editing goals. For example, - Remove valued moving images - Increase markup (in violation of another Wikipedia policy) - Change page titles (affecting google search results). Additionally, some authors simply disagreed with or reverted changes made by moderators. ??? - Is an animation a distraction or useful? - Is it ok to use the Pi symbol π - Semantic versus visual formatting: End user content creation tools limit changes to header formatting, pushing authors to do more visual formatting --- # Key Activity: Advocacy .quote[I could be [an author advocate] but what exactly would I say? "Don’t ban this person, they have [disability]."? That might not work.] .quote[No, how about "Hi, this user has identified herself as (having a condition), which is probably why (this problem arose). Before we ban her … she seems to have her heart in the right place and has a lot to offer our community. Obviously, this would not apply to users who are violent or angry…" ] ??? Those 10 moderators were concerned that not everyone would speak up, and discussed pro-actively addressing situations. Discussions of moderator-initiated author advocacy in the data lacked any reference to whether it had actually been done successfully or whether authors would want this. --- # Why is this interesting All of these may occur in a wide variety of settings - Complex organizational structure - Separation between authors and consumers - Use of GUIs for creating content --- # Case Study #2: Lymespace Users actively intervene to improve accessibility, relative to cognitive symptoms of Lyme disease “Brain fog” affects use of online content - Distraction when faced with visuals ranging from animated advertisements to bold text - Difficulty remembering the start of a long passage by the end of the same passage - Getting lost among multiple pages of content --- # Investigated Moderation What kinds of moderation occurs? How to users respond? - Qualitative analysis of 6 authors and their moderators When does moderation occur and what happens after? - Quantitative analysis of 600 threads --- # Quantitative Impact of Moderation .left-column60[ ![:img Diagram showing 300 unmoderated threads; 300 threads with moderating posts. Of the 300 moderated threads; 240 include responses (after moderation) and 60 do not. ,60%, width](img/assessment/moderation.png) ] .right-column40[ Metrics - Characters per paragraph - Capitalization (%) - Punctuation (%) - Bold (%) ] --- # Were moderated posts less accessible? ![:img Two bar charts showing random posts and pre-moderation posts. Pre-moderation posts had a significantly longer paragraph size (150 characters on average versus 100 for random posts). Both pre and post moderation posts had a similar percentage of punctuation (4%); boldface (less than 1%); and capitalization (around 5%) on average. ,60%, width](img/assessment/moderation-stats.png) --- # Were moderated posts less accessible? ![:img Comparing moderated posts before and after moderation, paragraph size dropped significantly to the same as random posts; capitalization use increased significantly from 5 to 6%; and punctuation increased significantly from just under to just over 4% ,60%, width](img/assessment/moderation-stats.png) --- # Alternatives to Moderation "Soft Paternalism" [Mack et al, 2021](https://dl.acm.org/doi/10.1145/3441852.3471207) Interfaces that suggest what to include in alt text are beneficial - Leads to more complete, detailed ALT text - Structure helps (e.g. types of content to cinclude) - Ok to prefill with automatically generated text